Authors: Yongyu Wang
Abstract: Graph Neural Networks (GNNs) have emerged as a dominant paradigm for learning on graph-structured data, thanks to their ability to jointly exploit node features and relational information encoded in the graph topology. This joint modeling, however, also introduces a critical weakness: perturbations or noise in either the structure or the features can be amplified through message passing, making GNNs highly vulnerable to adversarial attacks and spurious connections. In this work, we introduce a pruning framework that leverages adversarial robustness evaluation to explicitly identify and remove fragile or detrimental components of the graph. By using robustness scores as guidance, our method selectively prunes edges that are most likely to degrade model reliability, thereby yielding cleaner and more resilient graph representations. We instantiate this framework on three representative GNN architectures and conduct extensive experiments on benchmarks. The experimental results show that our approach can significantly enhance the defense capability of GNNs in the high-perturbation regime.
Authors: Hans Jarett J. Ong, Brian Godwin S. Lim, Dominic Dayta, Renzo Roel P. Tan, Kazushi Ikeda
Abstract: Unsupervised representation learning seeks to recover latent generative factors, yet standard methods relying on statistical independence often fail to capture causal dependencies. A central challenge is identifiability: as established in disentangled representation learning and nonlinear ICA literature, disentangling causal variables from observational data is impossible without supervision, auxiliary signals, or strong inductive biases. In this work, we propose the Latent Additive Noise Model Causal Autoencoder (LANCA) to operationalize the Additive Noise Model (ANM) as a strong inductive bias for unsupervised discovery. Theoretically, we prove that while the ANM constraint does not guarantee unique identifiability in the general mixing case, it resolves component-wise indeterminacy by restricting the admissible transformations from arbitrary diffeomorphisms to the affine class. Methodologically, arguing that the stochastic encoding inherent to VAEs obscures the structural residuals required for latent causal discovery, LANCA employs a deterministic Wasserstein Auto-Encoder (WAE) coupled with a differentiable ANM Layer. This architecture transforms residual independence from a passive assumption into an explicit optimization objective. Empirically, LANCA outperforms state-of-the-art baselines on synthetic physics benchmarks (Pendulum, Flow), and on photorealistic environments (CANDLE), where it demonstrates superior robustness to spurious correlations arising from complex background scenes.
Authors: Jiesong Lian, Ruizhe Zhong, Zixiang Zhou, Xiaoyue Mi, Yixue Hao, Yuan Zhou, Qinglin Lu, Long Hu, Junchi Yan
Abstract: Post-training alignment of video generation models with human preferences is a critical goal. Developing effective Reward Models (RMs) for this process faces significant methodological hurdles. Current data collection paradigms, reliant on in-prompt pairwise annotations, suffer from labeling noise. Concurrently, the architectural design of VLM-based RMs, particularly their output mechanisms, remains underexplored. Furthermore, RM is susceptible to reward hacking in post-training. To mitigate these limitations, we propose SoliReward, a systematic framework for video RM training. Our framework first sources high-quality, cost-efficient data via single-item binary annotations, then constructs preference pairs using a cross-prompt pairing strategy. Architecturally, we employ a Hierarchical Progressive Query Attention mechanism to enhance feature aggregation. Finally, we introduce a modified BT loss that explicitly accommodates win-tie scenarios. This approach regularizes the RM's score distribution for positive samples, providing more nuanced preference signals to alleviate over-focus on a small number of top-scoring samples. Our approach is validated on benchmarks evaluating physical plausibility, subject deformity, and semantic alignment, demonstrating improvements in direct RM evaluation metrics and in the efficacy of post-training on video generation models. Code and benchmark will be publicly available.
Authors: Chuanting Zhang, Haixia Zhang, Jingping Qiao, Zongzhang Li, Mohamed-Slim Alouini
Abstract: The growing demand for intelligent, adaptive resource management in next-generation wireless networks has underscored the importance of accurate and scalable wireless traffic prediction. While recent advancements in deep learning and foundation models such as large language models (LLMs) have demonstrated promising forecasting capabilities, they largely overlook the spatial dependencies inherent in city-scale traffic dynamics. In this paper, we propose TIDES (Traffic Intelligence with DeepSeek-Enhanced Spatial-temporal prediction), a novel LLM-based framework that captures spatial-temporal correlations for urban wireless traffic prediction. TIDES first identifies heterogeneous traffic patterns across regions through a clustering mechanism and trains personalized models for each region to balance generalization and specialization. To bridge the domain gap between numerical traffic data and language-based models, we introduce a prompt engineering scheme that embeds statistical traffic features as structured inputs. Furthermore, we design a DeepSeek module that enables spatial alignment via cross-domain attention, allowing the LLM to leverage information from spatially related regions. By fine-tuning only lightweight components while freezing core LLM layers, TIDES achieves efficient adaptation to domain-specific patterns without incurring excessive training overhead. Extensive experiments on real-world cellular traffic datasets demonstrate that TIDES significantly outperforms state-of-the-art baselines in both prediction accuracy and robustness. Our results indicate that integrating spatial awareness into LLM-based predictors is the key to unlocking scalable and intelligent network management in future 6G systems.
Authors: Rajeeb Thapa Chhetri, Zhixiong Chen, Saurab Thapa
Abstract: A fundamental limitation of supervised deep learning in high-dimensional tabular domains is "Generalization Collapse": models learn precise decision boundaries for known distributions but fail catastrophically when facing Out-of-Distribution (OOD) data. We hypothesize that this failure stems from the lack of topological constraints in the latent space, resulting in diffuse manifolds where novel anomalies remain statistically indistinguishable from benign data. To address this, we propose Latent Sculpting, a hierarchical two-stage representation learning framework. Stage 1 utilizes a hybrid 1D-CNN and Transformer Encoder trained with a novel Dual-Centroid Compactness Loss (DCCL) to actively "sculpt" benign traffic into a low-entropy, hyperspherical cluster. Unlike standard contrastive losses that rely on triplet mining, DCCL optimizes global cluster centroids to enforce absolute manifold density. Stage 2 conditions a Masked Autoregressive Flow (MAF) on this pre-structured manifold to learn an exact density estimate. We evaluate this methodology on the rigorous CIC-IDS-2017 benchmark, treating it as a proxy for complex, non-stationary data streams. Empirical results demonstrate that explicit manifold sculpting is a prerequisite for robust zero-shot generalization. While supervised baselines suffered catastrophic performance collapse on unseen distribution shifts (F1 approx 0.30) and the strongest unsupervised baseline achieved only 0.76, our framework achieved an F1-Score of 0.87 on strictly zero-shot anomalies. Notably, we report an 88.89% detection rate on "Infiltration" scenarios--a complex distributional shift where state-of-the-art supervised models achieved 0.00% accuracy. These findings suggest that decoupling structure learning from density estimation provides a scalable path toward generalized anomaly detection.
Authors: Vishnu Mohan
Abstract: Tennis strategy optimization is a challenging sequential decision-making problem involving hierarchical scoring, stochastic outcomes, long-horizon credit assignment, physical fatigue, and adaptation to opponent skill. I present a reinforcement learning framework that integrates a custom tennis simulation environment with a Dueling Double Deep Q-Network(DDQN) trained using curriculum learning. The environment models complete tennis scoring at the level of points, games, and sets, rally-level tactical decisions across ten discrete action categories, symmetric fatigue dynamics, and a continuous opponent skill parameter. The dueling architecture decomposes action-value estimation into state-value and advantage components, while double Q-learning reduces overestimation bias and improves training stability in this long-horizon stochastic domain. Curriculum learning progressively increases opponent difficulty from 0.40 to 0.50, enabling robust skill acquisition without the training collapse observed under fixed opponents. Across extensive evaluations, the trained agent achieves win rates between 98 and 100 percent against balanced opponents and maintains strong performance against more challenging opponents. Serve efficiency ranges from 63.0 to 67.5 percent, and return efficiency ranges from 52.8 to 57.1 percent. Ablation studies demonstrate that both the dueling architecture and curriculum learning are necessary for stable convergence, while a standard DQN baseline fails to learn effective policies. Despite strong performance, tactical analysis reveals a pronounced defensive bias, with the learned policy prioritizing error avoidance and prolonged rallies over aggressive point construction. These results highlight a limitation of win-rate driven optimization in simplified sports simulations and emphasize the importance of reward design for realistic sports reinforcement learning.
Authors: Jose I. Aizpurua
Abstract: The integration of physics-based knowledge with machine learning models is increasingly shaping the monitoring, diagnostics, and prognostics of electrical transformers. In this two-part series, the first paper introduced the foundations of Neural Networks (NNs) and their variants for health assessment tasks. This second paper focuses on integrating physics and uncertainty into the learning process. We begin with the fundamentals of Physics-Informed Neural Networks (PINNs), applied to spatiotemporal thermal modeling and solid insulation ageing. Building on this, we present Bayesian PINNs as a principled framework to quantify epistemic uncertainty and deliver robust predictions under sparse data. Finally, we outline emerging research directions that highlight the potential of physics-aware and trustworthy machine learning for critical power assets.
Authors: Jose I. Aizpurua
Abstract: Power transformers are critical assets in power networks, whose reliability directly impacts grid resilience and stability. Traditional condition monitoring approaches, often rule-based or purely physics-based, struggle with uncertainty, limited data availability, and the complexity of modern operating conditions. Recent advances in machine learning (ML) provide powerful tools to complement and extend these methods, enabling more accurate diagnostics, prognostics, and control. In this two-part series, we examine the role of Neural Networks (NNs) and their extensions in transformer condition monitoring and health management tasks. This first paper introduces the basic concepts of NNs, explores Convolutional Neural Networks (CNNs) for condition monitoring using diverse data modalities, and discusses the integration of NN concepts within the Reinforcement Learning (RL) paradigm for decision-making and control. Finally, perspectives on emerging research directions are also provided.
Authors: Jiahao Lu
Abstract: Regularization techniques such as L2 regularization (Weight Decay) and Dropout are fundamental to training deep neural networks, yet their underlying physical mechanisms regarding feature frequency selection remain poorly understood. In this work, we investigate the Spectral Bias of modern Convolutional Neural Networks (CNNs). We introduce a Visual Diagnostic Framework to track the dynamic evolution of weight frequencies during training and propose a novel metric, the Spectral Suppression Ratio (SSR), to quantify the "low-pass filtering" intensity of different regularizers. By addressing the aliasing issue in small kernels (e.g., 3x3) through discrete radial profiling, our empirical results on ResNet-18 and CIFAR-10 demonstrate that L2 regularization suppresses high-frequency energy accumulation by over 3x compared to unregularized baselines. Furthermore, we reveal a critical Accuracy-Robustness Trade-off: while L2 models are sensitive to broadband Gaussian noise due to over-specialization in low frequencies, they exhibit superior robustness against high-frequency information loss (e.g., low resolution), outperforming baselines by >6% in blurred scenarios. This work provides a signal-processing perspective on generalization, confirming that regularization enforces a strong spectral inductive bias towards low-frequency structures.
Authors: Dhruv Tiwari
Abstract: The ruling method in modern Artificial Intelligence spanning from Deep Reinforcement Learning (DRL) to Large Language Models (LLMs) relies on a surge of static, externally defined reward functions. While this "extrinsic maximization" approach has rendered superhuman performance in closed, stationary fields, it produces agents that are fragile in open-ended, real-world environments. Standard agents lack internal autonomy: they struggle to explore without dense feedback, fail to adapt to distribution shifts (non-stationarity), and require extensive manual tuning of static hyperparameters. This paper proposes that the unaddressed factor in robust autonomy is a functional analog to biological emotion, serving as a high-level homeostatic control mechanism. We introduce Emotion-Inspired Learning Signals (EILS), a unified framework that replaces scattered optimization heuristics with a coherent, bio-inspired internal feedback engine. Unlike traditional methods that treat emotions as semantic labels, EILS models them as continuous, homeostatic appraisal signals such as Curiosity, Stress, and Confidence. We formalize these signals as vector-valued internal states derived from interaction history. These states dynamically modulate the agent's optimization landscape in real time: curiosity regulates entropy to prevent mode collapse, stress modulates plasticity to overcome inactivity, and confidence adapts trust regions to stabilize convergence. We hypothesize that this closed-loop homeostatic regulation can enable EILS agents to outperform standard baselines in terms of sample efficiency and non-stationary adaptation.
Authors: Xiaowei Wang
Abstract: Since transformer was firstly published in 2017, several works have been proposed to optimize it. However, the major structure of transformer remains unchanged, ignoring one of its main intrinsic limitations, which is the same static value is used for every query in a head. Transformer itself tries to solve this problem by implementing multi-head attentions, yet the number of heads is limited by complexity. I propose a method to decide a value for each query dynamically, which could cut down all the redundant heads, keeping only one. Consequently, the following feed forward network could be cut down entirely, as each revised embedding has already fetched enough useful values far beyond the context. As a result, a single-head Dynamic Value Attention (DVA) is all you need in a transformer. According to the experiment, DVA may save 37.6% training time than the original transformer meanwhile increasing the learning capability.
Authors: Jeffrey T. H. Wong, Cheng Zhang, Louis Mahon, Wayne Luk, Anton Isopoussu, Yiren Zhao
Abstract: Attention sinks are tokens, often the beginning-of-sequence (BOS) token, that receive disproportionately high attention despite limited semantic relevance. In this work, we identify a class of attention sinks, which we term secondary sinks, that differ fundamentally from the sinks studied in prior works, which we term primary sinks. While prior works have identified that tokens other than BOS can sometimes become sinks, they were found to exhibit properties analogous to the BOS token. Specifically, they emerge at the same layer, persist throughout the network and draw a large amount of attention mass. Whereas, we find the existence of secondary sinks that arise primarily in middle layers and can persist for a variable number of layers, and draw a smaller, but still significant, amount of attention mass. Through extensive experiments across 11 model families, we analyze where these secondary sinks appear, their properties, how they are formed, and their impact on the attention mechanism. Specifically, we show that: (1) these sinks are formed by specific middle-layer MLP modules; these MLPs map token representations to vectors that align with the direction of the primary sink of that layer. (2) The $\ell_2$-norm of these vectors determines the sink score of the secondary sink, and also the number of layers it lasts for, thereby leading to different impacts on the attention mechanisms accordingly. (3) The primary sink weakens in middle layers, coinciding with the emergence of secondary sinks. We observe that in larger-scale models, the location and lifetime of the sinks, together referred to as sink levels, appear in a more deterministic and frequent manner. Specifically, we identify three sink levels in QwQ-32B and six levels in Qwen3-14B.
Authors: Soroush Vahidi
Abstract: Graph neural networks (GNNs) achieve strong performance on homophilic graphs but often struggle under heterophily, where adjacent nodes frequently belong to different classes. We propose an interpretable and adaptive framework for semi-supervised node classification based on explicit combinatorial inference rather than deep message passing. Our method assigns labels using a confidence-ordered greedy procedure driven by an additive scoring function that integrates class priors, neighborhood statistics, feature similarity, and training-derived label-label compatibility. A small set of transparent hyperparameters controls the relative influence of these components, enabling smooth adaptation between homophilic and heterophilic regimes. We further introduce a validation-gated hybrid strategy in which combinatorial predictions are optionally injected as priors into a lightweight neural model. Hybrid refinement is applied only when it improves validation performance, preserving interpretability when neuralization is unnecessary. All adaptation signals are computed strictly from training data, ensuring a leakage-free evaluation protocol. Experiments on heterophilic and transitional benchmarks demonstrate competitive performance with modern GNNs while offering advantages in interpretability, tunability, and computational efficiency.
Authors: Gnankan Landry Regis N'guessan
Abstract: Standard neural network architectures employ fixed activation functions (ReLU, tanh, sigmoid) that are poorly suited for approximating functions with singular or fractional power behavior, a structure that arises ubiquitously in physics, including boundary layers, fracture mechanics, and corner singularities. We introduce M\"untz-Sz\'asz Networks (MSN), a novel architecture that replaces fixed smooth activations with learnable fractional power bases grounded in classical approximation theory. Each MSN edge computes $\phi(x) = \sum_k a_k |x|^{\mu_k} + \sum_k b_k \mathrm{sign}(x)|x|^{\lambda_k}$, where the exponents $\{\mu_k, \lambda_k\}$ are learned alongside the coefficients. We prove that MSN inherits universal approximation from the M\"untz-Sz\'asz theorem and establish novel approximation rates: for functions of the form $|x|^\alpha$, MSN achieves error $\mathcal{O}(|\mu - \alpha|^2)$ with a single learned exponent, whereas standard MLPs require $\mathcal{O}(\epsilon^{-1/\alpha})$ neurons for comparable accuracy. On supervised regression with singular target functions, MSN achieves 5-8x lower error than MLPs with 10x fewer parameters. Physics-informed neural networks (PINNs) represent a particularly demanding application for singular function approximation; on PINN benchmarks including a singular ODE and stiff boundary-layer problems, MSN achieves 3-6x improvement while learning interpretable exponents that match the known solution structure. Our results demonstrate that theory-guided architectural design can yield dramatic improvements for scientifically-motivated function classes.
Authors: Shaghayegh Shajarian, Kennedy Marsh, James Benson, Sajad Khorsandroo, Mahmoud Abdelsalam
Abstract: Modern networks generate vast, heterogeneous traffic that must be continuously analyzed for security and performance. Traditional network traffic analysis systems, whether rule-based or machine learning-driven, often suffer from high false positives and lack interpretability, limiting analyst trust. In this paper, we present ReGAIN, a multi-stage framework that combines traffic summarization, retrieval-augmented generation (RAG), and Large Language Model (LLM) reasoning for transparent and accurate network traffic analysis. ReGAIN creates natural-language summaries from network traffic, embeds them into a multi-collection vector database, and utilizes a hierarchical retrieval pipeline to ground LLM responses with evidence citations. The pipeline features metadata-based filtering, MMR sampling, a two-stage cross-encoder reranking mechanism, and an abstention mechanism to reduce hallucinations and ensure grounded reasoning. Evaluated on ICMP ping flood and TCP SYN flood traces from the real-world traffic dataset, it demonstrates robust performance, achieving accuracy between 95.95% and 98.82% across different attack types and evaluation benchmarks. These results are validated against two complementary sources: dataset ground truth and human expert assessments. ReGAIN also outperforms rule-based, classical ML, and deep learning baselines while providing unique explainability through trustworthy, verifiable responses.
Authors: Ying Zhu, Jiaxin Wan, Xiaoran Liu, Siyanag He, Qiqi Wang, Xu Guo, Tianyi Liang, Zengfeng Huang, Ziwei He, Xipeng Qiu
Abstract: Diffusion Language Models (dLLMs) have emerged as promising alternatives to Auto-Regressive (AR) models. While recent efforts have validated their pre-training potential and accelerated inference speeds, the post-training landscape for dLLMs remains underdeveloped. Existing methods suffer from computational inefficiency and objective mismatches between training and inference, severely limiting performance on complex reasoning tasks such as mathematics. To address this, we introduce DiRL, an efficient post-training framework that tightly integrates FlexAttention-accelerated blockwise training with LMDeploy-optimized inference. This architecture enables a streamlined online model update loop, facilitating efficient two-stage post-training (Supervised Fine-Tuning followed by Reinforcement Learning). Building on this framework, we propose DiPO, the first unbiased Group Relative Policy Optimization (GRPO) implementation tailored for dLLMs. We validate our approach by training DiRL-8B-Instruct on high-quality math data. Our model achieves state-of-the-art math performance among dLLMs and surpasses comparable models in the Qwen2.5 series on several benchmarks.
Authors: Byung-Kwan Lee, Yu-Chiang Frank Wang, Ryo Hachiuma
Abstract: Large-scale vision-language models (VLMs) have recently achieved remarkable multimodal understanding, but their massive size makes them impractical for deployment on mobile or edge devices. This raises the need for compact yet capable VLMs that can efficiently learn from powerful large teachers. However, distilling knowledge from a large teacher to a small student remains challenging due to their large size gap: the student often fails to reproduce the teacher's complex, high-dimensional representations, leading to unstable learning and degraded performance. To address this, we propose Masters (Masking Teacher and Reinforcing Student), a mask-progressive reinforcement learning (RL) distillation framework. Masters first masks non-dominant weights of the teacher to reduce unnecessary complexity, then progressively restores the teacher by gradually increasing its capacity during training. This strategy allows the student to learn richer representations from the teacher in a smooth and stable manner. To further refine knowledge transfer, Masters integrates an offline RL stage with two complementary rewards: an accuracy reward that measures the correctness of the generated responses, and a distillation reward that quantifies the ease of transferring responses from teacher to student. Unlike online think-answer RL paradigms that are computationally expensive and generate lengthy responses, our offline RL leverages pre-generated responses from masked teachers. These provide rich yet efficient guidance, enabling students to achieve strong performance without requiring the think-answer process.
Authors: Chama Bensmail
Abstract: Machine learning models are primarily judged by predictive performance, especially in applied settings. Once a model reaches high accuracy, its explanation is often assumed to be correct and trustworthy. However, this assumption raises an overlooked question: when two models achieve high accuracy, do they rely on the same internal logic, or do they reach the same outcome via different -- and potentially competing -- mechanisms? We introduce EvoXplain, a diagnostic framework that measures the stability of model explanations across repeated training. Rather than analysing a single trained model, EvoXplain treats explanations as samples drawn from the stochastic optimisation process itself -- without aggregating predictions or constructing ensembles -- and examines whether these samples form a single coherent explanation or separate into multiple, distinct explanatory modes. We evaluate EvoXplain on the Breast Cancer and COMPAS datasets using two widely deployed model classes: Logistic Regression and Random Forests. Although all models achieve high predictive accuracy, their explanations frequently exhibit clear multimodality. Even models commonly assumed to be stable, such as Logistic Regression, can produce multiple well-separated explanatory basins under repeated training on the same data split. These differences are not explained by hyperparameter variation or simple performance trade-offs. EvoXplain does not attempt to select a 'correct' explanation. Instead, it makes explanatory instability visible and quantifiable, revealing when single-instance or averaged explanations obscure the existence of multiple underlying mechanisms. More broadly, EvoXplain reframes interpretability as a property of a model class under repeated instantiation, rather than of any single trained model.
Authors: Abdul Malik Al Mardhouf Al Saadi, Amrita Basak
Abstract: Accurate bead geometry prediction in laser-directed energy deposition (L-DED) is often hindered by the scarcity and heterogeneity of experimental datasets collected under different materials, machine configurations, and process parameters. To address this challenge, a cross-dataset knowledge transfer model based on meta-learning for predicting deposited track geometry in L-DED is proposed. Specifically, two gradient-based meta-learning algorithms, i.e., Model-Agnostic Meta-Learning (MAML) and Reptile, are investigated to enable rapid adaptation to new deposition conditions with limited data. The proposed framework is performed using multiple experimental datasets compiled from peer-reviewed literature and in-house experiments and evaluated across powder-fed, wire-fed, and hybrid wire-powder L-DED processes. Results show that both MAML and Reptile achieve accurate bead height predictions on unseen target tasks using as few as three to nine training examples, consistently outperforming conventional feedforward neural networks trained under comparable data constraints. Across multiple target tasks representing different printing conditions, the meta-learning models achieve strong generalization performance, with R-squared values reaching up to approximately 0.9 and mean absolute errors between 0.03-0.08 mm, demonstrating effective knowledge transfer across heterogeneous L-DED settings.
Authors: Shaurya Gaur, Michel Vitale, Alessa Hering, Johan Kwisthout, Colin Jacobs, Lena Philipp, Fennie van der Graaf
Abstract: Lung cancer is the leading cause of cancer-related mortality in adults worldwide. Screening high-risk individuals with annual low-dose CT (LDCT) can support earlier detection and reduce deaths, but widespread implementation may strain the already limited radiology workforce. AI models have shown potential in estimating lung cancer risk from LDCT scans. However, high-risk populations for lung cancer are diverse, and these models' performance across demographic groups remains an open question. In this study, we drew on the considerations on confounding factors and ethically significant biases outlined in the JustEFAB framework to evaluate potential performance disparities and fairness in two deep learning risk estimation models for lung cancer screening: the Sybil lung cancer risk model and the Venkadesh21 nodule risk estimator. We also examined disparities in the PanCan2b logistic regression model recommended in the British Thoracic Society nodule management guideline. Both deep learning models were trained on data from the US-based National Lung Screening Trial (NLST), and assessed on a held-out NLST validation set. We evaluated AUROC, sensitivity, and specificity across demographic subgroups, and explored potential confounding from clinical risk factors. We observed a statistically significant AUROC difference in Sybil's performance between women (0.88, 95% CI: 0.86, 0.90) and men (0.81, 95% CI: 0.78, 0.84, p < .001). At 90% specificity, Venkadesh21 showed lower sensitivity for Black (0.39, 95% CI: 0.23, 0.59) than White participants (0.69, 95% CI: 0.65, 0.73). These differences were not explained by available clinical confounders and thus may be classified as unfair biases according to JustEFAB. Our findings highlight the importance of improving and monitoring model performance across underrepresented subgroups, and further research on algorithmic fairness, in lung cancer screening.
Authors: Alan Inglis, Fiona Doohan, Subramani Natarajan, Breige McNulty, Chris Elliott, Anne Nugent, Julie Meneely, Brett Greer, Stephen Kildea, Diana Bucur, Martin Danaher, Melissa Di Rocco, Lisa Black, Adam Gauley, Naoise McKenna, Andrew Parnell
Abstract: Mycotoxin contamination poses a significant risk to cereal crop quality, food safety, and agricultural productivity. Accurate prediction of mycotoxin levels can support early intervention strategies and reduce economic losses. This study investigates the use of neural networks and transfer learning models to predict mycotoxin contamination in Irish oat crops as a multi-response prediction task. Our dataset comprises oat samples collected in Ireland, containing a mix of environmental, agronomic, and geographical predictors. Five modelling approaches were evaluated: a baseline multilayer perceptron (MLP), an MLP with pre-training, and three transfer learning models; TabPFN, TabNet, and FT-Transformer. Model performance was evaluated using regression (RMSE, $R^2$) and classification (AUC, F1) metrics, with results reported per toxin and on average. Additionally, permutation-based variable importance analysis was conducted to identify the most influential predictors across both prediction tasks. The transfer learning approach TabPFN provided the overall best performance, followed by the baseline MLP. Our variable importance analysis revealed that weather history patterns in the 90-day pre-harvest period were the most important predictors, alongside seed moisture content.
Authors: Bhaktipriya Radharapu, Eshika Saxena, Kenneth Li, Chenxi Whitehouse, Adina Williams, Nicola Cancedda
Abstract: As LLM-based judges become integral to industry applications, obtaining well-calibrated uncertainty estimates efficiently has become critical for production deployment. However, existing techniques, such as verbalized confidence and multi-generation methods, are often either poorly calibrated or computationally expensive. We introduce linear probes trained with a Brier score-based loss to provide calibrated uncertainty estimates from reasoning judges' hidden states, requiring no additional model training. We evaluate our approach on both objective tasks (reasoning, mathematics, factuality, coding) and subjective human preference judgments. Our results demonstrate that probes achieve superior calibration compared to existing methods with $\approx10$x computational savings, generalize robustly to unseen evaluation domains, and deliver higher accuracy on high-confidence predictions. However, probes produce conservative estimates that underperform on easier datasets but may benefit safety-critical deployments prioritizing low false-positive rates. Overall, our work demonstrates that interpretability-based uncertainty estimation provides a practical and scalable plug-and-play solution for LLM judges in production.
Authors: George Bird
Abstract: A systematic mismatch exists between mathematically ideal and effective activation updates during gradient descent. As intended, parameters update in their direction of steepest descent. However, activations are argued to constitute a more directly impactful quantity to prioritise in optimisation, as they are closer to the loss in the computational graph and carry sample-dependent information through the network. Yet their propagated updates do not take the optimal steepest-descent step. These quantities exhibit non-ideal sample-wise scaling across affine, convolutional, and attention layers. Solutions to correct for this are trivial and, entirely incidentally, derive normalisation from first principles despite motivational independence. Consequently, such considerations offer a fresh and conceptual reframe of normalisation's action, with auxiliary experiments bolstering this mechanistically. Moreover, this analysis makes clear a second possibility: a solution that is functionally distinct from modern normalisations, without scale-invariance, yet remains empirically successful, outperforming conventional normalisers across several tests. This is presented as an alternative to the affine map. This generalises to convolution via a new functional form, "PatchNorm", a compositionally inseparable normaliser. Together, these provide an alternative mechanistic framework that adds to, and counters some of, the discussion of normalisation. Further, it is argued that normalisers are better decomposed into activation-function-like maps with parameterised scaling, thereby aiding the prioritisation of representations during optimisation. Overall, this constitutes a theoretical-principled approach that yields several new functions that are empirically validated and raises questions about the affine + nonlinear approach to model creation.
Authors: Rohit Pandey, Rohan Pandey
Abstract: Accurate prediction of model rocket flight performance requires estimating aerodynamic parameters that are difficult to measure directly. Traditional approaches rely on computational fluid dynamics or empirical correlations, while data-driven methods require extensive real flight data that is expensive and time-consuming to collect. We present a simulation-based amortized inference approach that trains a neural network on synthetic flight data generated from a physics simulator, then applies the learned model to real flights without any fine-tuning. Our method learns to invert the forward physics model, directly predicting drag coefficient and thrust correction factor from a single apogee measurement combined with motor and configuration features. In this proof-of-concept study, we train on 10,000 synthetic flights and evaluate on 8 real flights, achieving a mean absolute error of 12.3 m in apogee prediction - demonstrating promising sim-to-real transfer with zero real training examples. Analysis reveals a systematic positive bias in predictions, providing quantitative insight into the gap between idealized physics and real-world flight conditions. We additionally compare against OpenRocket baseline predictions, showing that our learned approach reduces apogee prediction error. Our implementation is publicly available to support reproducibility and adoption in the amateur rocketry community.
Authors: Zheng Xing, Weibing Zhao
Abstract: Unsupervised human motion segmentation (HMS) can be effectively achieved using subspace clustering techniques. However, traditional methods overlook the role of temporal semantic exploration in HMS. This paper explores the use of temporal vision semantics (TVS) derived from human motion sequences, leveraging the image-to-text capabilities of a large language model (LLM) to enhance subspace clustering performance. The core idea is to extract textual motion information from consecutive frames via LLM and incorporate this learned information into the subspace clustering framework. The primary challenge lies in learning TVS from human motion sequences using LLM and integrating this information into subspace clustering. To address this, we determine whether consecutive frames depict the same motion by querying the LLM and subsequently learn temporal neighboring information based on its response. We then develop a TVS-integrated subspace clustering approach, incorporating subspace embedding with a temporal regularizer that induces each frame to share similar subspace embeddings with its temporal neighbors. Additionally, segmentation is performed based on subspace embedding with a temporal constraint that induces the grouping of each frame with its temporal neighbors. We also introduce a feedback-enabled framework that continuously optimizes subspace embedding based on the segmentation output. Experimental results demonstrate that the proposed method outperforms existing state-of-the-art approaches on four benchmark human motion datasets.
Authors: Pascal Passigan, Kevin zhu, Angelina Ning
Abstract: Understanding how small molecules perturb gene expression is essential for uncovering drug mechanisms, predicting off-target effects, and identifying repurposing opportunities. While prior deep learning frameworks have integrated multimodal embeddings into biomedical knowledge graphs (BKGs) and further improved these representations through graph neural network message-passing paradigms, these models have been applied to tasks such as link prediction and binary drug-disease association, rather than the task of gene perturbation, which may unveil more about mechanistic transcriptomic effects. To address this gap, we construct a merged biomedical graph that integrates (i) PrimeKG++, an augmentation of PrimeKG containing semantically rich embeddings for nodes with (ii) LINCS L1000 drug and cell line nodes, initialized with multimodal embeddings from foundation models such as MolFormerXL and BioBERT. Using this heterogeneous graph, we train a graph attention network (GAT) with a downstream prediction head that learns the delta expression profile of over 978 landmark genes for a given drug-cell pair. Our results show that our framework outperforms MLP baselines for differentially expressed genes (DEG) -- which predict the delta expression given a concatenated embedding of drug features, target features, and baseline cell expression -- under the scaffold and random splits. Ablation experiments with edge shuffling and node feature randomization further demonstrate that the edges provided by biomedical KGs enhance perturbation-level prediction. More broadly, our framework provides a path toward mechanistic drug modeling: moving beyond binary drug-disease association tasks to granular transcriptional effects of therapeutic intervention.
Authors: Huashen Lu, Wensheng Gan, Guoting Chen, Zhichao Huang, Philip S. Yu
Abstract: Graph neural networks (GNNs) have brought revolutionary advancements to the field of link prediction (LP), providing powerful tools for mining potential relationships in graphs. However, existing methods face challenges when dealing with large-scale sparse graphs and the need for a high degree of alignment between different datasets in transfer learning. Besides, although self-supervised methods have achieved remarkable success in many graph tasks, prior research has overlooked the potential of transfer learning to generalize across different graph datasets. To address these limitations, we propose a novel Graph Attention Adaptive Transfer Network (GAATNet). It combines the advantages of pre-training and fine-tuning to capture global node embedding information across datasets of different scales, ensuring efficient knowledge transfer and improved LP performance. To enhance the model's generalization ability and accelerate training, we design two key strategies: 1) Incorporate distant neighbor embeddings as biases in the self-attention module to capture global features. 2) Introduce a lightweight self-adapter module during fine-tuning to improve training efficiency. Comprehensive experiments on seven public datasets demonstrate that GAATNet achieves state-of-the-art performance in LP tasks. This study provides a general and scalable solution for LP tasks to effectively integrate GNNs with transfer learning. The source code and datasets are publicly available at https://github.com/DSI-Lab1/GAATNet
Authors: Daniil Burakov, Ivan Petrov, Dmitrii Khelimskii, Ivan Bessonov, Mikhail Lazarev
Abstract: Patient status, angiographic and procedural characteristics encode crucial signals for predicting long-term outcomes after percutaneous coronary intervention (PCI). The aim of the study was to develop a predictive model for assessing the risk of cardiac death based on the real and synthetic data of patients undergoing PCI and to identify the factors that have the greatest impact on mortality. We analyzed 2,044 patients, who underwent a PCI for bifurcation lesions. The primary outcome was cardiac death at 3-year follow-up. Several machine learning models were applied to predict three-year mortality after PCI. To address class imbalance and improve the representation of the minority class, an additional 500 synthetic samples were generated and added to the training set. To evaluate the contribution of individual features to model performance, we applied permutation feature importance. An additional experiment was conducted to evaluate how the model's predictions would change after removing non-informative features from the training and test datasets. Without oversampling, all models achieve high overall accuracy (0.92-0.93), yet they almost completely ignore the minority class. Across models, augmentation consistently increases minority-class recall with minimal loss of AUROC, improves probability quality, and yields more clinically reasonable risk estimates on the constructed severe profiles. According to feature importance analysis, four features emerged as the most influential: Age, Ejection Fraction, Peripheral Artery Disease, and Cerebrovascular Disease. These results show that straightforward augmentation with realistic and extreme cases can expose, quantify, and reduce brittleness in imbalanced clinical prediction using only tabular records, and motivate routine reporting of probability quality and stress tests alongside headline metrics.
Authors: Rahul D Ray
Abstract: Physics-constrained data generation is essential for machine learning in scientific domains where real data are scarce; however, existing approaches often over-constrain models without identifying which physical components are necessary. We present a systematic ablation study of a physics-informed grating coupler spectrum generator that maps five geometric parameters to 100-point spectral responses. By selectively removing explicit energy conservation enforcement, Fabry-Perot oscillations, bandwidth variation, and noise, we uncover a physics constraint paradox: explicit energy conservation enforcement is mathematically redundant when the underlying equations are physically consistent, with constrained and unconstrained variants achieving identical conservation accuracy (mean error approximately 7 x 10^-9). In contrast, Fabry-Perot oscillations dominate threshold-based bandwidth variability, accounting for a 72 percent reduction in half-maximum bandwidth spread when removed (with bandwidth spread reduced from 132.3 nm to 37.4 nm). We further identify a subtle pitfall: standard noise-addition-plus-renormalization pipelines introduce 0.5 percent unphysical negative absorption values. The generator operates at 200 samples per second, enabling high-throughput data generation and remaining orders of magnitude faster than typical full-wave solvers reported in the literature. Finally, downstream machine learning evaluation reveals a clear physics-learnability trade-off: while central wavelength prediction remains unaffected, removing Fabry-Perot oscillations improves bandwidth prediction accuracy by 31.3 percent in R-squared and reduces RMSE by 73.8 percent. These findings provide actionable guidance for physics-informed dataset design and highlight machine learning performance as a diagnostic tool for assessing constraint relevance.
Authors: Tzamn Melendez Carmona, Federico Marchesin, Marco P. Abrate, Peter Bienstman, Stefano Di Carlo, Alessandro Savino Senior
Abstract: PNNs present promising opportunities for accelerating machine learning by leveraging the unique benefits of photonic circuits. However, current state of the art PNN simulation tools face significant scalability challenges when training large-scale PNNs, due to the computational demands of transfer matrix calculations, resulting in high memory and time consumption. To overcome these limitations, we introduce the Slicing method, an efficient transfer matrix computation approach compatible with back-propagation. We integrate this method into LuxIA, a unified simulation and training framework. The Slicing method substantially reduces memory usage and execution time, enabling scalable simulation and training of large PNNs. Experimental evaluations across various photonic architectures and standard datasets, including MNIST, Digits, and Olivetti Faces, show that LuxIA consistently surpasses existing tools in speed and scalability. Our results advance the state of the art in PNN simulation, making it feasible to explore and optimize larger, more complex architectures. By addressing key computational bottlenecks, LuxIA facilitates broader adoption and accelerates innovation in AI hardware through photonic technologies. This work paves the way for more efficient and scalable photonic neural network research and development.
Authors: Bing Hao, Minglai Shao, Zengyi Wo, Yunlong Chu, Yuhang Liu, Ruijie Wang
Abstract: The widespread application of Large Language Models (LLMs) has motivated a growing interest in their capacity for processing dynamic graphs. Temporal motifs, as an elementary unit and important local property of dynamic graphs which can directly reflect anomalies and unique phenomena, are essential for understanding their evolutionary dynamics and structural features. However, leveraging LLMs for temporal motif analysis on dynamic graphs remains relatively unexplored. In this paper, we systematically study LLM performance on temporal motif-related tasks. Specifically, we propose a comprehensive benchmark, LLMTM (Large Language Models in Temporal Motifs), which includes six tailored tasks across nine temporal motif types. We then conduct extensive experiments to analyze the impacts of different prompting techniques and LLMs (including nine models: openPangu-7B, the DeepSeek-R1-Distill-Qwen series, Qwen2.5-32B-Instruct, GPT-4o-mini, DeepSeek-R1, and o3) on model performance. Informed by our benchmark findings, we develop a tool-augmented LLM agent that leverages precisely engineered prompts to solve these tasks with high accuracy. Nevertheless, the high accuracy of the agent incurs a substantial cost. To address this trade-off, we propose a simple yet effective structure-aware dispatcher that considers both the dynamic graph's structural properties and the LLM's cognitive load to intelligently dispatch queries between the standard LLM prompting and the more powerful agent. Our experiments demonstrate that the structure-aware dispatcher effectively maintains high accuracy while reducing cost.
Authors: Amgad Ahmed Ali Ibrahim, Hein Htet, Ryoji Asahi
Abstract: Radiation-induced grafting (RIG) enables precise functionalization of polymer films for ion-exchange membranes, CO2-separation membranes, and battery electrolytes by generating radicals on robust substrates to graft desired monomers. However, reproducibility remains limited due to unreported variability in base-film morphology (crystallinity, grain orientation, free volume), which governs monomer diffusion, radical distribution, and the Trommsdorff effect, leading to spatial graft gradients and performance inconsistencies. We present a hierarchical stacking optimization framework with a Dirichlet's Process (SoDip), a hierarchical data-driven framework integrating: (1) a decoder-only Transformer (DeepSeek-R1) to encode textual process descriptors (irradiation source, grafting type, substrate manufacturer); (2) TabNet and XGBoost for modelling multimodal feature interactions; (3) Gaussian Process Regression (GPR) with Dirichlet Process Mixture Models (DPMM) for uncertainty quantification and heteroscedasticity; and (4) Bayesian Optimization for efficient exploration of high-dimensional synthesis space. A diverse dataset was curated using ChemDataExtractor 2.0 and WebPlotDigitizer, incorporating numerical and textual variables across hundreds of RIG studies. In cross-validation, SoDip achieved ~33% improvement over GPR while providing calibrated confidence intervals that identify low-reproducibility regimes. Its stacked architecture integrates sparse textual and numerical inputs of varying quality, outperforming prior models and establishing a foundation for reproducible, morphology-aware design in graft polymerization research.
Authors: Varshith Gudur
Abstract: Modern AI systems rely on vector embeddings stored and searched using floating-point arithmetic. While effective for approximate similarity search, this design introduces fundamental non-determinism: identical models, inputs, and code can produce different memory states and retrieval results across hardware architectures (e.g., x86 vs. ARM). This prevents replayability and safe deployment, leading to silent data divergence that prevents post-hoc verification and compromises audit trails in regulated sectors. We present Valori, a deterministic AI memory substrate that replaces floating-point memory operations with fixed-point arithmetic (Q16.16) and models memory as a replayable state machine. Valori guarantees bit-identical memory states, snapshots, and search results across platforms. We demonstrate that non-determinism arises before indexing or retrieval and show how Valori enforces determinism at the memory boundary. Our results suggest that deterministic memory is a necessary primitive for trustworthy AI systems. The reference implementation is open-source and available at https://github.com/varshith-Git/Valori-Kernel (archived at https://zenodo.org/records/18022660).
URLs: https://github.com/varshith-Git/Valori-Kernel, https://zenodo.org/records/18022660).
Authors: Guokan Chen, Yao Xiao
Abstract: Physics-informed neural networks (PINNs) have led to significant advancements in scientific computing by integrating fundamental physical principles with advanced data-driven techniques. However, when dealing with problems characterized by multi-scale or high-frequency features, PINNs encounter persistent and severe challenges related to stiffness in gradient flow and spectral bias, which significantly limit their predictive capabilities. To address these issues, this paper proposes a Dynamic Balancing Adaptive Weighting Physics-Informed Kolmogorov-Arnold Network (DBAW-PIKAN), designed to mitigate such gradient-related failure modes and overcome the bottlenecks in function representation. The core of DBAW-PIKAN combines the Kolmogorov-Arnold network architecture, based on learnable B-splines, with an adaptive weighting strategy that incorporates a dynamic decay upper bound. Compared to baseline models, the proposed method accelerates the convergence process and improves solution accuracy by at least an order of magnitude without introducing additional computational complexity. A series of numerical benchmarks, including the Klein-Gordon, Burgers, and Helmholtz equations, demonstrate the significant advantages of DBAW-PIKAN in enhancing both accuracy and generalization performance.
Authors: Zikun Guoa, Adeyinka. P. Adedigbaa, Rammohan Mallipeddi
Abstract: Synthetic appliance data are essential for developing non-intrusive load monitoring algorithms and enabling privacy preserving energy research, yet the scarcity of labeled datasets remains a significant barrier. Recent GAN-based methods have demonstrated the feasibility of synthesizing load patterns, but most existing approaches treat all devices uniformly within a single model, neglecting the behavioral differences between intermittent and continuous appliances and resulting in unstable training and limited output fidelity. To address these limitations, we propose the Cluster Aggregated GAN framework, a hybrid generative approach that routes each appliance to a specialized branch based on its behavioral characteristics. For intermittent appliances, a clustering module groups similar activation patterns and allocates dedicated generators for each cluster, ensuring that both common and rare operational modes receive adequate modeling capacity. Continuous appliances follow a separate branch that employs an LSTM-based generator to capture gradual temporal evolution while maintaining training stability through sequence compression. Extensive experiments on the UVIC smart plug dataset demonstrate that the proposed framework consistently outperforms baseline methods across metrics measuring realism, diversity, and training stability, and that integrating clustering as an active generative component substantially improves both interpretability and scalability. These findings establish the proposed framework as an effective approach for synthetic load generation in non-intrusive load monitoring research.
Authors: Renping Zhou, Zanlin Ni, Tianyi Chen, Zeyu Liu, Yang Yue, Yulin Wang, Yuxuan Wang, Jingshu Liu, Gao Huang
Abstract: Recently, Masked Diffusion Models (MDMs) have shown promising potential across vision, language, and cross-modal generation. However, a notable discrepancy exists between their training and inference procedures. In particular, MDM inference is a multi-step, iterative process governed not only by the model itself but also by various schedules that dictate the token-decoding trajectory (e.g., how many tokens to decode at each step). In contrast, MDMs are typically trained using a simplified, single-step BERT-style objective that masks a subset of tokens and predicts all of them simultaneously. This step-level simplification fundamentally disconnects the training paradigm from the trajectory-level nature of inference, leaving the inference schedules never optimized during training. In this paper, we introduce Co-GRPO, which reformulates MDM generation as a unified Markov Decision Process (MDP) that jointly incorporates both the model and the inference schedule. By applying Group Relative Policy Optimization at the trajectory level, Co-GRPO cooperatively optimizes model parameters and schedule parameters under a shared reward, without requiring costly backpropagation through the multi-step generation process. This holistic optimization aligns training with inference more thoroughly and substantially improves generation quality. Empirical results across four benchmarks-ImageReward, HPS, GenEval, and DPG-Bench-demonstrate the effectiveness of our approach. For more details, please refer to our project page: https://co-grpo.github.io/ .
Authors: Arunkumar V, Nivethitha S, Sharan Srinivas, Gangadharan G. R
Abstract: A central question for the future of work is whether person centered management can survive when algorithms take on managerial roles. Standard tools often miss what is happening because worker responses to algorithmic systems are rarely linear. We use a Double Machine Learning framework to estimate a moderated mediation model without imposing restrictive functional forms. Using survey data from 464 gig workers, we find a clear nonmonotonic pattern. Supportive HR practices improve worker wellbeing, but their link to performance weakens in a murky middle where algorithmic oversight is present yet hard to interpret. The relationship strengthens again when oversight is transparent and explainable. These results show why simple linear specifications can miss the pattern and sometimes suggest the opposite conclusion. For platform design, the message is practical: control that is only partly defined creates confusion, but clear rules and credible recourse can make strong oversight workable. Methodologically, the paper shows how Double Machine Learning can be used to estimate conditional indirect effects in organizational research without forcing the data into a linear shape.
Authors: Qingyue Cao, Bo Jin, Changwei Gong, Xin Tong, Wenzheng Li, Xiaodong Zhou
Abstract: Graph anomaly detection technology has broad applications in financial fraud and risk control. However, existing graph anomaly detection methods often face significant challenges when dealing with complex and variable abnormal patterns, as anomalous nodes are often disguised and mixed with normal nodes, leading to the coexistence of homophily and heterophily in the graph domain. Recent spectral graph neural networks have made notable progress in addressing this issue; however, current techniques typically employ fixed, globally shared filters. This 'one-size-fits-all' approach can easily cause over-smoothing, erasing critical high-frequency signals needed for fraud detection, and lacks adaptive capabilities for different graph instances. To solve this problem, we propose a Multi-Head Spectral-Adaptive Graph Neural Network (MHSA-GNN). The core innovation is the design of a lightweight hypernetwork that, conditioned on a 'spectral fingerprint' containing structural statistics and Rayleigh quotient features, dynamically generates Chebyshev filter parameters tailored to each instance. This enables a customized filtering strategy for each node and its local subgraph. Additionally, to prevent mode collapse in the multi-head mechanism, we introduce a novel dual regularization strategy that combines teacher-student contrastive learning (TSC) to ensure representation accuracy and Barlow Twins diversity loss (BTD) to enforce orthogonality among heads. Extensive experiments on four real-world datasets demonstrate that our method effectively preserves high-frequency abnormal signals and significantly outperforms existing state-of-the-art methods, especially showing excellent robustness on highly heterogeneous datasets.
Authors: Tsogt-Ochir Enkhbayar
Abstract: Warning-framed content in training data (e.g., "DO NOT USE - this code is vulnerable") does not, it turns out, teach language models to avoid the warned-against behavior. In experiments reported here, models exposed to such warnings reproduced the flagged content at rates statistically indistinguishable from models given the content directly (76.7% vs. 83.3%). Why? Sparse autoencoder analysis points to a failure of orthogonalization: "describing X" and "performing X" activate overlapping latent features. Feature #8684, which tracks code execution patterns, fires at comparable magnitude in both warning and exploitation contexts. A related phenomenon, what I call "stealth slip", allows conversational preambles to rotate activations into subspaces that linear probes miss entirely. Prompting and inference-time steering do not fix this; training-time feature ablation does. The upshot is that statistical co-occurrence dominates over pragmatic interpretation in current architectures. Models learn what tends to follow a context, not why it appeared there.
Authors: Reda Heddad, Lamiae Bouanane
Abstract: The Mixture-of-Experts (MoE) architecture has emerged as a powerful paradigm for scaling deep learning models, yet it is fundamentally limited by challenges such as expert imbalance and the computational complexity of classical routing mechanisms. This paper investigates the potential of Quantum Machine Learning (QML) to address these limitations through a novel Hybrid Quantum-Classical Mixture of Experts (QMoE) architecture. Specifically, we conduct an ablation study using a Quantum Gating Network (Router) combined with classical experts to isolate the source of quantum advantage. Our central finding validates the Interference Hypothesis: by leveraging quantum feature maps (Angle Embedding) and wave interference, the Quantum Router acts as a high-dimensional kernel method, enabling the modeling of complex, non-linear decision boundaries with superior parameter efficiency compared to its classical counterparts. Experimental results on non-linearly separable data, such as the Two Moons dataset, demonstrate that the Quantum Router achieves a significant topological advantage, effectively "untangling" data distributions that linear classical routers fail to separate efficiently. Furthermore, we analyze the architecture's robustness against simulated quantum noise, confirming its feasibility for near-term intermediate-scale quantum (NISQ) hardware. We discuss practical applications in federated learning, privacy-preserving machine learning, and adaptive systems that could benefit from this quantum-enhanced routing paradigm.
Authors: Jennifer Sawyer, Julian Allagan
Abstract: This study extends previous hotspot and Chi-Square analysis by Sawyer \cite{sawyer2025hotspot} by integrating advanced statistical analysis, machine learning, and spatial modeling techniques to analyze five years (2019--2023) of traffic accident data from an 8.4-mile stretch of US 158 in Currituck County, NC. Building upon foundational statistical work, we apply Kernel Density Estimation (KDE), Negative Binomial Regression, Random Forest classification, and Highway Safety Manual (HSM) Safety Performance Function (SPF) comparisons to identify comprehensive temporal and spatial crash patterns. A Random Forest classifier predicts injury severity with 67\% accuracy, outperforming HSM SPF. Spatial clustering is confirmed via Moran's I test ($I = 0.32$, $p < 0.001$), and KDE analysis reveals hotspots near major intersections, validating and extending earlier hotspot identification methods. These results support targeted interventions to improve traffic safety on this vital transportation corridor. Our objective is to provide actionable insights for improving safety on US 158 while contributing to the broader understanding of rural highway safety analysis through methodological advancement beyond basic statistical techniques.
Authors: Sultan Amed, Chan Yu Hang, Sayantan Banerjee
Abstract: This paper presents PDx, an adaptive, machine learning operations (MLOps) driven decision system for forecasting credit risk using probability of default (PD) modeling in digital lending. While conventional PD models prioritize predictive accuracy during model development with complex machine learning algorithms, they often overlook continuous adaptation to changing borrower behaviour, resulting in static models that degrade over time in production and generate inaccurate default predictions. Many financial institutes also find it difficult transitioning ML models from development environment to production and maintaining their health. With PDx we aimed to addresses these limitations using a dynamic, end-to-end model lifecycle management approach that integrates continuous model monitoring, retraining, and validation through a robust MLOps pipeline. We introduced a dynamic champion-challenger framework for PDx to regularly update baseline models to recalibrate independent parameters with the latest data and select the best-performing model through out-of-time validation, ensuring resilience against data drift and changing credit risk patterns. Our empirical analysis shows that decision tree-based ensemble models consistently outperform others in classifying defaulters but require frequent updates to sustain performance. Linear models (e.g., logistic regression) and neural networks exhibit greater performance degradation. The study demonstrate with PDx we can mitigates value erosion for digital lenders, particularly in short-term, small-ticket loans, where borrower behavior shifts rapidly. We have validated the effectiveness of PDx using datasets from peer-to-peer lending, business loans, and auto loans, demonstrating its scalability and adaptability for modern credit risk forecasting.
Authors: Zehao Chen, Tianxiang Ai, Yifei Li, Gongxun Li, Yuyang Wei, Wang Zhou, Guanghui Li, Bin Yu, Zhijun Chen, Hailong Sun, Fuzhen Zhuang, Jianxin Li, Deqing Wang, Yikun Ban
Abstract: Ensemble learning of LLMs has emerged as a promising alternative to enhance performance, but existing approaches typically treat models as black boxes, combining the inputs or final outputs while overlooking the rich internal representations and interactions across models.In this work, we introduce LLMBoost, a novel ensemble fine-tuning framework that breaks this barrier by explicitly leveraging intermediate states of LLMs. Inspired by the boosting paradigm, LLMBoost incorporates three key innovations. First, a cross-model attention mechanism enables successor models to access and fuse hidden states from predecessors, facilitating hierarchical error correction and knowledge transfer. Second, a chain training paradigm progressively fine-tunes connected models with an error-suppression objective, ensuring that each model rectifies the mispredictions of its predecessor with minimal additional computation. Third, a near-parallel inference paradigm design pipelines hidden states across models layer by layer, achieving inference efficiency approaching single-model decoding. We further establish the theoretical foundations of LLMBoost, proving that sequential integration guarantees monotonic improvements under bounded correction assumptions. Extensive experiments on commonsense reasoning and arithmetic reasoning tasks demonstrate that LLMBoost consistently boosts accuracy while reducing inference latency.
Authors: Wenzhang Du
Abstract: Closed-loop decision-making systems (e.g., lending, screening, or recidivism risk assessment) often operate under fairness and service constraints while inducing feedback effects: decisions change who appears in the future, yielding non-stationary data and potentially amplifying disparities. We study online learning of a one-dimensional threshold policy from bandit feedback under demographic parity (DP) and, optionally, service-rate constraints. The learner observes only a scalar score each round and selects a threshold; reward and constraint residuals are revealed only for the chosen threshold. We propose Optimistic Feasible Search (OFS), a simple grid-based method that maintains confidence bounds for reward and constraint residuals for each candidate threshold. At each round, OFS selects a threshold that appears feasible under confidence bounds and, among those, maximizes optimistic reward; if no threshold appears feasible, OFS selects the threshold minimizing optimistic constraint violation. This design directly targets feasible high-utility thresholds and is particularly effective for low-dimensional, interpretable policy classes where discretization is natural. We evaluate OFS on (i) a synthetic closed-loop benchmark with stable contraction dynamics and (ii) two semi-synthetic closed-loop benchmarks grounded in German Credit and COMPAS, constructed by training a score model and feeding group-dependent acceptance decisions back into population composition. Across all environments, OFS achieves higher reward with smaller cumulative constraint violation than unconstrained and primal-dual bandit baselines, and is near-oracle relative to the best feasible fixed threshold under the same sweep procedure. Experiments are reproducible and organized with double-blind-friendly relative outputs.
Authors: Xudong Ling, Tianxi Huang, Qian Dong, Tao He, Chaorong Li, Guiduo Duan
Abstract: Short-term precipitation nowcasting is an inherently uncertain and under-constrained spatiotemporal forecasting problem, especially for rapidly evolving and extreme weather events. Existing generative approaches rely primarily on visual conditioning, leaving future motion weakly constrained and ambiguous. We propose a language-aware multimodal nowcasting framework(LangPrecip) that treats meteorological text as a semantic motion constraint on precipitation evolution. By formulating nowcasting as a semantically constrained trajectory generation problem under the Rectified Flow paradigm, our method enables efficient and physically consistent integration of textual and radar information in latent space.We further introduce LangPrecip-160k, a large-scale multimodal dataset with 160k paired radar sequences and motion descriptions. Experiments on Swedish and MRMS datasets show consistent improvements over state-of-the-art methods, achieving over 60 \% and 19\% gains in heavy-rainfall CSI at an 80-minute lead time.
Authors: Chorok Lee
Abstract: Probabilistic knowledge graph embeddings represent entities as distributions, using learned variances to quantify epistemic uncertainty. We identify a fundamental limitation: these variances are relation-agnostic, meaning an entity receives identical uncertainty regardless of relational context. This conflates two distinct out-of-distribution phenomena that behave oppositely: emerging entities (rare, poorly-learned) and novel relational contexts (familiar entities in unobserved relationships). We prove an impossibility result: any uncertainty estimator using only entity-level statistics independent of relation context achieves near-random OOD detection on novel contexts. We empirically validate this on three datasets, finding 100 percent of novel-context triples have frequency-matched in-distribution counterparts. This explains why existing probabilistic methods achieve 0.99 AUROC on random corruptions but only 0.52-0.64 on temporal distribution shift. We formalize uncertainty decomposition into complementary components: semantic uncertainty from entity embedding variance (detecting emerging entities) and structural uncertainty from entity-relation co-occurrence (detecting novel contexts). Our main theoretical result proves these signals are non-redundant, and that any convex combination strictly dominates either signal alone. Our method (CAGP) combines semantic and structural uncertainty via learned weights, achieving 0.94-0.99 AUROC on temporal OOD detection across multiple benchmarks, a 60-80 percent relative improvement over relation-agnostic baselines. Empirical validation confirms complete frequency overlap on three datasets (FB15k-237, WN18RR, YAGO3-10). On selective prediction, our method reduces errors by 43 percent at 85 percent answer rate.
Authors: Sravan Karthick T
Abstract: Bitcoin price forecasting is characterized by extreme volatility and non-stationarity, often defying traditional univariate time-series models over long horizons. This paper addresses a critical gap by integrating Global M2 Liquidity, aggregated from 18 major economies, as a leading exogenous variable with a 12-week lag structure. Using the TimeXer architecture, we compare a liquidity-conditioned forecasting model (TimeXer-Exog) against state-of-the-art benchmarks including LSTM, N-BEATS, PatchTST, and a standard univariate TimeXer. Experiments conducted on daily Bitcoin price data from January 2020 to August 2025 demonstrate that explicit macroeconomic conditioning significantly stabilizes long-horizon forecasts. At a 70-day forecast horizon, the proposed TimeXer-Exog model achieves a mean squared error (MSE) 1.08e8, outperforming the univariate TimeXer baseline by over 89 percent. These results highlight that conditioning deep learning models on global liquidity provides substantial improvements in long-horizon Bitcoin price forecasting.
Authors: Matthew Riemer, Erik Miehling, Miao Liu, Djallel Bouneffouf, Murray Campbell
Abstract: Although parameter-efficient fine-tuning methods, such as LoRA, only modify a small subset of parameters, they can have a significant impact on the model. Our instruction-tuning experiments show that LoRA-based supervised fine-tuning can catastrophically degrade model capabilities, even when trained on very small datasets for relatively few steps. With that said, we demonstrate that while the most straightforward approach (that is likely the most used in practice) fails spectacularly, small tweaks to the training procedure with very little overhead can virtually eliminate the problem. Particularly, in this paper we consider a regularized approximate replay approach which penalizes KL divergence with respect to the initial model and interleaves in data for next token prediction from a different, yet similar, open access corpus to what was used in pre-training. When applied to Qwen instruction-tuned models, we find that this recipe preserves general knowledge in the model without hindering plasticity to new tasks by adding a modest amount of computational overhead.
Authors: Bruno Mlodozeniec, Pierre Ablin, Louis B\'ethune, Dan Busbridge, Michal Klein, Jason Ramapuram, Marco Cuturi
Abstract: Hyperparameter tuning can dramatically impact training stability and final performance of large-scale models. Recent works on neural network parameterisations, such as $\mu$P, have enabled transfer of optimal global hyperparameters across model sizes. These works propose an empirical practice of search for optimal global base hyperparameters at a small model size, and transfer to a large size. We extend these works in two key ways. To handle scaling along most important scaling axes, we propose the Complete$^{(d)}$ Parameterisation that unifies scaling in width and depth -- using an adaptation of CompleteP -- as well as in batch-size and training duration. Secondly, with our parameterisation, we investigate per-module hyperparameter optimisation and transfer. We characterise the empirical challenges of navigating the high-dimensional hyperparameter landscape, and propose practical guidelines for tackling this optimisation problem. We demonstrate that, with the right parameterisation, hyperparameter transfer holds even in the per-module hyperparameter regime. Our study covers an extensive range of optimisation hyperparameters of modern models: learning rates, AdamW parameters, weight decay, initialisation scales, and residual block multipliers. Our experiments demonstrate significant training speed improvements in Large Language Models with the transferred per-module hyperparameters.
Authors: Omar Alsaqa, Linh Thi Hoang, Muhammed Fatih Balin
Abstract: Graph Neural Networks (GNNs) are powerful tools for learning from graph-structured data, but their application to large graphs is hindered by computational costs. The need to process every neighbor for each node creates memory and computational bottlenecks. To address this, we introduce BLISS, a Bandit Layer Importance Sampling Strategy. It uses multi-armed bandits to dynamically select the most informative nodes at each layer, balancing exploration and exploitation to ensure comprehensive graph coverage. Unlike existing static sampling methods, BLISS adapts to evolving node importance, leading to more informed node selection and improved performance. It demonstrates versatility by integrating with both Graph Convolutional Networks (GCNs) and Graph Attention Networks (GATs), adapting its selection policy to their specific aggregation mechanisms. Experiments show that BLISS maintains or exceeds the accuracy of full-batch training.
Authors: Jianxiang Xie, Yuncheng Hua
Abstract: While modern multivariate forecasters such as Transformers and GNNs achieve strong benchmark performance, they often suffer from systematic errors at specific variables or horizons and, critically, lack guarantees against performance degradation in deployment. Existing post-hoc residual correction methods attempt to fix these errors, but are inherently greedy: although they may improve average accuracy, they can also "help in the wrong way" by overcorrecting reliable predictions and causing local failures in unseen scenarios. To address this critical "safety gap," we propose CRC (Causality-inspired Safe Residual Correction), a plug-and-play framework explicitly designed to ensure non-degradation. CRC follows a divide-and-conquer philosophy: it employs a causality-inspired encoder to expose direction-aware structure by decoupling self- and cross-variable dynamics, and a hybrid corrector to model residual errors. Crucially, the correction process is governed by a strict four-fold safety mechanism that prevents harmful updates. Experiments across multiple datasets and forecasting backbones show that CRC consistently improves accuracy, while an in-depth ablation study confirms that its core safety mechanisms ensure exceptionally high non-degradation rates (NDR), making CRC a correction framework suited for safe and reliable deployment.
Authors: Jiacheng Li, Jianchao Tan, Zhidong Yang, Feiye Huo, Yerui Sun, Yuchen Xie, Xunliang Cai
Abstract: Low-Rank Adaptation (LoRA) is a widely adopted parameter-efficient fine-tuning (PEFT) method. However, its linear adaptation process limits its expressive power. This means there is a gap between the expressive power of linear training and non-linear training. To bridge this gap, we propose AFA-LoRA, a novel training strategy that brings non-linear expressivity to LoRA while maintaining its seamless mergeability. Our key innovation is an annealed activation function that transitions from a non-linear to a linear transformation during training, allowing the adapter to initially adopt stronger representational capabilities before converging to a mergeable linear form. We implement our method on supervised fine-tuning, reinforcement learning, and speculative decoding. The results show that AFA-LoRA reduces the performance gap between LoRA and full-parameter training. This work enables a more powerful and practical paradigm of parameter-efficient adaptation.
Authors: Qizhi Wang
Abstract: Origin-destination (OD) flow prediction remains a core task in GIS and urban analytics, yet practical deployments face two conflicting needs: high accuracy and clear interpretability. This paper develops AMBIT, a gray-box framework that augments physical mobility baselines with interpretable tree models. We begin with a comprehensive audit of classical spatial interaction models on a year-long, hourly NYC taxi OD dataset. The audit shows that most physical models are fragile at this temporal resolution; PPML gravity is the strongest physical baseline, while constrained variants improve when calibrated on full OD margins but remain notably weaker. We then build residual learners on top of physical baselines using gradient-boosted trees and SHAP analysis, demonstrating that (i) physics-grounded residuals approach the accuracy of a strong tree-based predictor while retaining interpretable structure, and (ii) POI-anchored residuals are consistently competitive and most robust under spatial generalization. We provide a reproducible pipeline, rich diagnostics, and spatial error analysis designed for urban decision-making.
Authors: Jong-Ik Park, Shreyas Chaudhari, Srinivasa Pranav, Carlee Joe-Wong, Jos\'e M. F. Moura
Abstract: In many deployed systems (multilingual ASR, cross-hospital imaging, region-specific perception), multiple pretrained specialist models coexist. Yet, new target domains often require domain expansion: a generalized model that performs well beyond any single specialist's domain. Given such a new target domain, prior works seek a single strong initialization prior for the model parameters by first blending expert models to initialize a target model. However, heuristic blending -- using coefficients based on data size or proxy metrics -- often yields lower target-domain test accuracy, and learning the coefficients on the target loss typically requires computationally-expensive full backpropagation through the network. We propose GLUE, Gradient-free Learning To Unify Experts, which initializes the target model as a convex combination of fixed experts, learning the mixture coefficients of this combination via a gradient-free two-point (SPSA) update that requires only two forward passes per step. Across experiments on three datasets and three network architectures, GLUE produces a single prior that can be fine-tuned effectively to outperform baselines. GLUE improves test accuracy by up to 8.5% over data-size weighting and by up to 9.1% over proxy-metric selection. GLUE either outperforms backpropagation-based full-gradient mixing or matches its performance within 1.4%.
Authors: Naman Aggarwal, Siddhartha R. Dalal, Vishal Misra
Abstract: Transformers often appear to perform Bayesian reasoning in context, but verifying this rigorously has been impossible: natural data lack analytic posteriors, and large models conflate reasoning with memorization. We address this by constructing \emph{Bayesian wind tunnels} -- controlled environments where the true posterior is known in closed form and memorization is provably impossible. In these settings, small transformers reproduce Bayesian posteriors with $10^{-3}$-$10^{-4}$ bit accuracy, while capacity-matched MLPs fail by orders of magnitude, establishing a clear architectural separation. Across two tasks -- bijection elimination and Hidden Markov Model (HMM) state tracking -- we find that transformers implement Bayesian inference through a consistent geometric mechanism: residual streams serve as the belief substrate, feed-forward networks perform the posterior update, and attention provides content-addressable routing. Geometric diagnostics reveal orthogonal key bases, progressive query-key alignment, and a low-dimensional value manifold parameterized by posterior entropy. During training this manifold unfurls while attention patterns remain stable, a \emph{frame-precision dissociation} predicted by recent gradient analyses. Taken together, these results demonstrate that hierarchical attention realizes Bayesian inference by geometric design, explaining both the necessity of attention and the failure of flat architectures. Bayesian wind tunnels provide a foundation for mechanistically connecting small, verifiable systems to reasoning phenomena observed in large language models.
Authors: Chuantao Li, Zhi Li, Jiahao Xu, Jie Li, Sheng Li
Abstract: Numerous studies attempt to mitigate classification bias caused by class imbalance. However, existing studies have yet to explore the collaborative optimization of imbalanced learning and model training. This constraint hinders further performance improvements. To bridge this gap, this study proposes a collaborative optimization Boosting model of multiclass imbalanced learning. This model is simple but effective by integrating the density factor and the confidence factor, this study designs a noise-resistant weight update mechanism and a dynamic sampling strategy. Rather than functioning as independent components, these modules are tightly integrated to orchestrate weight updates, sample region partitioning, and region-guided sampling. Thus, this study achieves the collaborative optimization of imbalanced learning and model training. Extensive experiments on 20 public imbalanced datasets demonstrate that the proposed model significantly outperforms eight state-of-the-art baselines. The code for the proposed model is available at: https://github.com/ChuantaoLi/DARG.
Authors: Hassan Wasswa, Timothy Lynar
Abstract: Although AI-based models have achieved high accuracy in IoT threat detection, their deployment in enterprise environments is constrained by reliance on stationary datasets that fail to reflect the dynamic nature of real-world IoT NetFlow traffic, which is frequently affected by concept drift. Existing solutions typically rely on periodic classifier retraining, resulting in high computational overhead and the risk of catastrophic forgetting. To address these challenges, this paper proposes a scalable framework for adaptive IoT threat detection that eliminates the need for continuous classifier retraining. The proposed approach trains a classifier once on latent-space representations of historical traffic, while an alignment model maps incoming traffic to the learned historical latent space prior to classification, thereby preserving knowledge of previously observed attacks. To capture inter-instance relationships among attack samples, the low-dimensional latent representations are further transformed into a graph-structured format and classified using a graph neural network. Experimental evaluations on real-world heterogeneous IoT traffic datasets demonstrate that the proposed framework maintains robust detection performance under concept drift. These results highlight the framework's potential for practical deployment in dynamic and large-scale IoT environments.
Authors: Hamed Damirchi, Cristian Rodriguez-Opazo, Ehsan Abbasnejad, Zhen Zhang, Javen Shi
Abstract: The Lottery Ticket Hypothesis (LTH) suggests that over-parameterized neural networks contain sparse subnetworks ("winning tickets") capable of matching full model performance when trained from scratch. With the growing reliance on fine-tuning large pretrained models, we investigate whether LTH extends to parameter-efficient fine-tuning (PEFT), specifically focusing on Low-Rank Adaptation (LoRA) methods. Our key finding is that LTH holds within LoRAs, revealing sparse subnetworks that can match the performance of dense adapters. In particular, we find that the effectiveness of sparse subnetworks depends more on how much sparsity is applied in each layer than on the exact weights included in the subnetwork. Building on this insight, we propose Partial-LoRA, a method that systematically identifies said subnetworks and trains sparse low-rank adapters aligned with task-relevant subspaces of the pre-trained model. Experiments across 8 vision and 12 language tasks in both single-task and multi-task settings show that Partial-LoRA reduces the number of trainable parameters by up to 87\%, while maintaining or improving accuracy. Our results not only deepen our theoretical understanding of transfer learning and the interplay between pretraining and fine-tuning but also open new avenues for developing more efficient adaptation strategies.
Authors: Lucky Susanto, Anasta Pranawijayana, Cortino Sukotjo, Soni Prasad, Derry Wijaya
Abstract: Large language models (LLMs) are increasingly adopted in high-stakes domains such as healthcare and medical education, where the risk of generating factually incorrect (i.e., hallucinated) information is a major concern. While significant efforts have been made to detect and mitigate such hallucinations, predicting whether an LLM's response is correct remains a critical yet underexplored problem. This study investigates the feasibility of predicting correctness by analyzing a general-purpose model (GPT-4o) and a reasoning-centric model (OSS-120B) on a multiple-choice prosthodontics exam. We utilize metadata and hallucination signals across three distinct prompting strategies to build a correctness predictor for each (model, prompting) pair. Our findings demonstrate that this metadata-based approach can improve accuracy by up to +7.14% and achieve a precision of 83.12% over a baseline that assumes all answers are correct. We further show that while actual hallucination is a strong indicator of incorrectness, metadata signals alone are not reliable predictors of hallucination. Finally, we reveal that prompting strategies, despite not affecting overall accuracy, significantly alter the models' internal behaviors and the predictive utility of their metadata. These results present a promising direction for developing reliability signals in LLMs but also highlight that the methods explored in this paper are not yet robust enough for critical, high-stakes deployment.
Authors: Hamed Damirchi, Ehsan Abbasnejad, Zhen Zhang, Javen Shi
Abstract: Large pre-trained models have transformed machine learning, yet adapting these models effectively to exhibit precise, concept-specific behaviors remains a significant challenge. Task vectors, defined as the difference between fine-tuned and pre-trained model parameters, provide a mechanism for steering neural networks toward desired behaviors. This has given rise to large repositories dedicated to task vectors tailored for specific behaviors. The arithmetic operation of these task vectors allows for the seamless combination of desired behaviors without the need for large datasets. However, these vectors often contain overlapping concepts that can interfere with each other during arithmetic operations, leading to unpredictable outcomes. We propose a principled decomposition method that separates each task vector into two components: one capturing shared knowledge across multiple task vectors, and another isolating information unique to each specific task. By identifying invariant subspaces across projections, our approach enables more precise control over concept manipulation without unintended amplification or diminution of other behaviors. We demonstrate the effectiveness of our decomposition method across three domains: improving multi-task merging in image classification by 5% using shared components as additional task vectors, enabling clean style mixing in diffusion models without generation degradation by mixing only the unique components, and achieving 47% toxicity reduction in language models while preserving performance on general knowledge tasks by negating the toxic information isolated to the unique component. Our approach provides a new framework for understanding and controlling task vector arithmetic, addressing fundamental limitations in model editing operations.
Authors: Jihang Wang, Dongcheng Zhao, Ruolin Chen, Qian Zhang, Yi Zeng
Abstract: Spiking Neural Networks (SNNs) utilize spike-based activations to mimic the brain's energy-efficient information processing. However, the binary and discontinuous nature of spike activations causes vanishing gradients, making adversarial robustness evaluation via gradient descent unreliable. While improved surrogate gradient methods have been proposed, their effectiveness under strong adversarial attacks remains unclear. We propose a more reliable framework for evaluating SNN adversarial robustness. We theoretically analyze the degree of gradient vanishing in surrogate gradients and introduce the Adaptive Sharpness Surrogate Gradient (ASSG), which adaptively evolves the shape of the surrogate function according to the input distribution during attack iterations, thereby enhancing gradient accuracy while mitigating gradient vanishing. In addition, we design an adversarial attack with adaptive step size under the $L_\infty$ constraint-Stable Adaptive Projected Gradient Descent (SA-PGD), achieving faster and more stable convergence under imprecise gradients. Extensive experiments show that our approach substantially increases attack success rates across diverse adversarial training schemes, SNN architectures and neuron models, providing a more generalized and reliable evaluation of SNN adversarial robustness. The experimental results further reveal that the robustness of current SNNs has been significantly overestimated and highlighting the need for more dependable adversarial training methods.
Authors: Jaebin Lee, Hankook Lee
Abstract: In machine learning, effective modeling requires a holistic consideration of how to encode inputs, make predictions (i.e., decoding), and train the model. However, in time-series forecasting, prior work has predominantly focused on encoder design, often treating prediction and training as separate or secondary concerns. In this paper, we propose TimePerceiver, a unified encoder-decoder forecasting framework that is tightly aligned with an effective training strategy. To be specific, we first generalize the forecasting task to include diverse temporal prediction objectives such as extrapolation, interpolation, and imputation. Since this generalization requires handling input and target segments that are arbitrarily positioned along the temporal axis, we design a novel encoder-decoder architecture that can flexibly perceive and adapt to these varying positions. For encoding, we introduce a set of latent bottleneck representations that can interact with all input segments to jointly capture temporal and cross-channel dependencies. For decoding, we leverage learnable queries corresponding to target timestamps to effectively retrieve relevant information. Extensive experiments demonstrate that our framework consistently and significantly outperforms prior state-of-the-art baselines across a wide range of benchmark datasets. The code is available at https://github.com/efficient-learning-lab/TimePerceiver.
URLs: https://github.com/efficient-learning-lab/TimePerceiver.
Authors: Taeyun Kim
Abstract: Rank-based input normalization is a workhorse of modern machine learning, prized for its robustness to scale, monotone transformations, and batch-to-batch variation. In many real systems, the ordering of feature values matters far more than their raw magnitudes - yet the structural conditions that a rank-based normalization operator must satisfy to remain stable under these invariances have never been formally pinned down. We show that widely used differentiable sorting and ranking operators fundamentally fail these criteria. Because they rely on value gaps and batch-level pairwise interactions, they are intrinsically unstable under strictly monotone transformations, shifts in mini-batch composition, and even tiny input perturbations. Crucially, these failures stem from the operators' structural design, not from incidental implementation choices. To address this, we propose three axioms that formalize the minimal invariance and stability properties required of rank-based input normalization. We prove that any operator satisfying these axioms must factor into (i) a feature-wise rank representation and (ii) a scalarization map that is both monotone and Lipschitz-continuous. We then construct a minimal operator that meets these criteria and empirically show that the resulting constraints are non-trivial in realistic setups. Together, our results sharply delineate the design space of valid rank-based normalization operators and formally separate them from existing continuous-relaxation-based sorting methods.
Authors: Jewel Rana Palit (Traffic Engineer/Project Manager-II, Collier County Government, Traffic Management Center, 2695 Francis Ave Unit D, Naples, Fl, 37221), Vijayalakshmi K Kumarasamy (Department of Computer Science and Engineering, University of Tennessee at Chattanooga, Chattanooga, TN, USA 37403), Osama A. Osman (Chief Scientist, Center of Urban Informatics and Progress, Chattanooga, TN, USA 37403)
Abstract: Automated Vehicles (AV) hold potential to reduce or eliminate human driving errors, enhance traffic safety, and support sustainable mobility. Recently, crash data has increasingly revealed that AV behavior can deviate from expected safety outcomes, raising concerns about the technology's safety and operational reliability in mixed traffic environments. While past research has investigated AV crash, most studies rely on small-size California-centered datasets, with a limited focus on understanding crash trends across various SAE Levels of automation. This study analyzes over 2,500 AV crash records from the United States National Highway Traffic Safety Administration (NHTSA), covering SAE Levels 2 and 4, to uncover underlying crash dynamics. A two-stage data mining framework is developed. K-means clustering is first applied to segment crash records into 4 distinct behavioral clusters based on temporal, spatial, and environmental factors. Then, Association Rule Mining (ARM) is used to extract interpretable multivariate relationships between crash patterns and crash contributors including lighting conditions, surface condition, vehicle dynamics, and environmental conditions within each cluster. These insights provide actionable guidance for AV developers, safety regulators, and policymakers in formulating AV deployment strategies and minimizing crash risks.
Authors: Guikun Xu, Xiaohan Yi, Peilin Zhao, Yatao Bian
Abstract: Generating low-energy conformer ensembles and identifying ground-state conformations from molecular graphs remain computationally demanding with physics-based pipelines. Current learning-based approaches often suffer from a fragmented paradigm: generative models capture diversity but lack reliable energy calibration, whereas deterministic predictors target a single structure and fail to represent ensemble variability. Here we present EnFlow, a unified framework that couples flow matching (FM) with an explicitly learned energy model through an energy-guided sampling scheme defined along a non-Gaussian FM path. By incorporating energy-gradient guidance during sampling, our method steers trajectories toward lower-energy regions, substantially improving conformational fidelity, particularly in the few-step regime. The learned energy function further enables efficient energy-based ranking of generated ensembles for accurate ground-state identification. Extensive experiments on GEOM-QM9 and GEOM-Drugs demonstrate that EnFlow simultaneously improves generation metrics with 1--2 ODE-steps and reduces ground-state prediction errors compared with state-of-the-art methods.
Authors: Milad Asadpour, Alireza Rezaee, Farshid Hajati
Abstract: According to the advent of cryptocurrencies and Bitcoin, many investments and businesses are now conducted online through cryptocurrencies. Among them, Bitcoin uses blockchain technology to make transactions secure, transparent, traceable, and immutable. It also exhibits significant price fluctuations and performance, which has attracted substantial attention, especially in financial sectors. Consequently, a wide range of investors and individuals have turned to investing in the cryptocurrency market. One of the most important challenges in economics is price forecasting for future trades. Cryptocurrencies are no exception, and investors are looking for methods to predict prices; various theories and methods have been proposed in this field. This paper presents a new deep model, called \emph{Parallel Gated Recurrent Units} (PGRU), for cryptocurrency price prediction. In this model, recurrent neural networks forecast prices in a parallel and independent way. The parallel networks utilize different inputs, each representing distinct price-related features. Finally, the outputs of the parallel networks are combined by a neural network to forecast the future price of cryptocurrencies. The experimental results indicate that the proposed model achieves mean absolute percentage errors (MAPE) of 3.243% and 2.641% for window lengths 20 and 15, respectively. Our method therefore attains higher accuracy and efficiency with fewer input data and lower computational cost compared to existing methods.
Authors: Hesam Taghipour, Alireza Rezaee, Farshid Hajati
Abstract: The global gold market, by its fundamentals, has long been home to many financial institutions, banks, governments, funds, and micro-investors. Due to the inherent complexity and relationship between important economic and political components, accurate forecasting of financial markets has always been challenging. Therefore, providing a model that can accurately predict the future of the markets is very important and will be of great benefit to their developers. In this paper, an artificial intelligence-based algorithm for daily and monthly gold forecasting is presented. Two Long short-term memory (LSTM) networks are responsible for daily and monthly forecasting, the results of which are integrated into a Multilayer perceptrons (MLP) network and provide the final forecast of the next day prices. The algorithm forecasts the highest, lowest, and closing prices on the daily and monthly time frame. Based on these forecasts, a trading strategy for live market trading was developed, according to which the proposed model had a return of 171% in three months. Also, the number of internal neurons in each network is optimized by the Gray Wolf optimization (GWO) algorithm based on the least RMSE error. The dataset was collected between 2010 and 2021 and includes data on macroeconomic, energy markets, stocks, and currency status of developed countries. Our proposed LSTM-MLP model predicted the daily closing price of gold with the Mean absolute error (MAE) of $ 0.21 and the next month's price with $ 22.23.
Authors: Tomas Ortega, Chun-Yin Huang, Xiaoxiao Li, Hamid Jafarkhani
Abstract: Distributed learning, particularly Federated Learning (FL), faces a significant bottleneck in the communication cost, particularly the uplink transmission of client-to-server updates, which is often constrained by asymmetric bandwidth limits at the edge. Biased compression techniques are effective in practice, but require error feedback mechanisms to provide theoretical guarantees and to ensure convergence when compression is aggressive. Standard error feedback, however, relies on client-specific control variates, which violates user privacy and is incompatible with stateless clients common in large-scale FL. This paper proposes two novel frameworks that enable biased compression without client-side state or control variates. The first, Compressed Aggregate Feedback (CAFe), uses the globally aggregated update from the previous round as a shared control variate for all clients. The second, Server-Guided Compressed Aggregate Feedback (CAFe-S), extends this idea to scenarios where the server possesses a small private dataset; it generates a server-guided candidate update to be used as a more accurate predictor. We consider Distributed Gradient Descent (DGD) as a representative algorithm and analytically prove CAFe's superiority to Distributed Compressed Gradient Descent (DCGD) with biased compression in the non-convex regime with bounded gradient dissimilarity. We further prove that CAFe-S converges to a stationary point, with a rate that improves as the server's data become more representative. Experimental results in FL scenarios validate the superiority of our approaches over existing compression schemes.
Authors: Shuyu Gan, James Mooney, Pan Hao, Renxiang Wang, Mingyi Hong, Qianwen Wang, Dongyeop Kang
Abstract: Large Language Model (LLM) agents can increasingly automate complex reasoning through Test-Time Scaling (TTS), iterative refinement guided by reward signals. However, many real-world tasks involve multi-stage pipeline whose final outcomes lack verifiable rewards or sufficient data to train robust reward models, making judge-based refinement prone to accumulate error over stages. We propose Selective TTS, a process-based refinement framework that scales inference across different stages in multi-agent pipeline, instead of repeated refinement over time by prior work. By distributing compute across stages and pruning low-quality branches early using process-specific judges, Selective TTS mitigates the judge drift and stabilizes refinement. Grounded in the data science pipeline, we build an end-to-end multi-agent pipeline for generating visually insightful charts and report of given dataset, and design a reliable LLM-based judge model, aligned with human experts (Kendall's {\tau}=0.55). Our proposed selective TTS then improves insight quality under a fixed compute budget, increasing mean scores from 61.64 to 65.86 while reducing variance. We hope our findings serve as the first step toward to scaling complex, open-ended tasks with unverifiable rewards, such as scientific discovery and story generation.
Authors: Achraf Hsain, Fouad Mohammed Abbou
Abstract: This paper presents the first application of quantum generative models to learned latent space representations of computational fluid dynamics (CFD) data. While recent work has explored quantum models for learning statistical properties of fluid systems, the combination of discrete latent space compression with quantum generative sampling for CFD remains unexplored. We develop a GPU-accelerated Lattice Boltzmann Method (LBM) simulator to generate fluid vorticity fields, which are compressed into a discrete 7-dimensional latent space using a Vector Quantized Variational Autoencoder (VQ-VAE). The central contribution is a comparative analysis of quantum and classical generative approaches for modeling this physics-derived latent distribution: we evaluate a Quantum Circuit Born Machine (QCBM) and Quantum Generative Adversarial Network (QGAN) against a classical Long Short-Term Memory (LSTM) baseline. Under our experimental conditions, both quantum models produced samples with lower average minimum distances to the true distribution compared to the LSTM, with the QCBM achieving the most favorable metrics. This work provides: (1)~a complete open-source pipeline bridging CFD simulation and quantum machine learning, (2)~the first empirical study of quantum generative modeling on compressed latent representations of physics simulations, and (3)~a foundation for future rigorous investigation at this intersection.
Authors: Donghwa Kang, Shana Moothedath
Abstract: Representation learning is a widely adopted framework for learning in data-scarce environments, aiming to extract common features from related tasks. While centralized approaches have been extensively studied, decentralized methods remain largely underexplored. We study decentralized multi-task representation learning in which the features share a low-rank structure. We consider multiple tasks, each with a finite number of data samples, where the observations follow a linear model with task-specific parameters. In the decentralized setting, task data are distributed across multiple nodes, and information exchange between nodes is constrained by a communication network. The goal is to recover the underlying feature matrix whose rank is much smaller than both the parameter dimension and the number of tasks. We propose a new alternating projected gradient and minimization algorithm with provable accuracy guarantees. We provide comprehensive characterizations of the time, communication, and sample complexities. Importantly, the communication complexity is independent of the target accuracy, which significantly reduces communication cost compared to prior methods. Numerical simulations validate the theoretical analysis across different dimensions and network topologies, and demonstrate regimes in which decentralized learning outperforms centralized federated approaches.
Authors: Andr\'e F. T. Martins
Abstract: Existing machine learning frameworks operate over the field of real numbers ($\mathbb{R}$) and learn representations in real (Euclidean or Hilbert) vector spaces (e.g., $\mathbb{R}^d$). Their underlying geometric properties align well with intuitive concepts such as linear separability, minimum enclosing balls, and subspace projection; and basic calculus provides a toolbox for learning through gradient-based optimization. But is this the only possible choice? In this paper, we study the suitability of a radically different field as an alternative to $\mathbb{R}$ -- the ultrametric and non-archimedean space of $p$-adic numbers, $\mathbb{Q}_p$. The hierarchical structure of the $p$-adics and their interpretation as infinite strings make them an appealing tool for code theory and hierarchical representation learning. Our exploratory theoretical work establishes the building blocks for classification, regression, and representation learning with the $p$-adics, providing learning models and algorithms. We illustrate how simple Quillian semantic networks can be represented as a compact $p$-adic linear network, a construction which is not possible with the field of reals. We finish by discussing open problems and opportunities for future research enabled by this new framework.
Authors: Antar Kumar Biswas, Masoud H. Nazari
Abstract: This paper presents a novel learning-based framework for predicting power outages caused by extreme events. The proposed approach specifically targets low-probability, high-consequence outage scenarios and leverages a comprehensive set of features derived from publicly available data sources. We integrate EAGLE-I outage records (2014-2024) with weather, socio-economic, infrastructure, and seasonal event data. Incorporating social and demographic indicators reveals underlying patterns of community vulnerability and provides a clearer understanding of outage risk during extreme conditions. Four machine learning models (Random Forest (RF), Support Vector Machine (SVM), Adaptive Boosting (AdaBoost), and Long Short-Term Memory (LSTM)) are evaluated. Experimental validation is performed on a large-scale dataset covering counties in the lower peninsula of Michigan. Among all models tested, the LSTM network achieves the lowest prediction error. Additionally, the results demonstrate that stronger economic conditions and more developed infrastructure are associated with lower outage occurrence.
Authors: Valentina Moretti, Andrea Cini, Ivan Marisca, Cesare Alippi
Abstract: Deep learning models have grown increasingly popular in time series applications. However, the large quantity of newly proposed architectures, together with often contradictory empirical results, makes it difficult to assess which components contribute significantly to final performance. We aim to make sense of the current design space of deep learning architectures for time series forecasting by discussing the design dimensions and trade-offs that can explain, often unexpected, observed results. This paper discusses the necessity of grounding model design on principles for forecasting groups of time series and how such principles can be applied to current models. In particular, we assess how concepts such as locality and globality apply to recent forecasting architectures. We show that accounting for these aspects can be more relevant for achieving accurate results than adopting specific sequence modeling layers and that simple, well-designed forecasting architectures can often match the state of the art. We discuss how overlooked implementation details in existing architectures (1) fundamentally change the class of the resulting forecasting method and (2) drastically affect the observed empirical results. Our results call for rethinking current faulty benchmarking practices and the need to focus on the foundational aspects of the forecasting problem when designing architectures. As a step in this direction, we propose an auxiliary forecasting model card, whose fields serve to characterize existing and new forecasting architectures based on key design choices.
Authors: Jiaqi Shao, Yufeng Miao, Wei Zhang, Bing Luo
Abstract: Long-horizon reinforcement learning (RL) for large language models faces critical scalability challenges from unbounded context growth, leading to context folding methods that compress interaction history during task execution. However, existing approaches treat summary actions as standard actions, overlooking that summaries fundamentally modify the agent's future observation space, creating a policy-dependent, non-stationary observation distribution that violates core RL assumptions. This introduces three fundamental challenges: (1) gradient dilution where summary tokens receive insufficient training signal, (2) self-conditioning where policy updates change summary distributions, creating a vicious cycle of training collapse, and (3) computational cost from processing unique contexts at each turn. We introduce \textbf{FoldAct}\footnote{https://github.com/SHAO-Jiaqi757/FoldAct}, a framework that explicitly addresses these challenges through three key innovations: separated loss computation for independent gradient signals on summary and action tokens, full context consistency loss to reduce distribution shift, and selective segment training to reduce computational cost. Our method enables stable training of long-horizon search agents with context folding, addressing the non-stationary observation problem while improving training efficiency with 5.19$\times$ speedup.
Authors: Sungwoo Kang
Abstract: Multi-task learning (MTL) assumes related material properties share underlying physics that can be leveraged for better predictions. We test this by simultaneously predicting electrical resistivity, Vickers hardness, and amorphous-forming ability using 54,028 alloy samples. We compare single-task models against standard and structured MTL. Results reveal a striking dichotomy: MTL significantly degrades regression performance (resistivity $R^2$: 0.897 $\to$ 0.844; hardness $R^2$: 0.832 $\to$ 0.694, $p < 0.01$) but improves classification (amorphous F1: 0.703 $\to$ 0.744, $p < 0.05$; recall +17%). Analysis shows near-zero inter-task weights, indicating property independence. Regression failure is attributed to negative transfer caused by severe data imbalance (52k vs. 800 samples). We recommend independent models for precise regression, while reserving MTL for classification tasks where recall is critical.
Authors: Rihong Qiu, Zhibang Yang, Xinke Jiang, Weibin Liao, Xin Gao, Xu Chu, Junfeng Zhao, Yasha Wang
Abstract: Text-to-SQL translates natural language questions into SQL statements grounded in a target database schema. Ensuring the reliability and executability of such systems requires validating generated SQL, but most existing approaches focus only on syntactic correctness, with few addressing semantic validation (detecting misalignments between questions and SQL). As a consequence, effective semantic validation still faces two key challenges: capturing both global user intent and SQL structural details, and constructing high-quality fine-grained sub-SQL annotations. To tackle these, we introduce HEROSQL, a hierarchical SQL representation approach that integrates global intent (via Logical Plans, LPs) and local details (via Abstract Syntax Trees, ASTs). To enable better information propagation, we employ a Nested Message Passing Neural Network (NMPNN) to capture inherent relational information in SQL and aggregate schema-guided semantics across LPs and ASTs. Additionally, to generate high-quality negative samples, we propose an AST-driven sub-SQL augmentation strategy, supporting robust optimization of fine-grained semantic inconsistencies. Extensive experiments conducted on Text-to-SQL validation benchmarks (both in-domain and out-of-domain settings) demonstrate that our approach outperforms existing state-of-the-art methods, achieving an average 9.40% improvement of AUPRC and 12.35% of AUROC in identifying semantic inconsistencies. It excels at detecting fine-grained semantic errors, provides large language models with more granular feedback, and ultimately enhances the reliability and interpretability of data querying platforms.
Authors: Rui Ai, David Simchi-Levi, Feng Zhu
Abstract: We study the pricing behavior of third-party platforms facing strategic agents. Assuming the platform is a revenue maximizer, it observes market features that generally affect demand. Since only the equilibrium price and quantity are observable, this presents a general demand learning problem under confounding. Mathematically, we develop an algorithm with optimal regret of $\Tilde{\cO}(\sqrt{T}\wedge\sigma_S^{-2})$. Our results reveal that supply-side noise fundamentally affects the learnability of demand, leading to a phase transition in regret. Technically, we show that non-i.i.d. actions can serve as instrumental variables for learning demand. We also propose a novel homeomorphic construction that allows us to establish estimation bounds without assuming star-shapedness, providing the first efficiency guarantee for learning demand with deep neural networks. Finally, we demonstrate the practical applicability of our approach through simulations and real-world data from Zomato and Lyft.
Authors: Eswarasanthosh Kumar Mamillapalli, Nishtha Sharma
Abstract: Childhood obesity remains a major public health challenge in the United States, strongly influenced by a combination of individual-level, household-level, and environmental-level risk factors. Traditional epidemiological studies typically analyze these levels independently, limiting insights into how structural environmental conditions interact with individual-level characteristics to influence health outcomes. In this study, we introduce a micro-macro machine learning framework that integrates (1) individual-level anthropometric and socioeconomic data from NHANES and (2) macro-level structural environment features, including food access, air quality, and socioeconomic vulnerability extracted from USDA and EPA datasets. Four machine learning models Logistic Regression, Random Forest, XGBoost, and LightGBM were trained to predict obesity using NHANES microdata. XGBoost achieved the strongest performance. A composite environmental vulnerability index (EnvScore) was constructed using normalized indicators from USDA and EPA at the state level. Multi-level comparison revealed strong geographic similarity between states with high environmental burden and the nationally predicted micro-level obesity risk distribution. This demonstrates the feasibility of integrating multi-scale datasets to identify environment-driven disparities in obesity risk. This work contributes a scalable, data-driven, multi-level modeling pipeline suitable for public health informatics, demonstrating strong potential for expansion into causal modeling, intervention planning, and real-time analytics.
Authors: Nikhil Ghosh, Denny Wu, Alberto Bietti
Abstract: The growing scale of deep learning models has rendered standard hyperparameter (HP) optimization prohibitively expensive. A promising solution is the use of scale-aware hyperparameters, which can enable direct transfer of optimal HPs from small-scale grid searches to large models with minimal performance loss. To understand the principles governing such transfer strategy, we develop a general conceptual framework for reasoning about HP transfer across scale, characterizing transfer as fast when the suboptimality it induces vanishes asymptotically faster than the finite-scale performance gap. We show formally that fast transfer is equivalent to useful transfer for compute-optimal grid search, meaning that transfer is asymptotically more compute-efficient than direct tuning. While empirical work has found that the Maximal Update Parameterization ($\mu$P) exhibits fast transfer when scaling model width, the mechanisms remain poorly understood. We show that this property depends critically on problem structure by presenting synthetic settings where transfer either offers provable computational advantage or fails to outperform direct tuning even under $\mu$P. To explain the fast transfer observed in practice, we conjecture that decomposing the optimization trajectory reveals two contributions to loss reduction: (1) a width-stable component that determines the optimal HPs, and (2) a width-sensitive component that improves with width but weakly perturbs the HP optimum. We present empirical evidence for this hypothesis across various settings, including large language model pretraining.
Authors: Xuyan Li, Jie Wang, Zheng Yan
Abstract: Dynamic graphs are widely used to represent evolving real-world networks. Temporal Graph Neural Networks (TGNNs) have emerged as a powerful tool for processing such graphs, but the lack of transparency and explainability limits their practical adoption. Research on TGNN explainability is still in its early stages and faces several key issues: (i) Current methods are tailored to specific TGNN types, restricting generality. (ii) They suffer from high computational costs, making them unsuitable for large-scale networks. (iii) They often overlook the structural connectivity of explanations and require prior knowledge, reducing user-friendliness. To address these issues, we propose GRExplainer, the first universal, efficient, and user-friendly explanation method for TGNNs. GRExplainer extracts node sequences as a unified feature representation, making it independent of specific input formats and thus applicable to both snapshot-based and event-based TGNNs (the major types of TGNNs). By utilizing breadth-first search and temporal information to construct input node sequences, GRExplainer reduces redundant computation and improves efficiency. To enhance user-friendliness, we design a generative model based on Recurrent Neural Networks (RNNs), enabling automated and continuous explanation generation. Experiments on six real-world datasets with three target TGNNs show that GRExplainer outperforms existing baseline methods in generality, efficiency, and user-friendliness.
Authors: Truong Son Nguyen
Abstract: We introduce \textbf{Schr\"{o}dinger AI}, a unified machine learning framework inspired by quantum mechanics. The system is defined by three tightly coupled components: (1) a {time-independent wave-energy solver} that treats perception and classification as spectral decomposition under a learned Hamiltonian; (2) a {time-dependent dynamical solver} governing the evolution of semantic wavefunctions over time, enabling context-aware decision revision, re-routing, and reasoning under environmental changes; and (3) a {low-rank operator calculus} that learns symbolic transformations such as modular arithmetic through learned quantum-like transition operators. Together, these components form a coherent physics-driven alternative to conventional cross-entropy training and transformer attention, providing robust generalization, interpretable semantics, and emergent topology. Empirically, Schr\"{o}dinger AI demonstrates: (a) emergent semantic manifolds that reflect human-conceived class relations without explicit supervision; (b) dynamic reasoning that adapts to changing environments, including maze navigation with real-time potential-field perturbations; and (c) exact operator generalization on modular arithmetic tasks, where the system learns group actions and composes them across sequences far beyond training length. These results suggest a new foundational direction for machine learning, where learning is cast as discovering and navigating an underlying semantic energy landscape.
Authors: Kasra Jalaldoust, Elias Bareinboim
Abstract: Generalization across the domains is not possible without asserting a structure that constrains the unseen target domain w.r.t. the source domain. Building on causal transportability theory, we design an algorithm for zero-shot compositional generalization which relies on access to qualitative domain knowledge in form of a causal graph for intra-domain structure and discrepancies oracle for inter-domain mechanism sharing. \textit{Circuit-TR} learns a collection of modules (i.e., local predictors) from the source data, and transport/compose them to obtain a circuit for prediction in the target domain if the causal structure licenses. Furthermore, circuit transportability enables us to design a supervised domain adaptation scheme that operates without access to an explicit causal structure, and instead uses limited target data. Our theoretical results characterize classes of few-shot learnable tasks in terms of graphical circuit transportability criteria, and connects few-shot generalizability with the established notion of circuit size complexity; controlled simulations corroborate our theoretical results.
Authors: Zhou Yang, Edward Dougherty, Chen Zhang, Zhenhe Pan, Fang Jin
Abstract: A comprehensive retrospective analysis of public health interventions, such as large scale testing, quarantining, and contact tracing, can help identify mechanisms most effective in mitigating COVID-19. We investigate China based SARS-CoV-2 transmission patterns (e.g., infection type and likely transmission source) using publicly released tracking data. We collect case reports from local health commissions, the Chinese CDC, and official local government social media, then apply NLP and manual curation to construct transmission/tracking chains. We further analyze tracking data together with Wuhan population mobility data to quantify and visualize temporal and spatial spread dynamics. Results indicate substantial regional differences, with larger cities showing more infections, likely driven by social activities. Most symptomatic individuals (79\%) were hospitalized within 5 days of symptom onset, and those with confirmed-case contact sought admission in under 5 days. Infection sources also shifted over time: early cases were largely linked to travel to (or contact with travelers from) Hubei Province, while later transmission was increasingly associated with social activities.
Authors: Shuai Chen, Chen Wang, Ziran Wang
Abstract: Electronic nose (E-nose) systems face dual challenges in open-set gas recognition: feature distribution shifts caused by signal drift and decision failures induced by unknown interference. Existing methods predominantly rely on Euclidean distance, failing to adequately account for anisotropic gas feature distributions and dynamic signal intensity variations. To address these issues, this study proposes SNM-Net, a universal deep learning framework for open-set gas recognition. The core innovation lies in a geometric decoupling mechanism achieved through cascaded batch normalization and L2 normalization, which projects high-dimensional features onto a unit hypersphere to eliminate signal intensity fluctuations. Additionally, Mahalanobis distance is introduced as the scoring mechanism, utilizing class-wise statistics to construct adaptive ellipsoidal decision boundaries. SNM-Net is architecture-agnostic and seamlessly integrates with CNN, RNN, and Transformer backbones. Systematic experiments on the Vergara dataset demonstrate that the Transformer+SNM configuration attains near-theoretical performance, achieving an AUROC of 0.9977 and an unknown gas detection rate of 99.57% (TPR at 5% FPR). This performance significantly outperforms state-of-the-art methods, showing a 3.0% improvement in AUROC and a 91.0% reduction in standard deviation compared to Class Anchor Clustering. The framework exhibits exceptional robustness across sensor positions with standard deviations below 0.0028. This work effectively resolves the trade-off between accuracy and stability, providing a solid technical foundation for industrial E-nose deployment.
Authors: Amirhossein Tighkhorshid, Zahra Dehghanian, Gholamali Aminian, Chengchun Shi, Hamid R. Rabiee
Abstract: Distillation addresses the slow sampling problem in diffusion models by creating models with smaller size or fewer steps that approximate the behavior of high-step teachers. In this work, we propose a reinforcement learning based distillation framework for diffusion models. Instead of relying on fixed reconstruction or consistency losses, we treat the distillation process as a policy optimization problem, where the student is trained using a reward signal derived from alignment with the teacher's outputs. This RL driven approach dynamically guides the student to explore multiple denoising paths, allowing it to take longer, optimized steps toward high-probability regions of the data distribution, rather than relying on incremental refinements. Our framework utilizes the inherent ability of diffusion models to handle larger steps and effectively manage the generative process. Experimental results show that our method achieves superior performance with significantly fewer inference steps and computational resources compared to existing distillation techniques. Additionally, the framework is model agnostic, applicable to any type of diffusion models with suitable reward functions, providing a general optimization paradigm for efficient diffusion learning.
Authors: Bor-Yiing Su, Peter Dykas, Mike Chrzanowski, Jatin Chhugani
Abstract: Mixed-precision training is a crucial technique for scaling deep learning models, but successful mixedprecision training requires identifying and applying the right combination of training methods. This paper presents our preliminary study on Mixture-of-Representations (MoR), a novel, per-tensor and sub-tensor level quantization framework that dynamically analyzes a tensor's numerical properties to select between a variety of different representations. Based on the framework, we have proposed and experimented concrete algorithms that choose dynamically between FP8 and BF16 representations for both per-tensor and sub-tensor level granularities. Our universal approach is designed to preserve model quality across various quantization partition strategies and datasets. Our initial findings show that this approach can achieve state-of-the-art results with 98.38% of tensors quantized to the FP8 format. This work highlights the potential of dynamic, property-aware quantization while preserving model quality. We believe this approach can generally improve the robustness of low precision training, as demonstrated by achieving FP8 accuracies that are on par with existing approaches without the need for fine-grain partitioning, or can be used in combination with other training methods to improve the leverage of even lower precision number formats such as NVFP4.
Authors: Scott A. Martin, Noah Brenowitz, Dale Durran, Michael Pritchard
Abstract: Accurate long-range weather forecasting remains a major challenge for AI models, both because errors accumulate over autoregressive rollouts and because reanalysis datasets used for training offer a limited sample of the slow modes of climate variability underpinning predictability. Most AI weather models are autoregressive, producing short lead forecasts that must be repeatedly applied to reach subseasonal-to-seasonal (S2S) or seasonal lead times, often resulting in instability and calibration issues. Long-timestep probabilistic models that generate long-range forecasts in a single step offer an attractive alternative, but training on the 40-year reanalysis record leads to overfitting, suggesting orders of magnitude more training data are required. We introduce long-range distillation, a method that trains a long-timestep probabilistic "student" model to forecast directly at long-range using a huge synthetic training dataset generated by a short-timestep autoregressive "teacher" model. Using the Deep Learning Earth System Model (DLESyM) as the teacher, we generate over 10,000 years of simulated climate to train distilled student models for forecasting across a range of timescales. In perfect-model experiments, the distilled models outperform climatology and approach the skill of their autoregressive teacher while replacing hundreds of autoregressive steps with a single timestep. In the real world, they achieve S2S forecast skill comparable to the ECMWF ensemble forecast after ERA5 fine-tuning. The skill of our distilled models scales with increasing synthetic training data, even when that data is orders of magnitude larger than ERA5. This represents the first demonstration that AI-generated synthetic training data can be used to scale long-range forecast skill.
Authors: Gaurav Chaudhary, Laxmidhar Behera
Abstract: Reinforcement Learning (RL) has achieved significant success in solving single-goal tasks. However, uniform goal selection often results in sample inefficiency in multi-goal settings where agents must learn a universal goal-conditioned policy. Inspired by the adaptive and structured learning processes observed in biological systems, we propose a novel Student-Teacher learning paradigm with a Temporal Variance-Driven Curriculum to accelerate Goal-Conditioned RL. In this framework, the teacher module dynamically prioritizes goals with the highest temporal variance in the policy's confidence score, parameterized by the state-action value (Q) function. The teacher provides an adaptive and focused learning signal by targeting these high-uncertainty goals, fostering continual and efficient progress. We establish a theoretical connection between the temporal variance of Q-values and the evolution of the policy, providing insights into the method's underlying principles. Our approach is algorithm-agnostic and integrates seamlessly with existing RL frameworks. We demonstrate this through evaluation across 11 diverse robotic manipulation and maze navigation tasks. The results show consistent and notable improvements over state-of-the-art curriculum learning and goal-selection methods.
Authors: Yutao Zhong
Abstract: In machine learning, the loss functions optimized during training often differ from the target loss that defines task performance due to computational intractability or lack of differentiability. We present an in-depth study of the target loss estimation error relative to the surrogate loss estimation error. Our analysis leads to $H$-consistency bounds, which are guarantees accounting for the hypothesis set $H$. These bounds offer stronger guarantees than Bayes-consistency or $H$-calibration and are more informative than excess error bounds. We begin with binary classification, establishing tight distribution-dependent and -independent bounds. We provide explicit bounds for convex surrogates (including linear models and neural networks) and analyze the adversarial setting for surrogates like $\rho$-margin and sigmoid loss. Extending to multi-class classification, we present the first $H$-consistency bounds for max, sum, and constrained losses, covering both non-adversarial and adversarial scenarios. We demonstrate that in some cases, non-trivial $H$-consistency bounds are unattainable. We also investigate comp-sum losses (e.g., cross-entropy, MAE), deriving their first $H$-consistency bounds and introducing smooth adversarial variants that yield robust learning algorithms. We develop a comprehensive framework for deriving these bounds across various surrogates, introducing new characterizations for constrained and comp-sum losses. Finally, we examine the growth rates of $H$-consistency bounds, establishing a universal square-root growth rate for smooth surrogates in binary and multi-class tasks, and analyze minimizability gaps to guide surrogate selection.
Authors: Anqi Mao
Abstract: Large language models (LLMs) have achieved remarkable performance but face critical challenges: hallucinations and high inference costs. Leveraging multiple experts offers a solution: deferring uncertain inputs to more capable experts improves reliability, while routing simpler queries to smaller, distilled models enhances efficiency. This motivates the problem of learning with multiple-expert deferral. This thesis presents a comprehensive study of this problem and the related problem of learning with abstention, supported by strong consistency guarantees. First, for learning with abstention (a special case of deferral), we analyze score-based and predictor-rejector formulations in multi-class classification. We introduce new families of surrogate losses and prove strong non-asymptotic, hypothesis set-specific consistency guarantees, resolving two existing open questions. We analyze both single-stage and practical two-stage settings, with experiments on CIFAR-10, CIFAR-100, and SVHN demonstrating the superior performance of our algorithms. Second, we address general multi-expert deferral in classification. We design new surrogate losses for both single-stage and two-stage scenarios and prove they benefit from strong $H$-consistency bounds. For the two-stage scenario, we show that our surrogate losses are realizable $H$-consistent for constant cost functions, leading to effective new algorithms. Finally, we introduce a novel framework for regression with deferral to address continuous label spaces. Our versatile framework accommodates multiple experts and various cost structures, supporting both single-stage and two-stage methods. It subsumes recent work on regression with abstention. We propose new surrogate losses with proven $H$-consistency and demonstrate the empirical effectiveness of the resulting algorithms.
Authors: S. Dai, G. Sun, F. Li, X. Tang, Q. Wang, Y. Cong
Abstract: Spectral clustering has emerged as one of the most effective clustering algorithms due to its superior performance. However, most existing models are designed for centralized settings, rendering them inapplicable in modern decentralized environments. Moreover, current federated learning approaches often suffer from poor generalization performance due to reliance on unreliable pseudo-labels, and fail to capture the latent correlations amongst heterogeneous clients. To tackle these limitations, this paper proposes a novel framework named Federated Multi-Task Clustering (i.e.,FMTC), which intends to learn personalized clustering models for heterogeneous clients while collaboratively leveraging their shared underlying structure in a privacy-preserving manner. More specifically, the FMTC framework is composed of two main components: client-side personalized clustering module, which learns a parameterized mapping model to support robust out-of-sample inference, bypassing the need for unreliable pseudo-labels; and server-side tensorial correlation module, which explicitly captures the shared knowledge across all clients. This is achieved by organizing all client models into a unified tensor and applying a low-rank regularization to discover their common subspace. To solve this joint optimization problem, we derive an efficient, privacy-preserving distributed algorithm based on the Alternating Direction Method of Multipliers, which decomposes the global problem into parallel local updates on clients and an aggregation step on the server. To the end, several extensive experiments on multiple real-world datasets demonstrate that our proposed FMTC framework significantly outperforms various baseline and state-of-the-art federated clustering algorithms.
Authors: Chumeng Liang, Zhanyang Jin, Zahaib Akhtar, Mona Pereira, Haofei Yu, Jiaxuan You
Abstract: Tabular log abstracts objects and events in the real-world system and reports their updates to reflect the change of the system, where one can detect real-world inconsistencies efficiently by debugging corresponding log entries. However, recent advances in processing text-enriched tabular log data overly depend on large language models (LLMs) and other heavy-load models, thus suffering from limited flexibility and scalability. This paper proposes a new framework, GraphLogDebugger, to debug tabular log based on dynamic graphs. By constructing heterogeneous nodes for objects and events and connecting node-wise edges, the framework recovers the system behind the tabular log as an evolving dynamic graph. With the help of our dynamic graph modeling, a simple dynamic Graph Neural Network (GNN) is representative enough to outperform LLMs in debugging tabular log, which is validated by experimental results on real-world log datasets of computer systems and academic papers.
Authors: Jin Wu, Chanjin Zheng
Abstract: Cognitive diagnosis is an essential research topic in intelligent education, aimed at assessing the level of mastery of different skills by students. So far, many research works have used deep learning models to explore the complex interactions between students, questions, and skills. However, the performance of existing method is frequently limited by the long-tailed distribution and dynamic changes in the data. To address these challenges, we propose a meta-learning framework for cognitive diagnosis based on continual learning (MetaCD). This framework can alleviate the long-tailed problem by utilizing meta-learning to learn the optimal initialization state, enabling the model to achieve good accuracy on new tasks with only a small amount of data. In addition, we utilize a continual learning method named parameter protection mechanism to give MetaCD the ability to adapt to new skills or new tasks, in order to adapt to dynamic changes in data. MetaCD can not only improve the plasticity of our model on a single task, but also ensure the stability and generalization of the model on sequential tasks. Comprehensive experiments on five real-world datasets show that MetaCD outperforms other baselines in both accuracy and generalization.
Authors: \"Unver \c{C}ift\c{c}i
Abstract: Deep Q-learning algorithms remain notoriously unstable, especially during early training when the maximization operator amplifies estimation errors. Inspired by bounded rationality theory and developmental learning, we introduce Sat-EnQ, a two-phase framework that first learns to be ``good enough'' before optimizing aggressively. In Phase 1, we train an ensemble of lightweight Q-networks under a satisficing objective that limits early value growth using a dynamic baseline, producing diverse, low-variance estimates while avoiding catastrophic overestimation. In Phase 2, the ensemble is distilled into a larger network and fine-tuned with standard Double DQN. We prove theoretically that satisficing induces bounded updates and cannot increase target variance, with a corollary quantifying conditions for substantial reduction. Empirically, Sat-EnQ achieves 3.8x variance reduction, eliminates catastrophic failures (0% vs 50% for DQN), maintains 79% performance under environmental noise}, and requires 2.5x less compute than bootstrapped ensembles. Our results highlight a principled path toward robust reinforcement learning by embracing satisficing before optimization.
Authors: Vincent Herrmann, Eric Alcaide, Michael Wand, J\"urgen Schmidhuber
Abstract: Measuring the in-context computational effort of language models is a key challenge, as metrics like next-token loss fail to capture reasoning complexity. Prior methods based on latent state compressibility can be invasive and unstable. We propose Multiple Token Divergence (MTD), a simple measure of computational effort defined as the KL divergence between a model's full output distribution and that of a shallow, auxiliary prediction head. MTD can be computed directly from pre-trained models with multiple prediction heads, requiring no additional training. Building on this, we introduce Divergence Steering, a novel decoding method to control the computational character of generated text. We empirically show that MTD is more effective than prior methods at distinguishing complex tasks from simple ones. On mathematical reasoning benchmarks, MTD correlates positively with problem difficulty. Lower MTD is associated with more accurate reasoning. MTD provides a practical, lightweight tool for analyzing and steering the computational dynamics of language models.
Authors: Wang Zixian
Abstract: Two divergence regimes dominate modern alignment practice. Supervised fine-tuning and many distillation-style objectives implicitly minimize the forward KL divergence KL(q || pi_theta), yielding stable mode-covering updates but often under-exploiting high-reward modes. In contrast, PPO-style online reinforcement learning from human feedback behaves closer to reverse KL divergence KL(pi_theta || q), enabling mode-seeking improvements but risking mode collapse. Recent anchored methods, such as ADPO, show that performing the projection in anchored coordinates can substantially improve stability, yet they typically commit to a single divergence. We introduce Alpha-Divergence Preference Optimization (APO), an anchored framework that uses Csiszar alpha-divergence to continuously interpolate between forward and reverse KL behavior within the same anchored geometry. We derive unified gradient dynamics parameterized by alpha, analyze gradient variance properties, and propose a practical reward-and-confidence-guarded alpha schedule that transitions from coverage to exploitation only when the policy is both improving and confidently calibrated. Experiments on Qwen3-1.7B with math-level3 demonstrate that APO achieves competitive performance with GRPO and GSPO baselines while maintaining training stability.
Authors: Wafaa El Husseini
Abstract: Access to longitudinal, individual-level data on work-life balance and wellbeing is limited by privacy, ethical, and logistical constraints. This poses challenges for reproducible research, methodological benchmarking, and education in domains such as stress modeling, behavioral analysis, and machine learning. We introduce FLOW, a synthetic longitudinal dataset designed to model daily interactions between workload, lifestyle behaviors, and wellbeing. FLOW is generated using a rule-based, feedback-driven simulation that produces coherent temporal dynamics across variables such as stress, sleep, mood, physical activity, and body weight. The dataset simulates 1{,}000 individuals over a two-year period with daily resolution and is released as a publicly available resource. In addition to the static dataset, we describe a configurable data generation tool that enables reproducible experimentation under adjustable behavioral and contextual assumptions. FLOW is intended as a controlled experimental environment rather than a proxy for observed human populations, supporting exploratory analysis, methodological development, and benchmarking where real-world data are inaccessible.
Authors: Amirali Vakili, Salar Jahanshiri, Armin Salimi-Badr
Abstract: Automatic sleep staging is a critical task in healthcare due to the global prevalence of sleep disorders. This study focuses on single-channel electroencephalography (EEG), a practical and widely available signal for automatic sleep staging. Existing approaches face challenges such as class imbalance, limited receptive-field modeling, and insufficient interpretability. This work proposes a context-aware and interpretable framework for single-channel EEG sleep staging, with particular emphasis on improving detection of the N1 stage. Many prior models operate as black boxes with stacked layers, lacking clearly defined and interpretable feature extraction roles.The proposed model combines compact multi-scale feature extraction with temporal modeling to capture both local and long-range dependencies. To address data imbalance, especially in the N1 stage, classweighted loss functions and data augmentation are applied. EEG signals are segmented into sub-epoch chunks, and final predictions are obtained by averaging softmax probabilities across chunks, enhancing contextual representation and robustness.The proposed framework achieves an overall accuracy of 89.72% and a macro-average F1-score of 85.46%. Notably, it attains an F1- score of 61.7% for the challenging N1 stage, demonstrating a substantial improvement over previous methods on the SleepEDF datasets. These results indicate that the proposed approach effectively improves sleep staging performance while maintaining interpretability and suitability for real-world clinical applications.
Authors: Tillmann Rheude, Roland Eils, Benjamin Wild
Abstract: Deep learning architectures for multimodal learning have increased in complexity, driven by the assumption that multimodal-specific methods improve performance. We challenge this assumption through a large-scale empirical study reimplementing 19 high-impact methods under standardized conditions, evaluating them across nine diverse datasets with up to 23 modalities, and testing their generalizability to new tasks beyond their original scope, including settings with missing modalities. We propose a Simple Baseline for Multimodal Learning (SimBaMM), a straightforward late-fusion Transformer architecture, and demonstrate that under standardized experimental conditions with rigorous hyperparameter tuning of all methods, more complex architectures do not reliably outperform SimBaMM. Statistical analysis indicates that more complex methods perform comparably to SimBaMM and frequently do not reliably outperform well-tuned unimodal baselines, especially in the small-data regime considered in many original studies. To support our findings, we include a case study of a recent multimodal learning method highlighting the methodological shortcomings in the literature. In addition, we provide a pragmatic reliability checklist to promote comparable, robust, and trustworthy future evaluations. In summary, we argue for a shift in focus: away from the pursuit of architectural novelty and toward methodological rigor.
Authors: Fuli Qiao, Mehrdad Mahdavi
Abstract: Parameter-efficient continual learning has emerged as a promising approach for large language models (LLMs) to mitigate catastrophic forgetting while enabling adaptation to new tasks. Current Low-Rank Adaptation (LoRA) continual learning techniques often retain and freeze previously learned LoRAs or generate data representations to overcome forgetting, typically utilizing these to support new LoRAs learn new tasks. However, these methods not only ignore growing computational memory with tasks and limited storage space but also suffer from potential task interference due to the lack of effective LoRA merging mechanisms. In this paper, we propose a novel continual learning method that orthogonally initializes and sequentially merges LoRAs updates into a single unified LoRA. Our method leverages orthogonal basis extraction from previously learned LoRA to initialize the learning of new tasks, further exploits the intrinsic asymmetry property of LoRA components by using a time-aware scaling mechanism to balance new and old knowledge during continual merging. Our approach maintains constant memory complexity with respect to the number of tasks, minimizes interference between past and new tasks via orthogonal basis initialization, and improves performance over asymmetric LoRA merging via adaptive scaling. We provide theoretical analysis to justify our design and conduct extensive experiments across diverse continual learning benchmarks using various Llama models, demonstrating the effectiveness and efficiency of our method.
Authors: Muhammad Haseeb, Salaar Masood, Muhammad Abdullah Sohail
Abstract: Federated Learning (FL) enables collaborative training of models on decentralized data, but its performance degrades significantly under Non-IID (non-independent and identically distributed) data conditions. While this accuracy loss is well-documented, the internal mechanistic causes remain a black box. This paper investigates the canonical FedAvg algorithm through the lens of Mechanistic Interpretability (MI) to diagnose this failure mode. We hypothesize that the aggregation of conflicting client updates leads to circuit collapse, the destructive interference of functional, sparse sub-networks responsible for specific class predictions. By training inherently interpretable, weight-sparse neural networks within an FL framework, we identify and track these circuits across clients and communication rounds. Using Intersection-over-Union (IoU) to quantify circuit preservation, we provide the first mechanistic evidence that Non-IID data distributions cause structurally distinct local circuits to diverge, leading to their degradation in the global model. Our findings reframe the problem of statistical drift in FL as a concrete, observable failure of mechanistic preservation, paving the way for more targeted solutions.
Authors: Min Zhu, Jingmin Sun, Zecheng Zhang, Hayden Schaeffer, Lu Lu
Abstract: Partial differential equations (PDEs) govern a wide range of physical systems, and recent multimodal foundation models have shown promise for learning PDE solution operators across diverse equation families. However, existing multi-operator learning approaches are data-hungry and neglect physics during training. Here, we propose a physics-informed multimodal foundation model (PI-MFM) framework that directly enforces governing equations during pretraining and adaptation. PI-MFM takes symbolic representations of PDEs as the input, and automatically assembles PDE residual losses from the input expression via a vectorized derivative computation. These designs enable any PDE-encoding multimodal foundation model to be trained or adapted with unified physics-informed objectives across equation families. On a benchmark of 13 parametric one-dimensional time-dependent PDE families, PI-MFM consistently outperforms purely data-driven counterparts, especially with sparse labeled spatiotemporal points, partially observed time domains, or few labeled function pairs. Physics losses further improve robustness against noise, and simple strategies such as resampling collocation points substantially improve accuracy. We also analyze the accuracy, precision, and computational cost of automatic differentiation and finite differences for derivative computation within PI-MFM. Finally, we demonstrate zero-shot physics-informed fine-tuning to unseen PDE families: starting from a physics-informed pretrained model, adapting using only PDE residuals and initial/boundary conditions, without any labeled solution data, rapidly reduces test errors to around 1% and clearly outperforms physics-only training from scratch. These results show that PI-MFM provides a practical and scalable path toward data-efficient, transferable PDE solvers.
Authors: Shuhuan Wang, Yuzhen Xie, Jiayi Li, Yinliang Diao
Abstract: Selective State Space Models (SSMs) achieve linear-time inference, yet their gradient-based sensitivity analysis remains bottlenecked by O(L) memory scaling during backpropagation. This memory constraint precludes genomic-scale modeling (L > 10^5) on consumer-grade hardware. We introduce Phase Gradient Flow (PGF), a framework that computes exact analytical derivatives by operating directly in the state-space manifold, bypassing the need to materialize the intermediate computational graph. By reframing SSM dynamics as Tiled Operator-Space Evolution (TOSE), our method delivers O(1) memory complexity relative to sequence length, yielding a 94% reduction in peak VRAM and a 23x increase in throughput compared to standard Autograd. Unlike parallel prefix scans that exhibit numerical divergence in stiff ODE regimes, PGF ensures stability through invariant error scaling, maintaining near-machine precision across extreme sequences. We demonstrate the utility of PGF on an impulse-response benchmark with 128,000-step sequences - a scale where conventional Autograd encounters prohibitive memory overhead, often leading to out-of-memory (OOM) failures in multi-layered models. Our work enables chromosome-scale sensitivity analysis on a single GPU, bridging the gap between theoretical infinite-context models and practical hardware limitations.
Authors: Boyang Zhang, Xiaobing Chen, Songyang Zhang, Shuai Zhang, Xiangwei Zhou, Mingxuan Sun
Abstract: Mixture-of-Experts (MoE) models enable scalable neural networks through conditional computation. However, their deployment with federated learning (FL) faces two critical challenges: 1) resource-constrained edge devices cannot store full expert sets, and 2) non-IID data distributions cause severe expert load imbalance that degrades model performance. To this end, we propose \textbf{FLEX-MoE}, a novel federated MoE framework that jointly optimizes expert assignment and load balancing under limited client capacity. Specifically, our approach introduces client-expert fitness scores that quantify the expert suitability for local datasets through training feedback, and employs an optimization-based algorithm to maximize client-expert specialization while enforcing balanced expert utilization system-wide. Unlike existing greedy methods that focus solely on personalization while ignoring load imbalance, our FLEX-MoE is capable of addressing the expert utilization skew, which is particularly severe in FL settings with heterogeneous data. Our comprehensive experiments on three different datasets demonstrate the superior performance of the proposed FLEX-MoE, together with its ability to maintain balanced expert utilization across diverse resource-constrained scenarios.
Authors: Mingyuan Zhang, Yue Bai, Yifan Wang, Yiyang Huang, Yun Fu
Abstract: Explorations in fine-tuning Vision-Language Models (VLMs), such as Low-Rank Adaptation (LoRA) from Parameter Efficient Fine-Tuning (PEFT), have made impressive progress. However, most approaches rely on explicit weight updates, overlooking the extensive representational structures already encoded in pre-trained models that remain underutilized. Recent works have demonstrated that Mask Fine-Tuning (MFT) can be a powerful and efficient post-training paradigm for language models. Instead of updating weights, MFT assigns learnable gating scores to each weight, allowing the model to reorganize its internal subnetworks for downstream task adaptation. In this paper, we rethink fine-tuning for VLMs from a structural reparameterization perspective grounded in MFT. We apply MFT to the language and projector components of VLMs with different language backbones and compare against strong PEFT baselines. Experiments show that MFT consistently surpasses LoRA variants and even full fine-tuning, achieving high performance without altering the frozen backbone. Our findings reveal that effective adaptation can emerge not only from updating weights but also from reestablishing connections among the model's existing knowledge. Code available at: https://github.com/Ming-K9/MFT-VLM
Authors: Yingru Li, Jiacai Liu, Jiawei Xu, Yuxuan Tong, Ziniu Li, Baoxiang Wang
Abstract: Policy gradient methods for large language models optimize a surrogate objective computed from samples of a rollout policy $\pi_{\text{roll}}$. When $\pi_{\text{roll}} \ne \pi_{\theta}$, there is approximation error between the surrogate and the true objective. Prior work has shown that this off-policy mismatch is unavoidable in modern LLM-RL due to implementation divergence, mixture-of-experts routing discontinuities, and distributed training staleness. Classical trust region bounds on the resulting error scale as $O(T^2)$ with sequence length $T$, rendering them vacuous for long-horizon tasks. We derive two tighter bounds: a Pinsker-Marginal bound scaling as $O(T^{3/2})$ and a Mixed bound scaling as $O(T)$. Crucially, both bounds depend on $D_{kl}^{tok,max}$ -- the maximum token-level KL divergence across all positions in a sequence. This is inherently a sequence-level quantity: it requires examining the entire trajectory to compute, and therefore cannot be controlled by token-independent methods like PPO clipping. We propose Trust Region Masking (TRM), which excludes entire sequences from gradient computation if any token violates the trust region, providing the first non-vacuous monotonic improvement guarantees for long-horizon LLM-RL.
Authors: Deyang Zheng, Tianyi Zhang, Wenming Zheng, Shujian Yu
Abstract: Emotional states manifest as coordinated yet heterogeneous physiological responses across central and autonomic systems, posing a fundamental challenge for multimodal representation learning in affective computing. Learning such joint dynamics is further complicated by the scarcity and subjectivity of affective annotations, which motivates the use of self-supervised learning (SSL). However, most existing SSL approaches rely on pairwise alignment objectives, which are insufficient to characterize dependencies among more than two modalities and fail to capture higher-order interactions arising from coordinated brain and autonomic responses. To address this limitation, we propose Multimodal Functional Maximum Correlation (MFMC), a principled SSL framework that maximizes higher-order multimodal dependence through a Dual Total Correlation (DTC) objective. By deriving a tight sandwich bound and optimizing it using a functional maximum correlation analysis (FMCA) based trace surrogate, MFMC captures joint multimodal interactions directly, without relying on pairwise contrastive losses. Experiments on three public affective computing benchmarks demonstrate that MFMC consistently achieves state-of-the-art or competitive performance under both subject-dependent and subject-independent evaluation protocols, highlighting its robustness to inter-subject variability. In particular, MFMC improves subject-dependent accuracy on CEAP-360VR from 78.9% to 86.8%, and subject-independent accuracy from 27.5% to 33.1% using the EDA signal alone. Moreover, MFMC remains within 0.8 percentage points of the best-performing method on the most challenging EEG subject-independent split of MAHNOB-HCI. Our code is available at https://github.com/DY9910/MFMC.
Authors: Yingru Li, Jiawei Xu, Jiacai Liu, Yuxuan Tong, Ziniu Li, Tianle Cai, Ge Zhang, Qian Liu, Baoxiang Wang
Abstract: Reinforcement learning for large language models (LLMs) faces a fundamental tension: high-throughput inference engines and numerically-precise training systems produce different probability distributions from the same parameters, creating a training-inference mismatch. We prove this mismatch has an asymmetric effect: the bound on log-probability mismatch scales as $(1-p)$ where $p$ is the token probability. For high-probability tokens, this bound vanishes, contributing negligibly to sequence-level mismatch. For low-probability tokens in the tail, the bound remains large, and moreover, when sampled, these tokens exhibit systematically biased mismatches that accumulate over sequences, destabilizing gradient estimation. Rather than applying post-hoc corrections, we propose constraining the RL objective to a dynamically-pruned ``safe'' vocabulary that excludes the extreme tail. By pruning such tokens, we trade large, systematically biased mismatches for a small, bounded optimization bias. Empirically, our method achieves stable training; theoretically, we bound the optimization bias introduced by vocabulary pruning.
Authors: Mario Colosi, Reza Farahani, Maria Fazio, Radu Prodan, Massimo Villari
Abstract: Data within a specific context gains deeper significance beyond its isolated interpretation. In distributed systems, interdependent data sources reveal hidden relationships and latent structures, representing valuable information for many applications. This paper introduces Osmotic Learning (OSM-L), a self-supervised distributed learning paradigm designed to uncover higher-level latent knowledge from distributed data. The core of OSM-L is osmosis, a process that synthesizes dense and compact representation by extracting contextual information, eliminating the need for raw data exchange between distributed entities. OSM-L iteratively aligns local data representations, enabling information diffusion and convergence into a dynamic equilibrium that captures contextual patterns. During training, it also identifies correlated data groups, functioning as a decentralized clustering mechanism. Experimental results confirm OSM-L's convergence and representation capabilities on structured datasets, achieving over 0.99 accuracy in local information alignment while preserving contextual integrity.
Authors: Yingru Li, Ziniu Li, Jiacai Liu
Abstract: We present a unified framework for Large Language Model (LLM) fine-tuning that integrates Imitation Learning and Reinforcement Learning. By analyzing the gradient of a composite objective combining trajectory-level KL divergence with task rewards, we derive a natural decomposition into two components: (1) an analytically computable Dense Gradient for token-level imitation, and (2) a Monte Carlo estimated Sparse Gradient for long-horizon reward optimization. The Dense Gradient admits a closed-form logit-level formula, enabling efficient GPU implementation.
Authors: Paul M. Thompson
Abstract: Modern generative and vision-language models (VLMs) are increasingly used in scientific and medical decision support, where predicted probabilities must be both accurate and well calibrated. Despite strong empirical results with moderate data, it remains unclear when such predictions generalize uniformly across inputs, classes, or subpopulations, rather than only on average-a critical issue in biomedicine, where rare conditions and specific groups can exhibit large errors even when overall loss is low. We study this question from a finite-sample perspective and ask: under what structural assumptions can generative and VLM-based predictors achieve uniformly accurate and calibrated behavior with practical sample sizes? Rather than analyzing arbitrary parameterizations, we focus on induced families of classifiers obtained by varying prompts or semantic embeddings within a restricted representation space. When model outputs depend smoothly on a low-dimensional semantic representation-an assumption supported by spectral structure in text and joint image-text embeddings-classical uniform convergence tools yield meaningful non-asymptotic guarantees. Our main results give finite-sample uniform convergence bounds for accuracy and calibration functionals of VLM-induced classifiers under Lipschitz stability with respect to prompt embeddings. The implied sample complexity depends on intrinsic/effective dimension, not ambient embedding dimension, and we further derive spectrum-dependent bounds that make explicit how eigenvalue decay governs data requirements. We conclude with implications for data-limited biomedical modeling, including when current dataset sizes can support uniformly reliable predictions and why average calibration metrics may miss worst-case miscalibration.
Authors: Yankang Li, Changsheng Li
Abstract: Accurate identification of the penetration process relies heavily on prior feature values of penetration acceleration. However, these feature values are typically obtained through long simulation cycles and expensive computations. To overcome this limitation, this paper proposes a multi-layer Perceptron architecture, termed squeeze and excitation multi-layer perceptron (SE-MLP), which integrates a channel attention mechanism with residual connections to enable rapid prediction of acceleration feature values. Using physical parameters under different working conditions as inputs, the model outputs layer-wise acceleration features, thereby establishing a nonlinear mapping between physical parameters and penetration characteristics. Comparative experiments against conventional MLP, XGBoost, and Transformer models demonstrate that SE-MLP achieves superior prediction accuracy, generalization, and stability. Ablation studies further confirm that both the channel attention module and residual structure contribute significantly to performance gains. Numerical simulations and range recovery tests show that the discrepancies between predicted and measured acceleration peaks and pulse widths remain within acceptable engineering tolerances. These results validate the feasibility and engineering applicability of the proposed method and provide a practical basis for rapidly generating prior feature values for penetration fuzes.
Authors: Anqi Mao, Mehryar Mohri, Yutao Zhong
Abstract: In applications with significant class imbalance or asymmetric costs, metrics such as the $F_\beta$-measure, AM measure, Jaccard similarity coefficient, and weighted accuracy offer more suitable evaluation criteria than standard binary classification loss. However, optimizing these metrics present significant computational and statistical challenges. Existing approaches often rely on the characterization of the Bayes-optimal classifier, and use threshold-based methods that first estimate class probabilities and then seek an optimal threshold. This leads to algorithms that are not tailored to restricted hypothesis sets and lack finite-sample performance guarantees. In this work, we introduce principled algorithms for optimizing generalized metrics, supported by $H$-consistency and finite-sample generalization bounds. Our approach reformulates metric optimization as a generalized cost-sensitive learning problem, enabling the design of novel surrogate loss functions with provable $H$-consistency guarantees. Leveraging this framework, we develop new algorithms, METRO (Metric Optimization), with strong theoretical performance guarantees. We report the results of experiments demonstrating the effectiveness of our methods compared to prior baselines.
Authors: Runzhi Zhou, Xi Luo
Abstract: Integrating non-Euclidean brain imaging data with Euclidean tabular data, such as clinical and demographic information, poses a substantial challenge for medical imaging analysis, particularly in forecasting future outcomes. While machine learning and deep learning techniques have been applied successfully to cross-sectional classification and prediction tasks, effectively forecasting outcomes in longitudinal imaging studies remains challenging. To address this challenge, we introduce a time-aware graph neural network model with transformer fusion (GNN-TF). This model flexibly integrates both tabular data and dynamic brain connectivity data, leveraging the temporal order of these variables within a coherent framework. By incorporating non-Euclidean and Euclidean sources of information from a longitudinal resting-state fMRI dataset from the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA), the GNN-TF enables a comprehensive analysis that captures critical aspects of longitudinal imaging data. Comparative analyses against a variety of established machine learning and deep learning models demonstrate that GNN-TF outperforms these state-of-the-art methods, delivering superior predictive accuracy for predicting future tobacco usage. The end-to-end, time-aware transformer fusion structure of the proposed GNN-TF model successfully integrates multiple data modalities and leverages temporal dynamics, making it a valuable analytic tool for functional brain imaging studies focused on clinical outcome prediction.
Authors: Xianqi Liu, Xiangru Li, Lefeng He, Ziyu Fang
Abstract: Weak signal learning (WSL) is a common challenge in many fields like fault diagnosis, medical imaging, and autonomous driving, where critical information is often masked by noise and interference, making feature identification difficult. Even in tasks with abundant strong signals, the key to improving model performance often lies in effectively extracting weak signals. However, the lack of dedicated datasets has long constrained research. To address this, we construct the first specialized dataset for weak signal feature learning, containing 13,158 spectral samples. It features low SNR dominance (over 55% samples with SNR below 50) and extreme class imbalance (class ratio up to 29:1), providing a challenging benchmark for classification and regression in weak signal scenarios. We also propose a dual-view representation (vector + time-frequency map) and a PDVFN model tailored to low SNR, distribution skew, and dual imbalance. PDVFN extracts local sequential features and global frequency-domain structures in parallel, following principles of local enhancement, sequential modeling, noise suppression, multi-scale capture, frequency extraction, and global perception. This multi-source complementarity enhances representation for low-SNR and imbalanced data, offering a novel solution for WSL tasks like astronomical spectroscopy. Experiments show our method achieves higher accuracy and robustness in handling weak signals, high noise, and extreme class imbalance, especially in low SNR and imbalanced scenarios. This study provides a dedicated dataset, a baseline model, and establishes a foundation for future WSL research.
Authors: Donghwa Kang, Shana Moothedath
Abstract: Representation learning is a widely adopted framework for learning in data-scarce environments to obtain a feature extractor or representation from various different yet related tasks. Despite extensive research on representation learning, decentralized approaches remain relatively underexplored. This work develops a decentralized projected gradient descent-based algorithm for multi-task representation learning. We focus on the problem of multi-task linear regression in which multiple linear regression models share a common, low-dimensional linear representation. We present an alternating projected gradient descent and minimization algorithm for recovering a low-rank feature matrix in a diffusion-based decentralized and federated fashion. We obtain constructive, provable guarantees that provide a lower bound on the required sample complexity and an upper bound on the iteration complexity of our proposed algorithm. We analyze the time and communication complexity of our algorithm and show that it is fast and communication-efficient. We performed numerical simulations to validate the performance of our algorithm and compared it with benchmark algorithms.
Authors: Qingyu Yin, Yulun Wu, Zhennan Shen, Sunbowen Li, Zhilin Wang, Yanshu Li, Chak Tou Leong, Jiale Kang, Jinjin Gu
Abstract: We systematically evaluate Parameter-Efficient Fine-Tuning (PEFT) methods under the paradigm of Reinforcement Learning with Verifiable Rewards (RLVR). RLVR incentivizes language models to enhance their reasoning capabilities through verifiable feedback; however, while methods like LoRA are commonly used, the optimal PEFT architecture for RLVR remains unidentified. In this work, we conduct the first comprehensive evaluation of over 12 PEFT methodologies across the DeepSeek-R1-Distill families on mathematical reasoning benchmarks. Our empirical results challenge the default adoption of standard LoRA with three main findings. First, we demonstrate that structural variants, such as DoRA, AdaLoRA, and MiSS, consistently outperform LoRA. Second, we uncover a spectral collapse phenomenon in SVD-informed initialization strategies (\textit{e.g.,} PiSSA, MiLoRA), attributing their failure to a fundamental misalignment between principal-component updates and RL optimization. Furthermore, our ablations reveal that extreme parameter reduction (\textit{e.g.,} VeRA, Rank-1) severely bottlenecks reasoning capacity. We further conduct ablation studies and scaling experiments to validate our findings. This work provides a definitive guide for advocating for more exploration for parameter-efficient RL methods.
Authors: Seungeon Lee, Takuto Koyama, Itsuki Maeda, Shigeyuki Matsumoto, Yasushi Okuno
Abstract: Therapeutic peptides have emerged as a pivotal modality in modern drug discovery, occupying a chemically and topologically rich space. While accurate prediction of their physicochemical properties is essential for accelerating peptide development, existing molecular language models rely on representations that fail to capture this complexity. Atom-level SMILES notation generates long token sequences and obscures cyclic topology, whereas amino-acid-level representations cannot encode the diverse chemical modifications central to modern peptide design. To bridge this representational gap, the Hierarchical Editing Language for Macromolecules (HELM) offers a unified framework enabling precise description of both monomer composition and connectivity, making it a promising foundation for peptide language modeling. Here, we propose HELM-BERT, the first encoder-based peptide language model trained on HELM notation. Based on DeBERTa, HELM-BERT is specifically designed to capture hierarchical dependencies within HELM sequences. The model is pre-trained on a curated corpus of 39,079 chemically diverse peptides spanning linear and cyclic structures. HELM-BERT significantly outperforms state-of-the-art SMILES-based language models in downstream tasks, including cyclic peptide membrane permeability prediction and peptide-protein interaction prediction. These results demonstrate that HELM's explicit monomer- and topology-aware representations offer substantial data-efficiency advantages for modeling therapeutic peptides, bridging a long-standing gap between small-molecule and protein language models.
Authors: Will Sebelik-Lassiter, Evan Schubert, Muhammad Alliyu, Quentin Robbins, Excel Olatunji, Mustafa Barry
Abstract: Intro: Vocal cord ultrasound (VCUS) has emerged as a less invasive and better tolerated examination technique, but its accuracy is operator dependent. This research aims to apply a machine learning-assisted algorithm to automatically identify the vocal cords and distinguish normal vocal cord images from vocal cord paralysis (VCP). Methods: VCUS videos were acquired from 30 volunteers, which were split into still frames and cropped to a uniform size. Healthy and simulated VCP images were used as training data for vocal cord segmentation and VCP classification models. Results: The vocal cord segmentation model achieved a validation accuracy of 96%, while the best classification model (VIPRnet) achieved a validation accuracy of 99%. Conclusion: Machine learning-assisted analysis of VCUS shows great promise in improving diagnostic accuracy over operator-dependent human interpretation.
Authors: Yi-Han Wang, Peng Zhao, Zhi-Hua Zhou
Abstract: Online eXp-concave Optimization (OXO) is a fundamental problem in online learning. The standard algorithm, Online Newton Step (ONS), balances statistical optimality and computational practicality, guaranteeing an optimal regret of $O(d \log T)$, where $d$ is the dimension and $T$ is the time horizon. ONS faces a computational bottleneck due to the Mahalanobis projections at each round. This step costs $\Omega(d^\omega)$ arithmetic operations for bounded domains, even for the unit ball, where $\omega \in (2,3]$ is the matrix-multiplication exponent. As a result, the total runtime can reach $\tilde{O}(d^\omega T)$, particularly when iterates frequently oscillate near the domain boundary. For Stochastic eXp-concave Optimization (SXO), computational cost is also a challenge. Deploying ONS with online-to-batch conversion for SXO requires $T = \tilde{O}(d/\epsilon)$ rounds to achieve an excess risk of $\epsilon$, and thereby necessitates an $\tilde{O}(d^{\omega+1}/\epsilon)$ runtime. A COLT'13 open problem posed by Koren [2013] asks for an SXO algorithm with runtime less than $\tilde{O}(d^{\omega+1}/\epsilon)$. This paper proposes a simple variant of ONS, LightONS, which reduces the total runtime to $O(d^2 T + d^\omega \sqrt{T \log T})$ while preserving the optimal $O(d \log T)$ regret. LightONS implies an SXO method with runtime $\tilde{O}(d^3/\epsilon)$, thereby answering the open problem. Importantly, LightONS preserves the elegant structure of ONS by leveraging domain-conversion techniques from parameter-free online learning to introduce a hysteresis mechanism that delays expensive Mahalanobis projections until necessary. This design enables LightONS to serve as an efficient plug-in replacement of ONS in broader scenarios, even beyond regret minimization, including gradient-norm adaptive regret, parametric stochastic bandits, and memory-efficient online learning.
Authors: Zhuo Zhang, Xi Yang, Yuan Zhao, Canqun Yang
Abstract: While Transformers have demonstrated remarkable potential in modeling Partial Differential Equations (PDEs), modeling large-scale unstructured meshes with complex geometries remains a significant challenge. Existing efficient architectures often employ feature dimensionality reduction strategies, which inadvertently induces Geometric Aliasing, resulting in the loss of critical physical boundary information. To address this, we propose the Physics-Geometry Operator Transformer (PGOT), designed to reconstruct physical feature learning through explicit geometry awareness. Specifically, we propose Spectrum-Preserving Geometric Attention (SpecGeo-Attention). Utilizing a ``physics slicing-geometry injection" mechanism, this module incorporates multi-scale geometric encodings to explicitly preserve multi-scale geometric features while maintaining linear computational complexity $O(N)$. Furthermore, PGOT dynamically routes computations to low-order linear paths for smooth regions and high-order non-linear paths for shock waves and discontinuities based on spatial coordinates, enabling spatially adaptive and high-precision physical field modeling. PGOT achieves consistent state-of-the-art performance across four standard benchmarks and excels in large-scale industrial tasks including airfoil and car designs.
Authors: Ziru Niu, Hai Dong, A. K. Qin, Tao Gu, Pengcheng Zhang
Abstract: Federated Learning (FL) has emerged as a privacy-preserving paradigm for training machine learning models across distributed edge devices in the Internet of Things (IoT). By keeping data local and coordinating model training through a central server, FL effectively addresses privacy concerns and reduces communication overhead. However, the limited computational power, memory, and bandwidth of IoT edge devices pose significant challenges to the efficiency and scalability of FL, especially when training deep neural networks. Various FL frameworks have been proposed to reduce computation and communication overheads through dropout or layer freezing. However, these approaches often sacrifice accuracy or neglect memory constraints. To this end, in this work, we introduce Federated Learning with Ordered Layer Freezing (FedOLF). FedOLF consistently freezes layers in a predefined order before training, significantly mitigating computation and memory requirements. To further reduce communication and energy costs, we incorporate Tensor Operation Approximation (TOA), a lightweight alternative to conventional quantization that better preserves model accuracy. Experimental results demonstrate that over non-iid data, FedOLF achieves at least 0.3%, 6.4%, 5.81%, 4.4%, 6.27% and 1.29% higher accuracy than existing works respectively on EMNIST (with CNN), CIFAR-10 (with AlexNet), CIFAR-100 (with ResNet20 and ResNet44), and CINIC-10 (with ResNet20 and ResNet44), along with higher energy efficiency and lower memory footprint.
Authors: Zihao Zhou, Shusen Yang, Fangyuan Zhao, Xuebin Ren
Abstract: Graph federated learning enables the collaborative extraction of high-order information from distributed subgraphs while preserving the privacy of raw data. However, graph data often exhibits overlap among different clients. Previous research has demonstrated certain benefits of overlapping data in mitigating data heterogeneity. However, the negative effects have not been explored, particularly in cases where the overlaps are imbalanced across clients. In this paper, we uncover the unfairness issue arising from imbalanced overlapping subgraphs through both empirical observations and theoretical reasoning. To address this issue, we propose FairGFL (FAIRness-aware subGraph Federated Learning), a novel algorithm that enhances cross-client fairness while maintaining model utility in a privacy-preserving manner. Specifically, FairGFL incorporates an interpretable weighted aggregation approach to enhance fairness across clients, leveraging privacy-preserving estimation of their overlapping ratios. Furthermore, FairGFL improves the tradeoff between model utility and fairness by integrating a carefully crafted regularizer into the federated composite loss function. Through extensive experiments on four benchmark graph datasets, we demonstrate that FairGFL outperforms four representative baseline algorithms in terms of both model utility and fairness.
Authors: Gang Liao, Hongsen Qin, Ying Wang, Alicia Golden, Michael Kuchnik, Yavuz Yetim, Jia Jiunn Ang, Chunli Fu, Yihan He, Samuel Hsia, Zewei Jiang, Dianshi Li, Uladzimir Pashkevich, Varna Puvvada, Feng Shi, Matt Steiner, Ruichao Xiao, Nathan Yan, Xiayu Yu, Zhou Fang, Abdul Zainul-Abedin, Ketan Singh, Hongtao Yu, Wenyuan Chi, Barney Huang, Sean Zhang, Noah Weller, Zach Marine, Wyatt Cook, Carole-Jean Wu, Gaoxiang Liu
Abstract: Making deep learning recommendation model (DLRM) training and inference fast and efficient is important. However, this presents three key system challenges - model architecture diversity, kernel primitive diversity, and hardware generation and architecture heterogeneity. This paper presents KernelEvolve-an agentic kernel coding framework-to tackle heterogeneity at-scale for DLRM. KernelEvolve is designed to take kernel specifications as input and automate the process of kernel generation and optimization for recommendation model across heterogeneous hardware architectures. KernelEvolve does so by operating at multiple programming abstractions, from Triton and CuTe DSL to low-level hardware agnostic languages, spanning the full hardware-software optimization stack. The kernel optimization process is described as graph-based search with selection policy, universal operator, fitness function, and termination rule, dynamically adapts to runtime execution context through retrieval-augmented prompt synthesis. We designed, implemented, and deployed KernelEvolve to optimize a wide variety of production recommendation models across generations of NVIDIA and AMD GPUs, as well as Meta's AI accelerators. We validate KernelEvolve on the publicly-available KernelBench suite, achieving 100% pass rate on all 250 problems across three difficulty levels, and 160 PyTorch ATen operators across three heterogeneous hardware platforms, demonstrating 100% correctness. KernelEvolve reduces development time from weeks to hours and achieves substantial performance improvements over PyTorch baselines across diverse production use cases and for heterogeneous AI systems at-scale. Beyond performance efficiency improvements, KernelEvolve significantly mitigates the programmability barrier for new AI hardware by enabling automated kernel generation for in-house developed AI hardware.
Authors: Tao Li, Peilin Li, Kui Lu, Yilei Wang, Junliang Shang, Guangshun Li, Huiyu Zhou
Abstract: The adverse drug reactions (ADRs) predicted based on the biased records in FAERS (U.S. Food and Drug Administration Adverse Event Reporting System) may mislead diagnosis online. Generally, such problems are solved by optimizing reporting odds ratio (ROR) or proportional reporting ratio (PRR). However, these methods that rely on statistical methods cannot eliminate the biased data, leading to inaccurate signal prediction. In this paper, we propose PFed-signal, a federated learning-based signal prediction model of ADR, which utilizes the Euclidean distance to eliminate the biased data from FAERS, thereby improving the accuracy of ADR prediction. Specifically, we first propose Pfed-Split, a method to split the original dataset into a split dataset based on ADR. Then we propose ADR-signal, an ADR prediction model, including a biased data identification method based on federated learning and an ADR prediction model based on Transformer. The former identifies the biased data according to the Euclidean distance and generates a clean dataset by deleting the biased data. The latter is an ADR prediction model based on Transformer trained on the clean data set. The results show that the ROR and PRR on the clean dataset are better than those of the traditional methods. Furthermore, the accuracy rate, F1 score, recall rate and AUC of PFed-Signal are 0.887, 0.890, 0.913 and 0.957 respectively, which are higher than the baselines.
Authors: Alexander Korotin, Gudmund Pammer
Abstract: This paper studies the inverse problem of flow matching (FM) between distributions with finite exponential moment, a problem motivated by modern generative AI applications such as the distillation of flow matching models. Uniqueness of the solution is established in two cases - the one-dimensional setting and the Gaussian case. The general multidimensional problem remains open for future studies.
Authors: Yuchen Xie, Honghang Chi, Haopeng Quan, Yahui Wang, Wei Wang, Yu Ma
Abstract: Physics-Informed Neural Networks with hard constraints (HC-PINNs) are increasingly favored for their ability to strictly enforce boundary conditions via a trial function ansatz $\tilde{u} = A + B \cdot N$, yet the theoretical mechanisms governing their training dynamics have remained unexplored. Unlike soft-constrained formulations where boundary terms act as additive penalties, this work reveals that the boundary function $B$ introduces a multiplicative spatial modulation that fundamentally alters the learning landscape. A rigorous Neural Tangent Kernel (NTK) framework for HC-PINNs is established, deriving the explicit kernel composition law. This relationship demonstrates that the boundary function $B(\vec{x})$ functions as a spectral filter, reshaping the eigenspectrum of the neural network's native kernel. Through spectral analysis, the effective rank of the residual kernel is identified as a deterministic predictor of training convergence, superior to classical condition numbers. It is shown that widely used boundary functions can inadvertently induce spectral collapse, leading to optimization stagnation despite exact boundary satisfaction. Validated across multi-dimensional benchmarks, this framework transforms the design of boundary functions from a heuristic choice into a principled spectral optimization problem, providing a solid theoretical foundation for geometric hard constraints in scientific machine learning.
Authors: Abolfazl Younesi, Abbas Shabrang Maryan, Elyas Oustad, Zahra Najafabadi Samani, Mohsen Ansari, Thomas Fahringer
Abstract: Deploying large language models (LLMs) on edge devices is challenging due to their limited memory and power resources. Cloud-only inference reduces device burden but introduces high latency and cost. Static edge-cloud partitions optimize a single metric and struggle when bandwidth fluctuates. We propose Splitwise, a novel Lyapunov-assisted deep reinforcement learning (DRL) framework for fine-grained, adaptive partitioning of LLMs across edge and cloud environments. Splitwise decomposes transformer layers into attention heads and feed-forward sub-blocks, exposing more partition choices than layer-wise schemes. A hierarchical DRL policy, guided by Lyapunov optimization, jointly minimizes latency, energy consumption, and accuracy degradation while guaranteeing queue stability under stochastic workloads and variable network bandwidth. Splitwise also guarantees robustness via partition checkpoints with exponential backoff recovery in case of communication failures. Experiments on Jetson Orin NX, Galaxy S23, and Raspberry Pi 5 with GPT-2 (1.5B), LLaMA-7B, and LLaMA-13B show that Splitwise reduces end-to-end latency by 1.4x-2.8x and cuts energy consumption by up to 41% compared with existing partitioners. It lowers the 95th-percentile latency by 53-61% relative to cloud-only execution, while maintaining accuracy and modest memory requirements.
Authors: Laurent Bou\'e
Abstract: This document is a follow-up to our previous paper dedicated to a vectorized derivation of backpropagation in CNNs. Following the same principles and notations already put in place there, we now focus on transformer-based next-token-prediction architectures. To this end, we apply our lightweight index-free methodology to new types of layers such as embedding, multi-headed self-attention and layer normalization. In addition, we also provide gradient expressions for LoRA layers to illustrate parameter-efficient fine-tuning. Why bother doing manual backpropagation when there are so many tools that do this automatically? Any gap in understanding of how values propagate forward will become evident when attempting to differentiate the loss function. By working through the backward pass manually, we gain a deeper intuition for how each operation influences the final output. A complete PyTorch implementation of a minimalistic GPT-like network is also provided along with analytical expressions for of all of its gradient updates.
Authors: Dakuan Lu, Jiaqi Zhang, Cheng Yuan, Jiawei Shao, Chi Zhang, Xuelong Li
Abstract: Recent advances in large language models (LLMs) have been largely driven by scaling laws for individual models, which predict performance improvements as model parameters and data volume increase. However, the capabilities of any single LLM are inherently bounded. One solution originates from intricate interactions among multiple LLMs, rendering their collective performance surpasses that of any constituent model. Despite the rapid proliferation of multi-model integration techniques such as model routing and post-hoc ensembling, a unifying theoretical framework of performance scaling for multi-model collaboration remains absent. In this work, we propose the Law of Multi-model Collaboration, a scaling law that predicts the performance limits of LLM ensembles based on their aggregated parameter budget. To quantify the intrinsic upper bound of multi-model collaboration, we adopt a method-agnostic formulation and assume an idealized integration oracle where the total cross-entropy loss of each sample is determined by the minimum loss of any model in the model pool. Experimental results reveal that multi-model systems follow a power-law scaling with respect to the total parameter count, exhibiting a more significant improvement trend and a lower theoretical loss floor compared to single model scaling. Moreover, ensembles of heterogeneous model families achieve better performance scaling than those formed within a single model family, indicating that model diversity is a primary driver of collaboration gains. These findings suggest that model collaboration represents a critical axis for extending the intelligence frontier of LLMs.
Authors: Hai Duong Nguyen, Xuan-The Tran
Abstract: Deep learning has achieved strong performance for electrocardiogram (ECG) classification within individual datasets, yet dependable generalization across heterogeneous acquisition settings remains a major obstacle to clinical deployment and longitudinal monitoring. A key limitation of many model architectures is the implicit entanglement of morphological waveform patterns and rhythm dynamics, which can promote shortcut learning and amplify sensitivity to distribution shifts. We propose ECG-RAMBA, a framework that separates morphology and rhythm and then re-integrates them through context-aware fusion. ECG-RAMBA combines: (i) deterministic morphological features extracted by MiniRocket, (ii) global rhythm descriptors computed from heart-rate variability (HRV), and (iii) long-range contextual modeling via a bi-directional Mamba backbone. To improve sensitivity to transient abnormalities under windowed inference, we introduce a numerically stable Power Mean pooling operator ($Q=3$) that emphasizes high-evidence segments while avoiding the brittleness of max pooling and the dilution of averaging. We evaluate under a protocol-faithful setting with subject-level cross-validation, a fixed decision threshold, and no test-time adaptation. On the Chapman--Shaoxing dataset, ECG-RAMBA achieves a macro ROC-AUC $\approx 0.85$. In zero-shot transfer, it attains PR-AUC $=0.708$ for atrial fibrillation detection on the external CPSC-2021 dataset, substantially outperforming a comparable raw-signal Mamba baseline, and shows consistent cross-dataset performance on PTB-XL. Ablation studies indicate that deterministic morphology provides a strong foundation, while explicit rhythm modeling and long-range context are critical drivers of cross-domain robustness.
Authors: Nilin Abrahamsen
Abstract: This note introduces Isometric Policy Optimization (ISOPO), an efficient method to approximate the natural policy gradient in a single gradient step. In comparison, existing proximal policy methods such as GRPO or CISPO use multiple gradient steps with variants of importance ratio clipping to approximate a natural gradient step relative to a reference policy. In its simplest form, ISOPO normalizes the log-probability gradient of each sequence in the Fisher metric before contracting with the advantages. Another variant of ISOPO transforms the microbatch advantages based on the neural tangent kernel in each layer. ISOPO applies this transformation layer-wise in a single backward pass and can be implemented with negligible computational overhead compared to vanilla REINFORCE.
Authors: Yilun Luo, HuaQing Zheng, Haoqian Meng, Wenyuan Liu, Peng Zhang
Abstract: Huawei's openPangu-Embedded-1B and openPangu-Embedded-7B, variants of the openPangu large language model, integrate three distinct Chain-of-Thought (CoT) reasoning paradigms, namely slow_think, auto_think, and no_think. While these CoT modes enhance reasoning capabilities, their generation of extended reasoning traces introduces substantial memory and latency overheads, posing challenges for practical deployment on Ascend NPUs. This paper addresses these computational constraints by leveraging low-bit quantization, which transforms FP16 computations into more efficient integer arithmetic. We introduce a unified low-bit inference framework, supporting INT8 (W8A8) and W4A8 quantization, specifically optimized for openPangu-Embedded models on the Atlas A2. Our comprehensive evaluation, conducted across all three CoT modes on code generation benchmarks (HumanEval and MBPP), demonstrates the efficacy of this approach. INT8 quantization consistently preserves over 90\% of the FP16 baseline accuracy and achieves a 1.5x prefill speedup on the Atlas A2. Furthermore, W4A8 quantization significantly reduces memory consumption, albeit with a moderate trade-off in accuracy. These findings collectively indicate that low-bit quantization effectively facilitates efficient CoT reasoning on Ascend NPUs, maintaining high model fidelity.
Authors: Xiao Ma, Mohammad Hasyim Taufik, Tariq Alkhalifah
Abstract: Velocity model building serves as a crucial component for achieving high precision subsurface imaging. However, conventional velocity model building methods are often computationally expensive and time consuming. In recent years, with the rapid advancement of deep learning, particularly the success of generative models and neural operators, deep learning based approaches that integrate data and their statistics have attracted increasing attention in addressing the limitations of traditional methods. In this study, we propose a novel framework that combines generative models with neural operators to obtain high resolution velocity models efficiently. Within this workflow, the neural operator functions as a forward mapping operator to rapidly generate time lag reverse time migration (RTM) extended images from the true and migration velocity models. In this framework, the neural operator is acting as a surrogate for modeling followed by migration, which uses the true and migration velocities, respectively. The trained neural operator is then employed, through automatic differentiation, to gradually update the migration velocity placed in the true velocity input channel with high resolution components so that the output of the network matches the time lag images of observed data obtained using the migration velocity. By embedding a generative model, trained on a high-resolution velocity model distribution, which corresponds to the true velocity model distribution used to train the neural operator, as a regularizer, the resulting predictions are cleaner with higher resolution information. Both synthetic and field data experiments demonstrate the effectiveness of the proposed generative neural operator based velocity model building approach.
Authors: Mohammad Nasirzadeh, Jafar Tahmoresnezhad, Parviz Rashidi-Khazaee
Abstract: Log anomaly detection is crucial for preserving the security of operating systems. Depending on the source of log data collection, various information is recorded in logs that can be considered log modalities. In light of this intuition, unimodal methods often struggle by ignoring the different modalities of log data. Meanwhile, multimodal methods fail to handle the interactions between these modalities. Applying multimodal sentiment analysis to log anomaly detection, we propose CoLog, a framework that collaboratively encodes logs utilizing various modalities. CoLog utilizes collaborative transformers and multi-head impressed attention to learn interactions among several modalities, ensuring comprehensive anomaly detection. To handle the heterogeneity caused by these interactions, CoLog incorporates a modality adaptation layer, which adapts the representations from different log modalities. This methodology enables CoLog to learn nuanced patterns and dependencies within the data, enhancing its anomaly detection capabilities. Extensive experiments demonstrate CoLog's superiority over existing state-of-the-art methods. Furthermore, in detecting both point and collective anomalies, CoLog achieves a mean precision of 99.63%, a mean recall of 99.59%, and a mean F1 score of 99.61% across seven benchmark datasets for log-based anomaly detection. The comprehensive detection capabilities of CoLog make it highly suitable for cybersecurity, system monitoring, and operational efficiency. CoLog represents a significant advancement in log anomaly detection, providing a sophisticated and effective solution to point and collective anomaly detection through a unified framework and a solution to the complex challenges automatic log data analysis poses. We also provide the implementation of CoLog at https://github.com/NasirzadehMoh/CoLog.
Authors: Nathan Buskulic, Luca Calatroni, Lorenzo Rosasco, Silvia Villa
Abstract: Blind inverse problems arise in many experimental settings where the forward operator is partially or entirely unknown. In this context, methods developed for the non-blind case cannot be adapted in a straightforward manner. Recently, data-driven approaches have been proposed to address blind inverse problems, demonstrating strong empirical performance and adaptability. However, these methods often lack interpretability and are not supported by rigorous theoretical guarantees, limiting their reliability in applied domains such as imaging inverse problems. In this work, we shed light on learning in blind inverse problems within the simplified yet insightful framework of Linear Minimum Mean Square Estimators (LMMSEs). We provide an in-depth theoretical analysis, deriving closed-form expressions for optimal estimators and extending classical results. In particular, we establish equivalences with suitably chosen Tikhonov-regularized formulations, where the regularization depends explicitly on the distributions of the unknown signal, the noise, and the random forward operators. We also prove convergence results under appropriate source condition assumptions. Furthermore, we derive rigorous finite-sample error bounds that characterize the performance of learned estimators as a function of the noise level, problem conditioning, and number of available samples. These bounds explicitly quantify the impact of operator randomness and reveal the associated convergence rates as this randomness vanishes. Finally, we validate our theoretical findings through illustrative numerical experiments that confirm the predicted convergence behavior.
Authors: Ayushman Raghuvanshi, Gonzalo Mateos, Sundeep Prabhakar Chepuri
Abstract: Graph neural networks (GNNs) often struggle to learn discriminative node representations for heterophilic graphs, where connected nodes tend to have dissimilar labels and feature similarity provides weak structural cues. We propose frequency-guided graph structure learning (FgGSL), an end-to-end graph inference framework that jointly learns homophilic and heterophilic graph structures along with a spectral encoder. FgGSL employs a learnable, symmetric, feature-driven masking function to infer said complementary graphs, which are processed using pre-designed low- and high-pass graph filter banks. A label-based structural loss explicitly promotes the recovery of homophilic and heterophilic edges, enabling task-driven graph structure learning. We derive stability bounds for the structural loss and establish robustness guarantees for the filter banks under graph perturbations. Experiments on six heterophilic benchmarks demonstrate that FgGSL consistently outperforms state-of-the-art GNNs and graph rewiring methods, highlighting the benefits of combining frequency information with supervised topology inference.
Authors: Huan Song, Qingfei Zhao, Ting Long, Shuyu Tian, Hongjun An, Jiawei Shao, Chi Zhang, Xuelong Li
Abstract: Neural scaling laws have become foundational for optimizing large language model (LLM) training, yet they typically assume a single dense model output. This limitation effectively overlooks "Familial models, a transformative paradigm essential for realizing ubiquitous intelligence across heterogeneous device-edge-cloud hierarchies. Transcending static architectures, familial models integrate early exits with relay-style inference to spawn G deployable sub-models from a single shared backbone. In this work, we theoretically and empirically extend the scaling law to capture this "one-run, many-models" paradigm by introducing Granularity (G) as a fundamental scaling variable alongside model size (N) and training tokens (D). To rigorously quantify this relationship, we propose a unified functional form L(N, D, G) and parameterize it using large-scale empirical runs. Specifically, we employ a rigorous IsoFLOP experimental design to strictly isolate architectural impact from computational scale. Across fixed budgets, we systematically sweep model sizes (N) and granularities (G) while dynamically adjusting tokens (D). This approach effectively decouples the marginal cost of granularity from the benefits of scale, ensuring high-fidelity parameterization of our unified scaling law. Our results reveal that the granularity penalty follows a multiplicative power law with an extremely small exponent. Theoretically, this bridges fixed-compute training with dynamic architectures. Practically, it validates the "train once, deploy many" paradigm, demonstrating that deployment flexibility is achievable without compromising the compute-optimality of dense baselines.
Authors: Yusuf Kalyoncuoglu
Abstract: While it is well-established that the weight matrices and feature manifolds of deep neural networks exhibit a low Intrinsic Dimension (ID), current state-of-the-art models still rely on massive high-dimensional widths. This redundancy is not required for representation, but is strictly necessary to solve the non-convex optimization search problem-finding a global minimum, which remains intractable for compact networks. In this work, we propose a constructive approach to bypass this optimization bottleneck. By decoupling the solution geometry from the ambient search space, we empirically demonstrate across ResNet-50, ViT, and BERT that the classification head can be compressed by even huge factors of 16 with negligible performance degradation. This motivates Subspace-Native Distillation as a novel paradigm: by defining the target directly in this constructed subspace, we provide a stable geometric coordinate system for student models, potentially allowing them to circumvent the high-dimensional search problem entirely and realize the vision of Train Big, Deploy Small.
Authors: Taha Emre, Arunava Chakravarty, Thomas Pinetz, Dmitrii Lachinov, Martin J. Menten, Hendrik Scholl, Sobha Sivaprasad, Daniel Rueckert, Andrew Lotery, Stefan Sacu, Ursula Schmidt-Erfurth, Hrvoje Bogunovi\'c
Abstract: Temporally aware image representations are crucial for capturing disease progression in 3D volumes of longitudinal medical datasets. However, recent state-of-the-art self-supervised learning approaches like Masked Autoencoding (MAE), despite their strong representation learning capabilities, lack temporal awareness. In this paper, we propose STAMP (Stochastic Temporal Autoencoder with Masked Pretraining), a Siamese MAE framework that encodes temporal information through a stochastic process by conditioning on the time difference between the 2 input volumes. Unlike deterministic Siamese approaches, which compare scans from different time points but fail to account for the inherent uncertainty in disease evolution, STAMP learns temporal dynamics stochastically by reframing the MAE reconstruction loss as a conditional variational inference objective. We evaluated STAMP on two OCT and one MRI datasets with multiple visits per patient. STAMP pretrained ViT models outperformed both existing temporal MAE methods and foundation models on different late stage Age-Related Macular Degeneration and Alzheimer's Disease progression prediction which require models to learn the underlying non-deterministic temporal dynamics of the diseases.
Authors: Vladimer Khasia
Abstract: Mixture of Experts (MoE) models scale capacity but often suffer from representation collapse and gradient instability. We propose Dynamic Subspace Composition (DSC), a framework that approximates context-dependent weights via a state-dependent, sparse expansion of a shared basis bank. Formally, DSC models the weight update as a residual trajectory within a Star- Shaped Domain, employing a Magnitude-Gated Simplex Interpolation to ensure continuity at the identity. Unlike standard Mixture-of-LoRAs, which incurs O(M rd) parameter complexity by retrieving independent rank-r matrices, DSC constructs a compositional rank-K approximation from decoupled unit-norm basis vectors. This reduces parameter complexity to O(M d) and memory traffic to O(Kd), while Frame-Theoretic regularization and spectral constraints provide rigorous worst-case bounds on the dynamic update. The code is available at https://github. com/VladimerKhasia/DSC
URLs: https://github.
Authors: Zhuo Li, Pengyu Cheng, Zhechao Yu, Feifei Tong, Anningzhe Gao, Tsung-Hui Chang, Xiang Wan, Erchao Zhao, Xiaoxi Jiang, Guanjun Jiang
Abstract: Reward models (RMs) are essential in reinforcement learning from human feedback (RLHF) to align large language models (LLMs) with human values. However, RM training data is commonly recognized as low-quality, containing inductive biases that can easily lead to overfitting and reward hacking. For example, more detailed and comprehensive responses are usually human-preferred but with more words, leading response length to become one of the inevitable inductive biases. A limited number of prior RM debiasing approaches either target a single specific type of bias or model the problem with only simple linear correlations, \textit{e.g.}, Pearson coefficients. To mitigate more complex and diverse inductive biases in reward modeling, we introduce a novel information-theoretic debiasing method called \textbf{D}ebiasing via \textbf{I}nformation optimization for \textbf{R}M (DIR). Inspired by the information bottleneck (IB), we maximize the mutual information (MI) between RM scores and human preference pairs, while minimizing the MI between RM outputs and biased attributes of preference inputs. With theoretical justification from information theory, DIR can handle more sophisticated types of biases with non-linear correlations, broadly extending the real-world application scenarios for RM debiasing methods. In experiments, we verify the effectiveness of DIR with three types of inductive biases: \textit{response length}, \textit{sycophancy}, and \textit{format}. We discover that DIR not only effectively mitigates target inductive biases but also enhances RLHF performance across diverse benchmarks, yielding better generalization abilities. The code and training recipes are available at https://github.com/Qwen-Applications/DIR.
Authors: Guoan Wan, Tianyu Chen, Fangzheng Feng, Haoyi Zhou, Runhua Xu
Abstract: Parameter-efficient fine-tuning (PEFT) methods have emerged as a practical solution for adapting large foundation models to downstream tasks, reducing computational and memory costs by updating only a small subset of parameters. Among them, approaches like LoRA aim to strike a balance between efficiency and expressiveness, but often suffer from slow convergence and limited adaptation capacity due to their inherent low-rank constraints. This trade-off hampers the ability of PEFT methods to capture complex patterns needed for diverse tasks. To address these challenges, we propose FRoD, a novel fine-tuning method that combines hierarchical joint decomposition with rotational degrees of freedom. By extracting a globally shared basis across layers and injecting sparse, learnable perturbations into scaling factors for flexible full-rank updates, FRoD enhances expressiveness and efficiency, leading to faster and more robust convergence. On 20 benchmarks spanning vision, reasoning, and language understanding, FRoD matches full model fine-tuning in accuracy, while using only 1.72% of trainable parameters under identical training budgets.
Authors: Vassilis Digalakis Jr, Ramayya Krishnan, Gonzalo Martin Fernandez, Agni Orfanoudaki
Abstract: We study how organizations should select among competing AI models when user utility, deployment costs, and compliance requirements jointly matter. Widely used capability leaderboards do not translate directly into deployment decisions, creating a capability -- deployment gap; to bridge it, we take a systems-level view in which model choice is tied to application outcomes, operating constraints, and a capability-cost frontier. We develop ML Compass, a framework that treats model selection as constrained optimization over this frontier. On the theory side, we characterize optimal model configurations under a parametric frontier and show a three-regime structure in optimal internal measures: some dimensions are pinned at compliance minima, some saturate at maximum levels, and the remainder take interior values governed by frontier curvature. We derive comparative statics that quantify how budget changes, regulatory tightening, and technological progress propagate across capability dimensions and costs. On the implementation side, we propose a pipeline that (i) extracts low-dimensional internal measures from heterogeneous model descriptors, (ii) estimates an empirical frontier from capability and cost data, (iii) learns a user- or task-specific utility function from interaction outcome data, and (iv) uses these components to target capability-cost profiles and recommend models. We validate ML Compass with two case studies: a general-purpose conversational setting using the PRISM Alignment dataset and a healthcare setting using a custom dataset we build using HealthBench. In both environments, our framework produces recommendations -- and deployment-aware leaderboards based on predicted deployment value under constraints -- that can differ materially from capability-only rankings, and clarifies how trade-offs between capability, cost, and safety shape optimal model choice.
Authors: Wei Gao, Paul Zheng, Peng Wu, Yulin Hu, Anke Schmeink
Abstract: In this article, we consider an industrial internet of things (IIoT) network supporting multi-device dynamic ultra-reliable low-latency communication (URLLC) while the channel state information (CSI) is imperfect. A joint link adaptation (LA) and device scheduling (including the order) design is provided, aiming at maximizing the total transmission rate under strict block error rate (BLER) constraints. In particular, a Bayesian optimization (BO) driven Twin Delayed Deep Deterministic Policy Gradient (TD3) method is proposed, which determines the device served order sequence and the corresponding modulation and coding scheme (MCS) adaptively based on the imperfect CSI. Note that the imperfection of CSI, error sample imbalance in URLLC networks, as well as the parameter sensitivity nature of the TD3 algorithm likely diminish the algorithm's convergence speed and reliability. To address such an issue, we proposed a BO based training mechanism for the convergence speed improvement, which provides a more reliable learning direction and sample selection method to track the imbalance sample problem. Via extensive simulations, we show that the proposed algorithm achieves faster convergence and higher sum-rate performance compared to existing solutions.
Authors: Zhuo Huang
Abstract: Machine Learning (ML) has been a foundational topic in artificial intelligence (AI), providing both theoretical groundwork and practical tools for its exciting advancements. From ResNet for visual recognition to Transformer for vision-language alignment, the AI models have achieved superior capability to humans. Furthermore, the scaling law has enabled AI to initially develop general intelligence, as demonstrated by Large Language Models (LLMs). To this stage, AI has had an enormous influence on society and yet still keeps shaping the future for humanity. However, distribution shift remains a persistent ``Achilles' heel'', fundamentally limiting the reliability and general usefulness of ML systems. Moreover, generalization under distribution shift would also cause trust issues for AIs. Motivated by these challenges, my research focuses on \textit{Trustworthy Machine Learning under Distribution Shifts}, with the goal of expanding AI's robustness, versatility, as well as its responsibility and reliability. We carefully study the three common distribution shifts into: (1) Perturbation Shift, (2) Domain Shift, and (3) Modality Shift. For all scenarios, we also rigorously investigate trustworthiness via three aspects: (1) Robustness, (2) Explainability, and (3) Adaptability. Based on these dimensions, we propose effective solutions and fundamental insights, meanwhile aiming to enhance the critical ML problems, such as efficiency, adaptability, and safety.
Authors: Maryam Mirzaei, Farzaneh Shayegh, Hamed Narimani
Abstract: Accurate recognition of human emotional states is critical for effective human-machine interaction. Electroencephalography (EEG) offers a reliable source for emotion recognition due to its high temporal resolution and its direct reflection of neural activity. Nevertheless, variations across recording sessions present a major challenge for model generalization. To address this issue, we propose EGDA, a framework that reduces cross-session discrepancies by jointly aligning the global (marginal) and class-specific (conditional) distributions, while preserving the intrinsic structure of EEG data through graph regularization. Experimental results on the SEED-IV dataset demonstrate that EGDA achieves robust cross-session performance, obtaining accuracies of 81.22%, 80.15%, and 83.27% across three transfer tasks, and surpassing several baseline methods. Furthermore, the analysis highlights the Gamma frequency band as the most discriminative and identifies the central-parietal and prefrontal brain regions as critical for reliable emotion recognition.
Authors: Zhehao Huang, Baijiong Lin, Jingyuan Zhang, Jingying Wang, Yuhang Liu, Ning Lu, Tao Li, Xiaolin Huang
Abstract: Multi-model routing has evolved from an engineering technique into essential infrastructure, yet existing work lacks a systematic, reproducible benchmark for evaluating vision-language models (VLMs). We present VL-RouterBench to assess the overall capability of VLM routing systems systematically. The benchmark is grounded in raw inference and scoring logs from VLMs and constructs quality and cost matrices over sample-model pairs. In scale, VL-RouterBench covers 14 datasets across 3 task groups, totaling 30,540 samples, and includes 15 open-source models and 2 API models, yielding 519,180 sample-model pairs and a total input-output token volume of 34,494,977. The evaluation protocol jointly measures average accuracy, average cost, and throughput, and builds a ranking score from the harmonic mean of normalized cost and accuracy to enable comparison across router configurations and cost budgets. On this benchmark, we evaluate 10 routing methods and baselines and observe a significant routability gain, while the best current routers still show a clear gap to the ideal Oracle, indicating considerable room for improvement in router architecture through finer visual cues and modeling of textual structure. We will open-source the complete data construction and evaluation toolchain to promote comparability, reproducibility, and practical deployment in multimodal routing research.
Authors: Christopher Burger
Abstract: Traditional Statistical Process Control (SPC) is essential for quality management but is limited by its reliance on often violated statistical assumptions, leading to unreliable monitoring in modern, complex manufacturing environments. This paper introduces a hybrid framework that enhances SPC by integrating the distribution free, model agnostic guarantees of Conformal Prediction. We propose two novel applications: Conformal-Enhanced Control Charts, which visualize process uncertainty and enable proactive signals like 'uncertainty spikes', and Conformal-Enhanced Process Monitoring, which reframes multivariate control as a formal anomaly detection problem using an intuitive p-value chart. Our framework provides a more robust and statistically rigorous approach to quality control while maintaining the interpretability and ease of use of classic methods.
Authors: Deniz Akdemir
Abstract: Distribution shift is the defining challenge of real-world machine learning. The dominant paradigm--Unsupervised Domain Adaptation (UDA)--enforces feature invariance, aligning source and target representations via symmetric divergence minimization [Ganin et al., 2016]. We demonstrate that this approach is fundamentally flawed: when domains are unequally informative (e.g., high-quality vs degraded sensors), strict invariance necessitates information destruction, causing "negative transfer" that can be catastrophic in safety-critical applications [Wang et al., 2019]. We propose a decision-theoretic framework grounded in Le Cam's theory of statistical experiments [Le Cam, 1986], using constructive approximations to replace symmetric invariance with directional simulability. We introduce Le Cam Distortion, quantified by the Deficiency Distance $\delta(E_1, E_2)$, as a rigorous upper bound for transfer risk conditional on simulability. Our framework enables transfer without source degradation by learning a kernel that simulates the target from the source. Across five experiments (genomics, vision, reinforcement learning), Le Cam Distortion achieves: (1) near-perfect frequency estimation in HLA genomics (correlation $r=0.999$, matching classical methods), (2) zero source utility loss in CIFAR-10 image classification (81.2% accuracy preserved vs 34.7% drop for CycleGAN), and (3) safe policy transfer in RL control where invariance-based methods suffer catastrophic collapse. Le Cam Distortion provides the first principled framework for risk-controlled transfer learning in domains where negative transfer is unacceptable: medical imaging, autonomous systems, and precision medicine.
Authors: Iris Xu, Guangtao Zeng, Zexue He, Charles Jin, Aldo Pareja, Dan Gutfreund, Chuang Gan, Zhang-Wei Hong
Abstract: Large language models (LLMs) have shown strong reasoning and coding capabilities, yet they struggle to generalize to real-world software engineering (SWE) problems that are long-horizon and out of distribution. Existing systems often rely on a single agent to handle the entire workflow-interpreting issues, navigating large codebases, and implementing fixes-within one reasoning chain. Such monolithic designs force the model to retain irrelevant context, leading to spurious correlations and poor generalization. Motivated by how human engineers decompose complex problems, we propose structuring SWE agents as orchestrators coordinating specialized sub-agents for sub-tasks such as localization, editing, and validation. The challenge lies in discovering effective hierarchies automatically: as the number of sub-agents grows, the search space becomes combinatorial, and it is difficult to attribute credit to individual sub-agents within a team. We address these challenges by formulating hierarchy discovery as a multi-armed bandit (MAB) problem, where each arm represents a candidate sub-agent and the reward measures its helpfulness when collaborating with others. This framework, termed Bandit Optimization for Agent Design (BOAD), enables efficient exploration of sub-agent designs under limited evaluation budgets. On SWE-bench-Verified, BOAD outperforms single-agent and manually designed multi-agent systems. On SWE-bench-Live, featuring more recent and out-of-distribution issues, our 36B system ranks second on the leaderboard at the time of evaluation, surpassing larger models such as GPT-4 and Claude. These results demonstrate that automatically discovered hierarchical multi-agent systems significantly improve generalization on challenging long-horizon SWE tasks. Code is available at https://github.com/iamxjy/BOAD-SWE-Agent.
Authors: Francesco Piatti, Thomas Cass, William F. Turner
Abstract: We introduce a training-efficient framework for time-series learning that combines random features with controlled differential equations (CDEs). In this approach, large randomly parameterized CDEs act as continuous-time reservoirs, mapping input paths to rich representations. Only a linear readout layer is trained, resulting in fast, scalable models with strong inductive bias. Building on this foundation, we propose two variants: (i) Random Fourier CDEs (RF-CDEs): these lift the input signal using random Fourier features prior to the dynamics, providing a kernel-free approximation of RBF-enhanced sequence models; (ii) Random Rough DEs (R-RDEs): these operate directly on rough-path inputs via a log-ODE discretization, using log-signatures to capture higher-order temporal interactions while remaining stable and efficient. We prove that in the infinite-width limit, these model induces the RBF-lifted signature kernel and the rough signature kernel, respectively, offering a unified perspective on random-feature reservoirs, continuous-time deep architectures, and path-signature theory. We evaluate both models across a range of time-series benchmarks, demonstrating competitive or state-of-the-art performance. These methods provide a practical alternative to explicit signature computations, retaining their inductive bias while benefiting from the efficiency of random features.
Authors: Arnuv Tandon, Karan Dalal, Xinhao Li, Daniel Koceja, Marcel R{\o}d, Sam Buchanan, Xiaolong Wang, Jure Leskovec, Sanmi Koyejo, Tatsunori Hashimoto, Carlos Guestrin, Jed McCaleb, Yejin Choi, Yu Sun
Abstract: We formulate long-context language modeling as a problem in continual learning rather than architecture design. Under this formulation, we only use a standard architecture -- a Transformer with sliding-window attention. However, our model continues learning at test time via next-token prediction on the given context, compressing the context it reads into its weights. In addition, we improve the model's initialization for learning at test time via meta-learning at training time. Overall, our method, a form of Test-Time Training (TTT), is End-to-End (E2E) both at test time (via next-token prediction) and training time (via meta-learning), in contrast to previous forms. We conduct extensive experiments with a focus on scaling properties. In particular, for 3B models trained with 164B tokens, our method (TTT-E2E) scales with context length in the same way as Transformer with full attention, while others, such as Mamba 2 and Gated DeltaNet, do not. However, similar to RNNs, TTT-E2E has constant inference latency regardless of context length, making it 2.7 times faster than full attention for 128K context. Our code is publicly available.
Authors: Shashwat Goel, Rishi Hazra, Dulhan Jayalath, Timon Willi, Parag Jain, William F. Shen, Ilias Leontiadis, Francesco Barbieri, Yoram Bachrach, Jonas Geiping, Chenxi Whitehouse
Abstract: AI co-scientists are emerging as a tool to assist human researchers in achieving their research goals. A crucial feature of these AI co-scientists is the ability to generate a research plan given a set of aims and constraints. The plan may be used by researchers for brainstorming, or may even be implemented after further refinement. However, language models currently struggle to generate research plans that follow all constraints and implicit requirements. In this work, we study how to leverage the vast corpus of existing research papers to train language models that generate better research plans. We build a scalable, diverse training corpus by automatically extracting research goals and goal-specific grading rubrics from papers across several domains. We then train models for research plan generation via reinforcement learning with self-grading. A frozen copy of the initial policy acts as the grader during training, with the rubrics creating a generator-verifier gap that enables improvements without external human supervision. To validate this approach, we conduct a study with human experts for machine learning research goals, spanning 225 hours. The experts prefer plans generated by our finetuned Qwen3-30B-A3B model over the initial model for 70% of research goals, and approve 84% of the automatically extracted goal-specific grading rubrics. To assess generality, we also extend our approach to research goals from medical papers, and new arXiv preprints, evaluating with a jury of frontier models. Our finetuning yields 12-22% relative improvements and significant cross-domain generalization, proving effective even in problem settings like medical research where execution feedback is infeasible. Together, these findings demonstrate the potential of a scalable, automated training recipe as a step towards improving general AI co-scientists.
Authors: Conor Wallace, Umer Siddique, Yongcan Cao
Abstract: Ad-hoc teamwork (AHT) requires agents to infer the behavior of previously unseen teammates and adapt their policy accordingly. Conventional approaches often rely on fixed probabilistic models or classifiers, which can be brittle under partial observability and limited interaction. Large language models (LLMs) offer a flexible alternative: by mapping short behavioral traces into high-level hypotheses, they can serve as world models over teammate behavior. We introduce \Collab, a language-based framework that classifies partner types using a behavior rubric derived from trajectory features, and extend it to \ReCollab, which incorporates retrieval-augmented generation (RAG) to stabilize inference with exemplar trajectories. In the cooperative Overcooked environment, \Collab effectively distinguishes teammate types, while \ReCollab consistently improves adaptation across layouts, achieving Pareto-optimal trade-offs between classification accuracy and episodic return. These findings demonstrate the potential of LLMs as behavioral world models for AHT and highlight the importance of retrieval grounding in challenging coordination settings.
Authors: Leyang Xue, Meghana Madhyastha, Myungjin Lee, Amos Storkey, Randal Burns, Mahesh K. Marina
Abstract: The ecosystem behind foundation model development today is highly centralized and limited to large-scale cloud data center operators: training foundation models is costly, needing immense compute resources. Decentralized foundation model training across edge devices, leveraging their spare compute, promises a democratized alternative. However, existing edge-training approaches fall short: they struggle to match cloud-based training performance, exhibit limited scalability with model size, exceed device memory capacity, and have prohibitive communication overhead. They also fail to satisfactorily handle device heterogeneity and dynamism. We introduce a new paradigm, Cleave, which finely partitions training operations through a novel selective hybrid tensor parallelism method. Together with a parameter server centric training framework, Cleave copes with device memory limits and avoids communication bottlenecks, thereby enabling efficient training of large models on par with the cloud. Further, with a cost optimization model to guide device selection and training workload distribution, Cleave effectively accounts for device heterogeneity and churn. Our evaluations show that Cleave matches cloud-based GPU training by scaling efficiently to larger models and thousands of devices, supporting up to 8x more devices than baseline edge-training approaches. It outperforms state-of-the-art edge training methods by up to a factor of 10 in per-batch training time and efficiently handles device failures, achieving at least 100x faster recovery than prior methods.
Authors: Gaofeng Dong, Kang Yang, Mani Srivastava
Abstract: Existing Wi-Fi sensing systems rely on injecting high-rate probing packets to extract channel state information (CSI), leading to communication degradation and poor deployability. Although Integrated Sensing and Communication (ISAC) is a promising direction, existing solutions still rely on auxiliary packet injection because they exploit only CSI from data frames. We present UniFi, the first Wi-Fi-based ISAC framework that fully eliminates intrusive packet injection by directly exploiting irregularly sampled CSI from diverse communication packets across multiple frequency bands. UniFi integrates a CSI sanitization pipeline to harmonize heterogeneous packets and remove burst-induced redundancy, together with a time-aware attention model that learns directly from non-uniform CSI sequences without resampling. We further introduce CommCSI-HAR, the first dataset with irregularly sampled CSI from real-world dual-band communication traffic. Extensive evaluations on this dataset and four public benchmarks show that UniFi achieves state-of-the-art accuracy with a compact model size, while fully preserving communication throughput.
Authors: Xiao Zhou, Yuze Sun, Jie Wu, Xiaomeng Huang
Abstract: Accurately defining the life cycle of the Madden-Julian Oscillation (MJO), the dominant mode of intraseasonal climate variability, remains a foundational challenge due to its propagating nature. The established linear-projection method (RMM index) often conflates mathematical artifacts with physical states, while direct clustering in raw data space is confounded by a "propagation penalty." Here, we introduce an "AI-for-theory" paradigm to objectively discover the MJO's intrinsic structure. We develop a deep learning model, PhysAnchor-MJO-AE, to learn a latent representation where vector distance corresponds to physical-feature similarity, enabling objective clustering of MJO dynamical states. Clustering these "MJO fingerprints" reveals the first complete, six-phase anatomical map of its life cycle. This taxonomy refines and critically completes the classical view by objectively isolating two long-hypothesized transitional phases: organizational growth over the Indian Ocean and the northward shift over the Philippine Sea. Derived from this anatomy, we construct a new physics-coherent monitoring framework that decouples location and intensity diagnostics. This framework reduces the rates of spurious propagation and convective misplacement by over an order of magnitude compared to the classical index. Our work transforms AI from a forecasting tool into a discovery microscope, establishing a reproducible template for extracting fundamental dynamical constructs from complex systems.
Authors: Hanbeot Park, Yunjeong Cho, Hunhee Kim
Abstract: Restoring speech communication from neural signals is a central goal of brain-computer interface research, yet EEG-based speech reconstruction remains challenging due to limited spatial resolution, susceptibility to noise, and the absence of temporally aligned acoustic targets in imagined speech. In this study, we propose an EEG-to-Voice paradigm that directly reconstructs speech from non-invasive EEG signals without dynamic time warping (DTW) or explicit temporal alignment. The proposed pipeline generates mel-spectrograms from EEG in an open-loop manner using a subject-specific generator, followed by pretrained vocoder and automatic speech recognition (ASR) modules to synthesize speech waveforms and decode text. Separate generators were trained for spoken speech and imagined speech, and transfer learning-based domain adaptation was applied by pretraining on spoken speech and adapting to imagined speech. A minimal language model-based correction module was optionally applied to correct limited ASR errors while preserving semantic structure. The framework was evaluated under 2 s and 4 s speech conditions using acoustic-level metrics (PCC, RMSE, MCD) and linguistic-level metrics (CER, WER). Stable acoustic reconstruction and comparable linguistic accuracy were observed for both spoken speech and imagined speech. While acoustic similarity decreased for longer utterances, text-level decoding performance was largely preserved, and word-position analysis revealed a mild increase in decoding errors toward later parts of sentences. The language model-based correction consistently reduced CER and WER without introducing semantic distortion. These results demonstrate the feasibility of direct, open-loop EEG-to-Voice reconstruction for spoken speech and imagined speech without explicit temporal alignment.
Authors: Ruifan Chu, Anbang Wang, Xiuxiu Bai, Shuai Liu, Xiaoshe Dong
Abstract: In high-performance computing, hotspot GPU kernels are primary bottlenecks, and expert manual tuning is costly and hard to port. Large language model methods often assume kernels can be compiled and executed cheaply, which fails in large applications where full builds and runs are expensive. We present an end-to-end LLM framework with performance feedback that optimizes kernels without building the full application. From independently extracted hotspot kernels, it automatically completes code into a Minimal Executable Program (MEP), then performs multi-round iterative optimization and evaluation outside the full application. The framework integrates Automatic Error Repair and Performance Pattern Inheritance to fix faults, preserve correctness, reuse effective tiling/memory/synchronization strategies, and reduce search cost. Optimized variants are reintegrated into the original application for validation. We evaluate on NVIDIA GPUs and the Haiguang Deep Computing Unit (DCU) platform (AMD-licensed architecture) using PolyBench, the AMD APP SDK, and hotspot kernels from large-scale supercomputing applications. The method achieves average speedups of 5.05x (PolyBench on NVIDIA), 7.77x (PolyBench on DCU), 1.77x (AMD APP SDK), and 1.25x on three hotspot kernels, surpassing direct LLM optimization. The approach requires no full-source dependencies, offers cross-platform portability, and enables practical, low-cost GPU kernel optimization.
Authors: Emna Bouzid, Noura Baccar, Kamran Iqbal, Yassine Chaouch, Fares Ben Youssef, Amine Regayeg, Sarra Toumi, Houda Nsir, Amina Mseddi, Leila Costelle
Abstract: As agriculture faces increasing pressure from water scarcity, especially in regions like Tunisia, innovative, resource-efficient solutions are urgently needed. This work explores the integration of indoor vertical hydroponics with Machine Learning (ML) techniques to optimize basil yield while saving water. This research develops a prediction system that uses different ML models and assesses their performance. The models were systematically trained and tested using data collected from IoT sensors of various environmental parameters like CO2, light. The experimental setup features 21 basil crops and uses Raspberry Pi and Arduino. 10k data points were collected and used to train and evaluate three ML models: Linear Regression (LR), Long Short-Term Memory (LSTM), and Deep Neural Networks (DNN). The comparative analysis of the performance of each model revealed that, while LSTM showed high predictive capability and accuracy of 99%, its execution time was 10 times longer than LR and its RAM usage was about 3 times higher than DNN's when simulated on a standard CPU environment. Conversely, the DNN model had an accuracy rate of 98%. This proves an efficient balance between computational speed and prediction quality, which makes this model well-suited for real-life deployment. Moreover, LR excelled in fast processing of basic prediction with an execution time of 11 seconds. This makes the LR model more suitable for low-complexity or resource-limited applications. These performance trade-offs highlight the potential of DNN-based solutions for building responsive, high-accuracy decision-support systems tailored to agricultural environments, making it suitable for future edge-device deployment.
Authors: Daria Botvynko (Lab-STICC\_OSE, IMT Atlantique - MEE, IMT Atlantique), Pierre Hasl\'ee (Lab-STICC\_OSE, IMT Atlantique - MEE, IMT Atlantique), Lucile Gaultier (ODL), Bertrand Chapron (LOPS), Clement de Boyer Mont\'egut (LOPS), Anass El Aouni (MOi), Julien Le Sommer (IGE), Ronan Fablet (IMT Atlantique - MEE, Lab-STICC\_OSE, ODYSSEY)
Abstract: We present an end-to-end deep learning framework for short-term forecasting of global sea surface dynamics based on sparse satellite altimetry data. Building on two state-of-the-art architectures: U-Net and 4DVarNet, originally developed for image segmentation and spatiotemporal interpolation respectively, we adapt the models to forecast the sea level anomaly and sea surface currents over a 7-day horizon using sequences of sparse nadir altimeters observations. The model is trained on data from the GLORYS12 operational ocean reanalysis, with synthetic nadir sampling patterns applied to simulate realistic observational coverage. The forecasting task is formulated as a sequence-to-sequence mapping, with the input comprising partial sea level anomaly (SLA) snapshots and the target being the corresponding future full-field SLA maps. We evaluate model performance using (i) normalized root mean squared error (nRMSE), (ii) averaged effective resolution, (iii) percentage of correctly predicted velocities magnitudes and angles, and benchmark results against the operational Mercator Ocean forecast product. Results show that end-to-end neural forecasts outperform the baseline across all lead times, with particularly notable improvements in high variability regions. Our framework is developed within the OceanBench benchmarking initiative, promoting reproducibility and standardized evaluation in ocean machine learning. These results demonstrate the feasibility and potential of end-to-end neural forecasting models for operational oceanography, even in data-sparse conditions.
Authors: Nicolas Zilberstein, Santiago Segarra, Luiz Chamon
Abstract: We introduce shielded Langevin Monte Carlo (LMC), a constrained sampler inspired by navigation functions, capable of sampling from unnormalized target distributions defined over punctured supports. In other words, this approach samples from non-convex spaces defined as convex sets with convex holes. This defines a novel and challenging problem in constrained sampling. To do so, the sampler incorporates a combination of a spatially adaptive temperature and a repulsive drift to ensure that samples remain within the feasible region. Experiments on a 2D Gaussian mixture and multiple-input multiple-output (MIMO) symbol detection showcase the advantages of the proposed shielded LMC in contrast to unconstrained cases.
Authors: Md Shahriar Sajid, Abhijit Kumar Ghosh, Fariha Nusrat
Abstract: Electroencephalography (EEG) signals contain rich temporal-spectral structure but are difficult to model due to noise, subject variability, and multi-scale dynamics. Lightweight deep learning models have shown promise, yet many either rely solely on local convolutions or require heavy recurrent modules. This paper presents PaperNet, a compact hybrid architecture that combines temporal convolutions, a channel-wise residual attention module, and a lightweight bidirectional recurrent block which is used for short-window classification. Using the publicly available BEED: Bangalore EEG Epilepsy Dataset, we evaluate PaperNet under a clearly defined subject-independent training protocol and compare it against established and widely used lightweight baselines. The model achieves a macro-F1 of 0.96 on the held-out test set with approximately 0.6M parameters, while maintaining balanced performance across all four classes. An ablation study demonstrates the contribution of temporal convolutions, residual attention, and recurrent aggregation. Channel-wise attention weights further offer insights into electrode relevance. Computational profiling shows that PaperNet remains efficient enough for practical deployment on resource-constrained systems through out the whole process. These results indicate that carefully combining temporal filtering, channel reweighting, and recurrent context modeling can yield strong EEG classification performance without excessive computational cost.
Authors: Hassan Khalid, Muhammad Mahad Khaliq, Muhammad Jawad Bashir
Abstract: The rapid evolution of Artificial intelligence in healthcare has opened avenues for enhancing various processes, including medical billing and transcription. This paper introduces an innovative approach by integrating AI with Locally Linear Embedding (LLE) to revolutionize the handling of high-dimensional medical data. This AI-enhanced LLE model is specifically tailored to improve the accuracy and efficiency of medical billing systems and transcription services. By automating these processes, the model aims to reduce human error and streamline operations, thereby facilitating faster and more accurate patient care documentation and financial transactions. This paper provides a comprehensive mathematical model of AI-enhanced LLE, demonstrating its application in real-world healthcare scenarios through a series of experiments. The results indicate a significant improvement in data processing accuracy and operational efficiency. This study not only underscores the potential of AI-enhanced LLE in medical data analysis but also sets a foundation for future research into broader healthcare applications.
Authors: Kun-Woo Shin (Seoul National University, Korea), Jay H. Park (Samsung Electronics, Korea), Moonwook Oh (Samsung Electronics, Korea), Yohan Jo (Seoul National University, Korea), Jaeyoung Do (Seoul National University, Korea), Sang-Won Lee (Seoul National University, Korea)
Abstract: We observe two major trends in LLM-based generative AI: (1) inference is becoming the dominant factor in terms of cost and power consumption, surpassing training, and (2) retrieval augmented generation (RAG) is becoming prevalent. When processing long inputs in RAG, the prefill phase of computing the key-value vectors of input text is energy-intensive and time-consuming even with high-end GPUs. Thus, it is crucial to make the prefill phase in RAG inference efficient. To address this issue, we propose MatKV, a scheme that precomputes the key-value vectors (KVs) of RAG objects (e.g., documents), materializes them in inexpensive but fast and power-efficient flash storage, and reuses them at inference time instead of recomputing the KVs using costly and power-inefficient GPU. Experimental results using Hugging Face's Transformers library across state-of-the-art GPUs and flash memory SSDs confirm that, compared to full KV computation on GPUs, MatKV reduces both inference time and power consumption by half for RAG workloads, without severely impacting accuracy in the question-answering task. Furthermore, we demonstrate that MatKV enables additional optimizations in two ways. First, a GPU can decode text while simultaneously loading the materialized KVs for the next instance, reducing load latency. Second, since decoding speed is less sensitive to GPU performance than KV computation, low-end GPUs can be leveraged for decoding without significantly compromising speed once the materialized KVs are loaded into GPU memory. These findings underscore MatKV's potential to make large-scale generative AI applications more cost-effective, power-efficient, and accessible across a wider range of tasks and hardware environments.
Authors: Qizhi Wang
Abstract: Digital-humanities work on semantic shift often alternates between handcrafted close readings and opaque embedding machinery. We present a reproducible expert-system style pipeline that quantifies and visualises lexical drift in the Old Bailey Corpus (1720--1913), coupling interpretable trajectories with legally meaningful axes. We bin proceedings by decade with dynamic merging for low-resource slices, train skip-gram embeddings, align spaces through orthogonal Procrustes, and measure both geometric displacement and neighborhood turnover. Three visual analytics outputs, which are drift magnitudes, semantic trajectories, and movement along a mercy-versus-retribution axis, expose how justice, crime, poverty, and insanity evolve with penal reforms, transportation debates, and Victorian moral politics. The pipeline is implemented as auditable scripts so results can be reproduced in other historical corpora.
Authors: Yogeswar Reddy Thota
Abstract: Modern deep residual networks perform substantial redundant computation by evaluating all residual blocks for every input, even when identity mappings suffice. We introduce CosineGate, an end-to-end differentiable architecture for dynamic routing in residual networks that uses cosine incompatibility between identity and residual feature representations as a self-supervised skip signal. CosineGate measures semantic redundancy through the Cosine Incompatibility Ratio (CIR), defined as 1 - cos(x, F(x)), and uses Gumbel-Softmax relaxation to enable per-sample, per-block gating during training. A progressive FLOPs regularization term controls average compute usage without destabilizing optimization. On CIFAR-10, CosineGate spans the accuracy-efficiency Pareto frontier: an aggressive configuration achieves 89.9 percent accuracy with 24.1 percent FLOPs savings, a balanced configuration achieves 91.3 percent accuracy with 28.5 percent savings at epoch 160, and a conservative configuration reaches a peak of 93.2 percent accuracy with minimal compute reduction. These results match or exceed ResNet-20 (91.3 percent) while reducing computation, without auxiliary supervision, distillation, or task-specific heuristics. Our results demonstrate that simple geometric measures of feature incompatibility provide a principled and effective signal for dynamic residual routing.
Authors: Pu Zhao, Xuan Shen, Zhenglun Kong, Yixin Shen, Sung-En Chang, Arash Akbari, Timothy Rupprecht, Lei Lu, Enfu Nan, Changdi Yang, Yumei He, Weiyan Shi, Xingchen Xu, Yu Huang, Wei Jiang, Wei Wang, Yue Chen, Yong He, Yanzhi Wang
Abstract: Recently, Large Language Models (LLMs) have undergone a significant transformation, marked by a rapid rise in both their popularity and capabilities. Leading this evolution are proprietary LLMs like GPT-4 and GPT-o1, which have captured widespread attention in the AI community due to their remarkable performance and versatility. Simultaneously, open-source LLMs, such as LLaMA and Mistral, have made great contributions to the ever-increasing popularity of LLMs due to the ease to customize and deploy the models across diverse applications. Moxin 7B is introduced as a fully open-source LLM developed in accordance with the Model Openness Framework, which moves beyond the simple sharing of model weights to embrace complete transparency in training, datasets, and implementation detail, thus fostering a more inclusive and collaborative research environment that can sustain a healthy open-source ecosystem. To further equip Moxin with various capabilities in different tasks, we develop three variants based on Moxin, including Moxin-VLM, Moxin-VLA, and Moxin-Chinese, which target the vision-language, vision-language-action, and Chinese capabilities, respectively. Experiments show that our models achieve superior performance in various evaluations. We adopt open-source framework and open data for the training. We release our models, along with the available data and code to derive these models.
Authors: Xinhao Cheng, Zhihao Zhang, Yu Zhou, Jianan Ji, Jinchen Jiang, Zepeng Zhao, Ziruo Xiao, Zihao Ye, Yingyi Huang, Ruihang Lai, Hongyi Jin, Bohan Hou, Mengdi Wu, Yixin Dong, Anthony Yip, Zihao Ye, Songting Wang, Wenqin Yang, Xupeng Miao, Tianqi Chen, Zhihao Jia
Abstract: We introduce Mirage Persistent Kernel (MPK), the first compiler and runtime system that automatically transforms multi-GPU model inference into a single high-performance megakernel. MPK introduces an SM-level graph representation that captures data dependencies at the granularity of individual streaming multiprocessors (SMs), enabling cross-operator software pipelining, fine-grained kernel overlap, and other previously infeasible GPU optimizations. The MPK compiler lowers tensor programs into highly optimized SM-level task graphs and generates optimized CUDA implementations for all tasks, while the MPK in-kernel parallel runtime executes these tasks within a single mega-kernel using decentralized scheduling across SMs. Together, these components provide end-to-end kernel fusion with minimal developer effort, while preserving the flexibility of existing programming models. Our evaluation shows that MPK significantly outperforms existing kernel-per-operator LLM serving systems by reducing end-to-end inference latency by up to 1.7x, pushing LLM inference performance close to hardware limits. MPK is publicly available at https://github.com/mirage-project/mirage.
Authors: Naishan Zheng, Jie Huang, Qingpei Guo, Feng Zhao
Abstract: Understanding long videos with multimodal large language models (MLLMs) remains challenging due to the heavy redundancy across frames and the need for temporally coherent representations. Existing static strategies, such as sparse sampling, frame compression, and clustering, are optimized for offline settings and often produce fragmented or over-compressed outputs when applied to continuous video streams. We present VideoScaffold, a dynamic representation framework designed for streaming video understanding. It adaptively adjusts event granularity according to video duration while preserving fine-grained visual semantics. VideoScaffold introduces two key components: Elastic-Scale Event Segmentation (EES), which performs prediction-guided segmentation to dynamically refine event boundaries, and Hierarchical Event Consolidation (HEC), which progressively aggregates semantically related segments into multi-level abstractions. Working in concert, EES and HEC enable VideoScaffold to transition smoothly from fine-grained frame understanding to abstract event reasoning as the video stream unfolds. Extensive experiments across both offline and streaming video understanding benchmarks demonstrate that VideoScaffold achieves state-of-the-art performance. The framework is modular and plug-and-play, seamlessly extending existing image-based MLLMs to continuous video comprehension. The code is available at https://github.com/zheng980629/VideoScaffold.
Authors: Sophie Zhao
Abstract: Recent work has shown that transformer-based language models learn rich geometric structure in their embedding spaces, yet the presence of higher-level cognitive organization within these representations remains underexplored. In this work, we investigate whether sentence embeddings encode a graded, hierarchical structure aligned with human-interpretable cognitive or psychological attributes. We construct a dataset of 480 natural-language sentences annotated with continuous ordinal energy scores and discrete tier labels spanning seven ordered cognitive categories. Using fixed sentence embeddings from multiple transformer models, we evaluate the recoverability of these annotations via linear and shallow nonlinear probes. Across models, both continuous scores and tier labels are reliably decodable, with shallow nonlinear probes providing consistent performance gains over linear probes. Lexical TF-IDF baselines perform substantially worse, indicating that the observed structure is not attributable to surface word statistics alone. Nonparametric permutation tests further confirm that probe performance exceeds chance under label-randomization nulls. Qualitative analyses using UMAP visualizations and confusion matrices reveal smooth low-to-high gradients and predominantly adjacent-tier confusions in embedding space. Taken together, these results provide evidence that transformer embedding spaces exhibit a hierarchical geometric organization aligned with human-defined cognitive attributes, while remaining agnostic to claims of internal awareness or phenomenology.
Authors: Adrien Pav\~ao
Abstract: AI-generated images are now pervasive online, yet many people believe they can easily tell them apart from real photographs. We test this assumption through an interactive web experiment where participants classify 20 images as real or AI-generated. Our dataset contains 120 difficult cases: real images sampled from CC12M, and carefully curated AI-generated counterparts produced with MidJourney. In total, 165 users completed 233 sessions. Their average accuracy was 54%, only slightly above random guessing, with limited improvement across repeated attempts. Response times averaged 7.3 seconds, and some images were consistently more deceptive than others. These results indicate that, even on relatively simple portrait images, humans struggle to reliably detect AI-generated content. As synthetic media continues to improve, human judgment alone is becoming insufficient for distinguishing real from artificial data. These findings highlight the need for greater awareness and ethical guidelines as AI-generated media becomes increasingly indistinguishable from reality.
Authors: Sarthak Sarkar, Supratim Das, Purushottam Saha, Diganta Mukherjee, Tridib Mukherjee
Abstract: Online fantasy cricket has emerged as large-scale competitive systems in which participants construct virtual teams and compete based on real-world player performances. This massive growth has been accompanied by important questions about whether outcomes are primarily driven by skill or chance. We develop a statistical framework to assess the role of skill in determining success on these platforms. We construct and analyze a range of deterministic and stochastic team selection strategies, based on recent form, historical statistics, statistical optimization, and multi-criteria decision making. Strategy performance is evaluated based on points, ranks, and payoff under two contest structures Mega and 4x or Nothing. An extensive comparison between different strategies is made to find an optimal set of strategies. To capture adaptive behavior, we further introduce a dynamic tournament model in which agent populations evolve through a softmax reweighting mechanism proportional to positive payoff realizations. We demonstrate our work by running extensive numerical experiments on the IPL 2024 dataset. The results provide quantitative evidence in favor of the skill element present in online fantasy cricket platforms.
Authors: Abhranil Chandra, Ayush Agrawal, Arian Hosseini, Sebastian Fischmeister, Rishabh Agarwal, Navin Goyal, Aaron Courville
Abstract: We present the surprising finding that a language model's reasoning capabilities can be improved by training on synthetic datasets of chain-of-thought (CoT) traces from more capable models, even when all of those traces lead to an incorrect final answer. Our experiments show this approach can yield better performance on reasoning tasks than training on human-annotated datasets. We hypothesize that two key factors explain this phenomenon: first, the distribution of synthetic data is inherently closer to the language model's own distribution, making it more amenable to learning. Second, these `incorrect' traces are often only partially flawed and contain valid reasoning steps from which the model can learn. To further test the first hypothesis, we use a language model to paraphrase human-annotated traces -- shifting their distribution closer to the model's own distribution -- and show that this improves performance. For the second hypothesis, we introduce increasingly flawed CoT traces and study to what extent models are tolerant to these flaws. We demonstrate our findings across various reasoning domains like math, algorithmic reasoning and code generation using MATH, GSM8K, Countdown and MBPP datasets on various language models ranging from 1.5B to 9B across Qwen, Llama, and Gemma models. Our study shows that curating datasets that are closer to the model's distribution is a critical aspect to consider. We also show that a correct final answer is not always a reliable indicator of a faithful reasoning process.
Authors: Satvik Tripathi
Abstract: Large language models (LLMs) excel at natural language reasoning but remain unreliable on tasks requiring strict rule adherence, determinism, and auditability. Logic Sketch Prompting (LSP) is a lightweight prompting framework that introduces typed variables, deterministic condition evaluators, and a rule based validator that produces traceable and repeatable outputs. Using two pharmacologic logic compliance tasks, we benchmark LSP against zero shot prompting, chain of thought prompting, and concise prompting across three open weight models: Gemma 2, Mistral, and Llama 3. Across both tasks and all models, LSP consistently achieves the highest accuracy (0.83 to 0.89) and F1 score (0.83 to 0.89), substantially outperforming zero shot prompting (0.24 to 0.60), concise prompts (0.16 to 0.30), and chain of thought prompting (0.56 to 0.75). McNemar tests show statistically significant gains for LSP across nearly all comparisons (p < 0.01). These results demonstrate that LSP improves determinism, interpretability, and consistency without sacrificing performance, supporting its use in clinical, regulated, and safety critical decision support systems.
Authors: Piotr Keller, Mark Eastwood, Zedong Hu, Aim\'ee Selten, Ruqayya Awan, Gertjan Rasschaert, Sara Verbandt, Vlad Popovici, Hubert Piessevaux, Hayley T Morris, Petros Tsantoulis, Thomas Alexander McKee, Andr\'e D'Hoore, C\'edric Schraepen, Xavier Sagaert, Gert De Hertogh, Sabine Tejpar, Fayyaz Minhas
Abstract: Routine histology contains rich prognostic information in stage II/III colorectal cancer, much of which is embedded in complex spatial tissue organisation. We present INSIGHT, a graph neural network that predicts survival directly from routine histology images. Trained and cross-validated on TCGA (n=342) and SURGEN (n=336), INSIGHT produces patient-level spatially resolved risk scores. Large independent validation showed superior prognostic performance compared with pTNM staging (C-index 0.68-0.69 vs 0.44-0.58). INSIGHT spatial risk maps recapitulated canonical prognostic histopathology and identified nuclear solidity and circularity as quantitative risk correlates. Integrating spatial risk with data-driven spatial transcriptomic signatures, spatial proteomics, bulk RNA-seq, and single-cell references revealed an epithelium-immune risk manifold capturing epithelial dedifferentiation and fetal programs, myeloid-driven stromal states including $\mathrm{SPP1}^{+}$ macrophages and $\mathrm{LAMP3}^{+}$ dendritic cells, and adaptive immune dysfunction. This analysis exposed patient-specific epithelial heterogeneity, stratification within MSI-High tumours, and high-risk routes of CDX2/HNF4A loss and CEACAM5/6-associated proliferative programs, highlighting coordinated therapeutic vulnerabilities.
Authors: Aahan Sachdeva, Dhanvinkumar Ganeshkumar, James E. Gallagher, Tyler Treat, Edward J. Oughton
Abstract: Autonomous robotic platforms are playing a growing role across the emergency services sector, supporting missions such as search and rescue operations in disaster zones and reconnaissance. However, traditional red-green-blue (RGB) detection pipelines struggle in low-light environments, and thermal-based systems lack color and texture information. To overcome these limitations, we present an adaptive framework that fuses RGB and long-wave infrared (LWIR) video streams at multiple fusion ratios and dynamically selects the optimal detection model for each illumination condition. We trained 33 You Only Look Once (YOLO) models on over 22,000 annotated images spanning three light levels: no-light (<10 lux), dim-light (10-1000 lux), and full-light (>1000 lux). To integrate both modalities, fusion was performed by blending aligned RGB and LWIR frames at eleven ratios, from full RGB (100/0) to full LWIR (0/100) in 10% increments. Evaluation showed that the best full-light model (80/20 RGB-LWIR) and dim-light model (90/10 fusion) achieved 92.8% and 92.0% mean confidence; both significantly outperformed the YOLOv5 nano (YOLOv5n) and YOLOv11 nano (YOLOv11n) baselines. Under no-light conditions, the top 40/60 fusion reached 71.0%, exceeding baselines though not statistically significant. Adaptive RGB-LWIR fusion improved detection confidence and reliability across all illumination conditions, enhancing autonomous robotic vision performance.
Authors: Qianqian Qi, Peter G. M. van der Heijden
Abstract: Across fields such as machine learning, social science, geography, considerable attention has been given to models that factorize a nonnegative matrix into the product of two or three matrices, subject to nonnegative or row-sum-to-1 constraints. Although these models are to a large extend similar or even equivalent, they are presented under different names, and their similarity is not well known. This paper highlights similarities among five popular models, latent budget analysis (LBA), latent class analysis (LCA), end-member analysis (EMA), probabilistic latent semantic analysis (PLSA), and nonnegative matrix factorization (NMF). We focus on an essential issue-identifiability-of these models and prove that the solution of LBA, EMA, LCA, PLSA is unique if and only if the solution of NMF is unique. We also provide a brief review for algorithms of these models. We illustrate the models with a time budget dataset from social science, and end the paper with a discussion of closely related models such as archetypal analysis.
Authors: Ernest Fokou\'e
Abstract: Nature rarely reveals her secrets bluntly, yet in the Fibonacci sequence she grants us a glimpse of her quiet architecture of growth, harmony, and recursive stability \citep{Koshy2001Fibonacci, Livio2002GoldenRatio}. From spiral galaxies to the unfolding of leaves, this humble sequence reflects a universal grammar of balance. In this work, we introduce \emph{Fibonacci Ensembles}, a mathematically principled yet philosophically inspired framework for ensemble learning that complements and extends classical aggregation schemes such as bagging, boosting, and random forests \citep{Breiman1996Bagging, Breiman2001RandomForests, Friedman2001GBM, Zhou2012Ensemble, HastieTibshiraniFriedman2009ESL}. Two intertwined formulations unfold: (1) the use of normalized Fibonacci weights -- tempered through orthogonalization and Rao--Blackwell optimization -- to achieve systematic variance reduction among base learners, and (2) a second-order recursive ensemble dynamic that mirrors the Fibonacci flow itself, enriching representational depth beyond classical boosting. The resulting methodology is at once rigorous and poetic: a reminder that learning systems flourish when guided by the same intrinsic harmonies that shape the natural world. Through controlled one-dimensional regression experiments using both random Fourier feature ensembles \citep{RahimiRecht2007RFF} and polynomial ensembles, we exhibit regimes in which Fibonacci weighting matches or improves upon uniform averaging and interacts in a principled way with orthogonal Rao--Blackwellization. These findings suggest that Fibonacci ensembles form a natural and interpretable design point within the broader theory of ensemble learning.
Authors: Ernest Fokou\'e
Abstract: Ensemble learning is traditionally justified as a variance-reduction strategy, explaining its strong performance for unstable predictors such as decision trees. This explanation, however, does not account for ensembles constructed from intrinsically stable estimators-including smoothing splines, kernel ridge regression, Gaussian process regression, and other regularized reproducing kernel Hilbert space (RKHS) methods whose variance is already tightly controlled by regularization and spectral shrinkage. This paper develops a general weighting theory for ensemble learning that moves beyond classical variance-reduction arguments. We formalize ensembles as linear operators acting on a hypothesis space and endow the space of weighting sequences with geometric and spectral constraints. Within this framework, we derive a refined bias-variance approximation decomposition showing how non-uniform, structured weights can outperform uniform averaging by reshaping approximation geometry and redistributing spectral complexity, even when variance reduction is negligible. Our main results provide conditions under which structured weighting provably dominates uniform ensembles, and show that optimal weights arise as solutions to constrained quadratic programs. Classical averaging, stacking, and recently proposed Fibonacci-based ensembles appear as special cases of this unified theory, which further accommodates geometric, sub-exponential, and heavy-tailed weighting laws. Overall, the work establishes a principled foundation for structure-driven ensemble learning, explaining why ensembles remain effective for smooth, low-variance base learners and setting the stage for distribution-adaptive and dynamically evolving weighting schemes developed in subsequent work.
Authors: Vesal Ahsani, Babak Hossein Khalaj
Abstract: In-cabin Driver Monitoring Systems (DMS) must recognize distraction- and drowsiness-related behaviors with low latency under strict constraints on compute, power, and cost. We present a single-camera in-cabin driver behavior recognition system designed for deployment on two low-cost edge platforms: Raspberry Pi 5 (CPU-only) and Google Coral Edge TPU. The proposed pipeline combines (i) a compact per-frame vision model, (ii) a confounder-aware label design to reduce visually similar false positives, and (iii) a temporal decision head that triggers alerts only when predictions are both confident and sustained. The system covers 17 behavior classes, including multiple phone-use modes, eating/drinking, smoking, reaching behind, gaze/attention shifts, passenger interaction, grooming, control-panel interaction, yawning, and eyes-closed sleep. Training and evaluation use licensed datasets spanning diverse drivers, vehicles, and lighting conditions (details in Section 6), and we further validate runtime behavior in real in-vehicle tests. The optimized deployments achieve about 16 FPS on Raspberry Pi 5 with INT8 inference (per-frame latency under 60 ms) and about 25 FPS on Coral Edge TPU, enabling real-time monitoring and stable alert generation on inexpensive hardware. Finally, we discuss how reliable in-cabin human-state perception can serve as an upstream input for human-centered vehicle intelligence, including emerging agentic vehicle concepts.
Authors: You Li, Guannan Zhao, Yuhao Ju, Yunqi He, Jie Gu, Hai Zhou
Abstract: We introduce LLA, an effective intellectual property (IP) protection scheme for generative AI models. LLA leverages the synergy between hardware and software to defend against various supply chain threats, including model theft, model corruption, and information leakage. On the software side, it embeds key bits into neurons that can trigger outliers to degrade performance and applies invariance transformations to obscure the key values. On the hardware side, it integrates a lightweight locking module into the AI accelerator while maintaining compatibility with various dataflow patterns and toolchains. An accelerator with a pre-stored secret key acts as a license to access the model services provided by the IP owner. The evaluation results show that LLA can withstand a broad range of oracle-guided key optimization attacks, while incurring a minimal computational overhead of less than 0.1% for 7,168 key bits.
Authors: Shaofei Cai, Yulei Qin, Haojia Lin, Zihan Xu, Gang Li, Yuchen Shi, Zongyi Li, Yong Mao, Siqi Cai, Xiaoyu Tan, Yitao Liang, Ke Li, Xing Sun
Abstract: Agentic reinforcement learning (RL) holds great promise for the development of autonomous agents under complex GUI tasks, but its scalability remains severely hampered by the verification of task completion. Existing task verification is treated as a passive, post-hoc process: a verifier (i.e., rule-based scoring script, reward or critic model, and LLM-as-a-Judge) analyzes the agent's entire interaction trajectory to determine if the agent succeeds. Such processing of verbose context that contains irrelevant, noisy history poses challenges to the verification protocols and therefore leads to prohibitive cost and low reliability. To overcome this bottleneck, we propose SmartSnap, a paradigm shift from this passive, post-hoc verification to proactive, in-situ self-verification by the agent itself. We introduce the Self-Verifying Agent, a new type of agent designed with dual missions: to not only complete a task but also to prove its accomplishment with curated snapshot evidences. Guided by our proposed 3C Principles (Completeness, Conciseness, and Creativity), the agent leverages its accessibility to the online environment to perform self-verification on a minimal, decisive set of snapshots. Such evidences are provided as the sole materials for a general LLM-as-a-Judge verifier to determine their validity and relevance. Experiments on mobile tasks across model families and scales demonstrate that our SmartSnap paradigm allows training LLM-driven agents in a scalable manner, bringing performance gains up to 26.08% and 16.66% respectively to 8B and 30B models. The synergizing between solution finding and evidence seeking facilitates the cultivation of efficient, self-verifying agents with competitive performance against DeepSeek V3.1 and Qwen3-235B-A22B.
Authors: Ricardo Vasquez, Diego Riofr\'io-Luzcando, Joe Carrion-Jumbo, Cesar Guevara
Abstract: Emotions are one of the important components of the human being, thus they are a valuable part of daily activities such as interaction with people, decision making and learning. For this reason, it is important to detect, recognize and understand emotions using computational systems to improve communication between people and machines, which would facilitate the ability of computers to understand the communication between humans. This study proposes the creation of a model that allows the classification of people's emotions based on their EEG signals, for which the brain-computer interface EMOTIV EPOC was used. This allowed the collection of electroencephalographic information from 50 people, all of whom were shown audiovisual resources that helped to provoke the desired mood. The information obtained was stored in a database for the generation of the model and the corresponding classification analysis. Random Forest model was created for emotion prediction (happiness, sadness and relaxation), based on the signals of any person. The results obtained were 97.21% accurate for happiness, 76% for relaxation and 76% for sadness. Finally, the model was used to generate a real-time emotion prediction algorithm; it captures the person's EEG signals, executes the generated algorithm and displays the result on the screen with the help of images representative of each emotion.
Authors: Alaa Alahmadi, Mohamed Hasan
Abstract: Machine vision models, particularly deep neural networks, are increasingly applied to physiological signal interpretation, including electrocardiography (ECG), yet they typically require large training datasets and offer limited insight into the causal features underlying their predictions. This lack of data efficiency and interpretability constrains their clinical reliability and alignment with human reasoning. Here, we show that a perception-informed pseudo-colouring technique, previously demonstrated to enhance human ECG interpretation, can improve both explainability and few-shot learning in deep neural networks analysing complex physiological data. We focus on acquired, drug-induced long QT syndrome (LQTS) as a challenging case study characterised by heterogeneous signal morphology, variable heart rate, and scarce positive cases associated with life-threatening arrhythmias such as torsades de pointes. This setting provides a stringent test of model generalisation under extreme data scarcity. By encoding clinically salient temporal features, such as QT-interval duration, into structured colour representations, models learn discriminative and interpretable features from as few as one or five training examples. Using prototypical networks and a ResNet-18 architecture, we evaluate one-shot and few-shot learning on ECG images derived from single cardiac cycles and full 10-second rhythms. Explainability analyses show that pseudo-colouring guides attention toward clinically meaningful ECG features while suppressing irrelevant signal components. Aggregating multiple cardiac cycles further improves performance, mirroring human perceptual averaging across heartbeats. Together, these findings demonstrate that human-like perceptual encoding can bridge data efficiency, explainability, and causal reasoning in medical machine intelligence.
Authors: Xin Yu, Xiaojuan Qi, Zhengqi Li, Kai Zhang, Richard Zhang, Zhe Lin, Eli Shechtman, Tianyu Wang, Yotam Nitzan
Abstract: We introduce the Self-Evaluating Model (Self-E), a novel, from-scratch training approach for text-to-image generation that supports any-step inference. Self-E learns from data similarly to a Flow Matching model, while simultaneously employing a novel self-evaluation mechanism: it evaluates its own generated samples using its current score estimates, effectively serving as a dynamic self-teacher. Unlike traditional diffusion or flow models, it does not rely solely on local supervision, which typically necessitates many inference steps. Unlike distillation-based approaches, it does not require a pretrained teacher. This combination of instantaneous local learning and self-driven global matching bridges the gap between the two paradigms, enabling the training of a high-quality text-to-image model from scratch that excels even at very low step counts. Extensive experiments on large-scale text-to-image benchmarks show that Self-E not only excels in few-step generation, but is also competitive with state-of-the-art Flow Matching models at 50 steps. We further find that its performance improves monotonically as inference steps increase, enabling both ultra-fast few-step generation and high-quality long-trajectory sampling within a single unified model. To our knowledge, Self-E is the first from-scratch, any-step text-to-image model, offering a unified framework for efficient and scalable generation.
Authors: Mohammad Zakaria Haider, Amit Kumar Podder, Prabin Mali, Aranya Chakrabortty, Sumit Paudyal, Mohammad Ashiqur Rahman
Abstract: The rapid deployment of electric vehicle charging stations (EVCS) within distribution networks necessitates intelligent and adaptive control to maintain the grid's resilience and reliability. In this work, we propose PHANTOM, a physics-aware adversarial network that is trained and optimized through a multi-agent reinforcement learning model. PHANTOM integrates a physics-informed neural network (PINN) enabled by federated learning (FL) that functions as a digital twin of EVCS-integrated systems, ensuring physically consistent modeling of operational dynamics and constraints. Building on this digital twin, we construct a multi-agent RL environment that utilizes deep Q-networks (DQN) and soft actor-critic (SAC) methods to derive adversarial false data injection (FDI) strategies capable of bypassing conventional detection mechanisms. To examine the broader grid-level consequences, a transmission and distribution (T and D) dual simulation platform is developed, allowing us to capture cascading interactions between EVCS disturbances at the distribution level and the operations of the bulk transmission system. Results demonstrate how learned attack policies disrupt load balancing and induce voltage instabilities that propagate across T and D boundaries. These findings highlight the critical need for physics-aware cybersecurity to ensure the resilience of large-scale vehicle-grid integration.
Authors: Behzad Parvaresh, Rahmat K. Adesunkanmi, Adel Alaeddini
Abstract: Continuous fiber-reinforced composite manufactured by additive manufacturing (CFRC-AM) offers opportunities for printing lightweight materials with high specific strength. However, their performance is sensitive to the interaction of process and material parameters, making exhaustive experimental testing impractical. In this study, we introduce a data-efficient, multi-input, multi-target learning approach that integrates Latin Hypercube Sampling (LHS)-guided experimentation with a squeeze-and-excitation wide and deep neural network (SE-WDNN) to jointly predict multiple mechanical and manufacturing properties of CFRC-AMs based on different manufacturing parameters. We printed and tested 155 specimens selected from a design space of 4,320 combinations using a Markforged Mark Two 3D printer. The processed data formed the input-output set for our proposed model. We compared the results with those from commonly used machine learning models, including feedforward neural networks, Kolmogorov-Arnold networks, XGBoost, CatBoost, and random forests. Our model achieved the lowest overall test error (MAPE = 12.33%) and showed statistically significant improvements over the baseline wide and deep neural network for several target variables (paired t-tests, p <= 0.05). SHapley Additive exPlanations (SHAP) analysis revealed that reinforcement strategy was the major influence on mechanical performance. Overall, this study demonstrates that the integration of LHS and SE-WDNN enables interpretable and sample-efficient multi-target predictions, guiding parameter selection in CFRC-AM with a balance between mechanical behavior and manufacturing metrics.
Authors: Zihan Lin, QiZhi He
Abstract: We present a latent diffusion-based differentiable inversion method (LD-DIM) for PDE-constrained inverse problems involving high-dimensional spatially distributed coefficients. LD-DIM couples a pretrained latent diffusion prior with an end-to-end differentiable numerical solver to reconstruct unknown heterogeneous parameter fields in a low-dimensional nonlinear manifold, improving numerical conditioning and enabling stable gradient-based optimization under sparse observations. The proposed framework integrates a latent diffusion model (LDM), trained in a compact latent space, with a differentiable finite-volume discretization of the forward PDE. Sensitivities are propagated through the discretization using adjoint-based gradients combined with reverse-mode automatic differentiation. Inversion is performed directly in latent space, which implicitly suppresses ill-conditioned degrees of freedom while preserving dominant structural modes, including sharp material interfaces. The effectiveness of LD-DIM is demonstrated using a representative inverse problem for flow in porous media, where heterogeneous conductivity fields are reconstructed from spatially sparse hydraulic head measurements. Numerical experiments assess convergence behavior and reconstruction quality for both Gaussian random fields and bimaterial coefficient distributions. The results show that LD-DIM achieves consistently improved numerical stability and reconstruction accuracy of both parameter fields and corresponding PDE solutions compared with physics-informed neural networks (PINNs) and physics-embedded variational autoencoder (VAE) baselines, while maintaining sharp discontinuities and reducing sensitivity to initialization.
Authors: Amil Khan, Matheus Palhares Viana, Suraj Mishra, B. S. Manjunath
Abstract: Label-free 3D brightfield microscopy offers a fast and noninvasive way to visualize cellular morphology, yet robust volumetric segmentation still typically depends on fluorescence or heavy post-processing. We address this gap by introducing Bright-4B, a 4 billion parameter foundation model that learns on the unit hypersphere to segment subcellular structures directly from 3D brightfield volumes. Bright-4B combines a hardware-aligned Native Sparse Attention mechanism (capturing local, coarse, and selected global context), depth-width residual HyperConnections that stabilize representation flow, and a soft Mixture-of-Experts for adaptive capacity. A plug-and-play anisotropic patch embed further respects confocal point-spread and axial thinning, enabling geometry-faithful 3D tokenization. The resulting model produces morphology-accurate segmentations of nuclei, mitochondria, and other organelles from brightfield stacks alone--without fluorescence, auxiliary channels, or handcrafted post-processing. Across multiple confocal datasets, Bright-4B preserves fine structural detail across depth and cell types, outperforming contemporary CNN and Transformer baselines. All code, pretrained weights, and models for downstream finetuning will be released to advance large-scale, label-free 3D cell mapping.
Authors: Y. Sungtaek Ju
Abstract: Reconstructing time-resolved flow fields from temporally sparse velocimetry measurements is critical for characterizing many complex thermal-fluid systems. We introduce a machine learning framework for uncertainty-aware flow reconstruction using sparse variational Gaussian processes in the Kolmogorov-Arnold network topology (SVGP-KAN). This approach extends the classical foundations of Linear Stochastic Estimation (LSE) and Spectral Analysis Modal Methods (SAMM) while enabling principled epistemic uncertainty quantification. We perform a systematic comparison of our framework with the classical reconstruction methods as well as Kalman filtering. Using synthetic data from pulsed impingement jet flows, we assess performance across fractional PIV sampling rates ranging from 0.5% to 10%. Evaluation metrics include reconstruction error, generalization gap, structure preservation, and uncertainty calibration. Our SVGP-KAN methods achieve reconstruction accuracy comparable to established methods, while also providing well-calibrated uncertainty estimates that reliably indicate when and where predictions degrade. The results demonstrate a robust, data-driven framework for flow field reconstruction with meaningful uncertainty quantification and offer practical guidance for experimental design in periodic flows.
Authors: Zining Wang, Jian Gao, Weimin Fu, Xiaolong Guo, Xuan Zhang
Abstract: Analog circuit design remains a knowledge- and experience-intensive process that relies heavily on human intuition for topology generation and device parameter tuning. Existing LLM-based approaches typically depend on prompt-driven netlist generation or predefined topology templates, limiting their ability to satisfy complex specification requirements. We propose AnalogSAGE, an open-source self-evolving multi-agent framework that coordinates three-stage agent explorations through four stratified memory layers, enabling iterative refinement with simulation-grounded feedback. To support reproducibility and generality, we release the source code. Our benchmark spans ten specification-driven operational amplifier design problems of varying difficulty, enabling quantitative and cross-task comparison under identical conditions. Evaluated under the open-source SKY130 PDK with ngspice, AnalogSAGE achieves a 10$\times$ overall pass rate, a 48$\times$ Pass@1, and a 4$\times$ reduction in parameter search space compared with existing frameworks, demonstrating that stratified memory and grounded reasoning substantially enhance the reliability and autonomy of analog design automation in practice.
Authors: Cattalyya Nuengsigkapian
Abstract: Retrieval-Augmented Generation (RAG) in open-domain settings faces significant challenges regarding irrelevant information in retrieved documents and the alignment of generated answers with user intent. We present HiFi-RAG (Hierarchical Filtering RAG), the winning closed-source system in the Text-to-Text static evaluation of the MMU-RAGent NeurIPS 2025 Competition. Our approach moves beyond standard embedding-based retrieval via a multi-stage pipeline. We leverage the speed and cost-efficiency of Gemini 2.5 Flash (4-6x cheaper than Pro) for query formulation, hierarchical content filtering, and citation attribution, while reserving the reasoning capabilities of Gemini 2.5 Pro for final answer generation. On the MMU-RAGent validation set, our system outperformed the baseline, improving ROUGE-L to 0.274 (+19.6%) and DeBERTaScore to 0.677 (+6.2%). On Test2025, our custom dataset evaluating questions that require post-cutoff knowledge (post January 2025), HiFi-RAG outperforms the parametric baseline by 57.4% in ROUGE-L and 14.9% in DeBERTaScore.
Authors: Naman Aggarwal, Siddhartha R. Dalal, Vishal Misra
Abstract: Transformers empirically perform precise probabilistic reasoning in carefully constructed ``Bayesian wind tunnels'' and in large-scale language models, yet the mechanisms by which gradient-based learning creates the required internal geometry remain opaque. We provide a complete first-order analysis of how cross-entropy training reshapes attention scores and value vectors in a transformer attention head. Our core result is an \emph{advantage-based routing law} for attention scores, \[ \frac{\partial L}{\partial s_{ij}} = \alpha_{ij}\bigl(b_{ij}-\mathbb{E}_{\alpha_i}[b]\bigr), \qquad b_{ij} := u_i^\top v_j, \] coupled with a \emph{responsibility-weighted update} for values, \[ \Delta v_j = -\eta\sum_i \alpha_{ij} u_i, \] where $u_i$ is the upstream gradient at position $i$ and $\alpha_{ij}$ are attention weights. These equations induce a positive feedback loop in which routing and content specialize together: queries route more strongly to values that are above-average for their error signal, and those values are pulled toward the queries that use them. We show that this coupled specialization behaves like a two-timescale EM procedure: attention weights implement an E-step (soft responsibilities), while values implement an M-step (responsibility-weighted prototype updates), with queries and keys adjusting the hypothesis frame. Through controlled simulations, including a sticky Markov-chain task where we compare a closed-form EM-style update to standard SGD, we demonstrate that the same gradient dynamics that minimize cross-entropy also sculpt the low-dimensional manifolds identified in our companion work as implementing Bayesian inference. This yields a unified picture in which optimization (gradient flow) gives rise to geometry (Bayesian manifolds), which in turn supports function (in-context probabilistic reasoning).
Authors: Zihan Weng, Chanlin Yi, Pouya Bashivan, Jing Lu, Fali Li, Dezhong Yao, Jingming Hou, Yangsong Zhang, Peng Xu
Abstract: Decoding fine-grained movement from non-invasive surface Electromyography (sEMG) is a challenge for prosthetic control due to signal non-stationarity and low signal-to-noise ratios. Generic self-supervised learning (SSL) frameworks often yield suboptimal results on sEMG as they attempt to reconstruct noisy raw signals and lack the inductive bias to model the cylindrical topology of electrode arrays. To overcome these limitations, we introduce SPECTRE, a domain-specific SSL framework. SPECTRE features two primary contributions: a physiologically-grounded pre-training task and a novel positional encoding. The pre-training involves masked prediction of discrete pseudo-labels from clustered Short-Time Fourier Transform (STFT) representations, compelling the model to learn robust, physiologically relevant frequency patterns. Additionally, our Cylindrical Rotary Position Embedding (CyRoPE) factorizes embeddings along linear temporal and annular spatial dimensions, explicitly modeling the forearm sensor topology to capture muscle synergies. Evaluations on multiple datasets, including challenging data from individuals with amputation, demonstrate that SPECTRE establishes a new state-of-the-art for movement decoding, significantly outperforming both supervised baselines and generic SSL approaches. Ablation studies validate the critical roles of both spectral pre-training and CyRoPE. SPECTRE provides a robust foundation for practical myoelectric interfaces capable of handling real-world sEMG complexities.
Authors: Zhenqian Chen, Baoquan Zhong, Xiang Li, Qing Dai, Xinkui Zhao, Miao Ye, Ren Cheng, Lufei Zhang, Jianwei Yin
Abstract: RL post-training for LLMs has been widely scaled to enhance reasoning and tool-using capabilities. However, RL post-training interleaves training and inference workloads, exposing the system to faults from both sides. Existing fault tolerance frameworks for LLMs target either training or inference, leaving the optimization potential in the asynchronous execution unexplored for RL. Our key insight is role-based fault isolation so the failure in one machine does not affect the others. We treat trainer, rollout, and other management roles in RL training as distinct distributed sub-tasks. Instead of restarting the entire RL task in ByteRobust, we recover only the failed role and reconnect it to living ones, thereby eliminating the full-restart overhead including rollout replay and initialization delay. We present RobustRL, the first comprehensive robust system to handle GPU machine errors for RL post-training Effective Training Time Ratio improvement. (1) \textit{Detect}. We implement role-aware monitoring to distinguish actual failures from role-specific behaviors to avoid the false positive and delayed detection. (2) \textit{Restart}. For trainers, we implement a non-disruptive recovery where rollouts persist state and continue trajectory generation, while the trainer is rapidly restored via rollout warm standbys. For rollout, we perform isolated machine replacement without interrupting the RL task. (3) \textit{Reconnect}. We replace static collective communication with dynamic, UCX-based (Unified Communication X) point-to-point communication, enabling immediate weight synchronization between recovered roles. In an RL training task on a 256-GPU cluster with Qwen3-8B-Math workload under 10\% failure injection frequency, RobustRL can achieve an ETTR of over 80\% compared with the 60\% in ByteRobust and achieves 8.4\%-17.4\% faster in end-to-end training time.
Authors: Chuang-Chieh Lin, Chi-Jen Lu, Po-An Chen, Chih-Chieh Hung
Abstract: We formulate two-party policy competition as a two-player non-cooperative game, generalizing Lin et al.'s work (2021). Each party selects a real-valued policy vector as its strategy from a compact subset of Euclidean space, and a voter's utility for a policy is given by the inner product with their preference vector. To capture the uncertainty in the competition, we assume that a policy's winning probability increases monotonically with its total utility across all voters, and we formalize this via an affine isotonic function. A player's payoff is defined as the expected utility received by its supporters. In this work, we first test and validate the isotonicity hypothesis through voting simulations. Next, we prove the existence of a pure-strategy Nash equilibrium (PSNE) in both one- and multi-dimensional settings. Although we construct a counterexample demonstrating the game's non-monotonicity, our experiments show that a decentralized gradient-based algorithm typically converges rapidly to an approximate PSNE. Finally, we present a grid-based search algorithm that finds an $\epsilon$-approximate PSNE of the game in time polynomial in the input size and $1/\epsilon$.
Authors: Wei Gao, Yuheng Zhao, Tianyuan Wu, Shaopan Xiong, Weixun Wang, Dakai An, Lunxi Cao, Dilxat Muhtar, Zichen Liu, Haizhou Zhao, Ju Huang, Siran Yang, Yongbin Li, Wenbo Su, Jiamang Wang, Lin Qu, Bo Zheng, Wei Wang
Abstract: Agentic Reinforcement Learning (RL) enables Large Language Models (LLMs) to perform autonomous decision-making and long-term planning. Unlike standard LLM post-training, agentic RL workloads are highly heterogeneous, combining compute-intensive prefill phases, bandwidth-bound decoding, and stateful, CPU-heavy environment simulations. We argue that efficient agentic RL training requires disaggregated infrastructure to leverage specialized, best-fit hardware. However, naive disaggregation introduces substantial synchronization overhead and resource underutilization due to the complex dependencies between stages. We present RollArc, a distributed system designed to maximize throughput for multi-task agentic RL on disaggregated infrastructure. RollArc is built on three core principles: (1) hardware-affinity workload mapping, which routes compute-bound and bandwidth-bound tasks to bestfit GPU devices, (2) fine-grained asynchrony, which manages execution at the trajectory level to mitigate resource bubbles, and (3) statefulness-aware computation, which offloads stateless components (e.g., reward models) to serverless infrastructure for elastic scaling. Our results demonstrate that RollArc effectively improves training throughput and achieves 1.35-2.05\(\times\) end-to-end training time reduction compared to monolithic and synchronous baselines. We also evaluate RollArc by training a hundreds-of-billions-parameter MoE model for Qoder product on an Alibaba cluster with more than 3,000 GPUs, further demonstrating RollArc scalability and robustness. The code is available at https://github.com/alibaba/ROLL.
Authors: Atakan I\c{s}{\i}k, Selin Vulga I\c{s}{\i}k, Ahmet Feridun I\c{s}{\i}k, Mah\c{s}uk Taylan
Abstract: Respiratory sound classification is hindered by the limited size, high noise levels, and severe class imbalance of benchmark datasets like ICBHI 2017. While Transformer-based models offer powerful feature extraction capabilities, they are prone to overfitting and often converge to sharp minima in the loss landscape when trained on such constrained medical data. To address this, we introduce a framework that enhances the Audio Spectrogram Transformer (AST) using Sharpness-Aware Minimization (SAM). Instead of merely minimizing the training loss, our approach optimizes the geometry of the loss surface, guiding the model toward flatter minima that generalize better to unseen patients. We also implement a weighted sampling strategy to handle class imbalance effectively. Our method achieves a state-of-the-art score of 68.10% on the ICBHI 2017 dataset, outperforming existing CNN and hybrid baselines. More importantly, it reaches a sensitivity of 68.31%, a crucial improvement for reliable clinical screening. Further analysis using t-SNE and attention maps confirms that the model learns robust, discriminative features rather than memorizing background noise.
Authors: Deniz Akdemir
Abstract: Modern machine learning embeddings provide powerful compression of high-dimensional data, yet they typically destroy the geometric structure required for classical likelihood-based statistical inference. This paper develops a rigorous theory of likelihood-preserving embeddings: learned representations that can replace raw data in likelihood-based workflows -- hypothesis testing, confidence interval construction, model selection -- without altering inferential conclusions. We introduce the Likelihood-Ratio Distortion metric $\Delta_n$, which measures the maximum error in log-likelihood ratios induced by an embedding. Our main theoretical contribution is the Hinge Theorem, which establishes that controlling $\Delta_n$ is necessary and sufficient for preserving inference. Specifically, if the distortion satisfies $\Delta_n = o_p(1)$, then (i) all likelihood-ratio based tests and Bayes factors are asymptotically preserved, and (ii) surrogate maximum likelihood estimators are asymptotically equivalent to full-data MLEs. We prove an impossibility result showing that universal likelihood preservation requires essentially invertible embeddings, motivating the need for model-class-specific guarantees. We then provide a constructive framework using neural networks as approximate sufficient statistics, deriving explicit bounds connecting training loss to inferential guarantees. Experiments on Gaussian and Cauchy distributions validate the sharp phase transition predicted by exponential family theory, and applications to distributed clinical inference demonstrate practical utility.
Authors: Irched Chafaa, Giacomo Bacci, Luca Sanguinetti
Abstract: Power allocation remains a fundamental challenge in wireless communication networks, particularly under dynamic user loads and large-scale deployments. While Transformerbased models have demonstrated strong performance, their computational cost scales poorly with the number of users. In this work, we propose a novel hybrid Tree-Transformer architecture that achieves scalable per-user power allocation. Our model compresses user features via a binary tree into a global root representation, applies a Transformer encoder solely to this root, and decodes per-user uplink and downlink powers through a shared decoder. This design achieves logarithmic depth and linear total complexity, enabling efficient inference across large and variable user sets without retraining or architectural changes. We evaluate our model on the max-min fairness problem in cellfree massive MIMO systems and demonstrate that it achieves near-optimal performance while significantly reducing inference time compared to full-attention baselines.
Authors: Argha Kamal Samanta, Deepak Mewada, Monalisa Sarma, Debasis Samanta
Abstract: Clinical electroencephalography is routinely used to evaluate patients with diverse and often overlapping neurological conditions, yet interpretation remains manual, time-intensive, and variable across experts. While automated EEG analysis has been widely studied, most existing methods target isolated diagnostic problems, particularly seizure detection, and provide limited support for multi-disorder clinical screening. This study examines automated EEG-based classification across eleven clinically relevant neurological disorder categories, encompassing acute time-critical conditions, chronic neurocognitive and developmental disorders, and disorders with indirect or weak electrophysiological signatures. EEG recordings are processed using a standard longitudinal bipolar montage and represented through a multi-domain feature set capturing temporal statistics, spectral structure, signal complexity, and inter-channel relationships. Disorder-aware machine learning models are trained under severe class imbalance, with decision thresholds explicitly calibrated to prioritize diagnostic sensitivity. Evaluation on a large, heterogeneous clinical EEG dataset demonstrates that sensitivity-oriented modeling achieves recall exceeding 80% for the majority of disorder categories, with several low-prevalence conditions showing absolute recall gains of 15-30% after threshold calibration compared to default operating points. Feature importance analysis reveals physiologically plausible patterns consistent with established clinical EEG markers. These results establish realistic performance baselines for multi-disorder EEG classification and provide quantitative evidence that sensitivity-prioritized automated analysis can support scalable EEG screening and triage in real-world clinical settings.
Authors: Shravan Saranyan, Pramit Saha
Abstract: Left ventricular ejection fraction (LVEF) is a key indicator of cardiac function and plays a central role in the diagnosis and management of cardiovascular disease. Echocardiography, as a readily accessible and non-invasive imaging modality, is widely used in clinical practice to estimate LVEF. However, manual assessment of cardiac function from echocardiograms is time-consuming and subject to considerable inter-observer variability. Deep learning approaches offer a promising alternative, with the potential to achieve performance comparable to that of experienced human experts. In this study, we investigate the effectiveness of several deep learning architectures for LVEF estimation from echocardiography videos, including 3D Inception, two-stream, and CNN-RNN models. We systematically evaluate architectural modifications and fusion strategies to identify configurations that maximize prediction accuracy. Models were trained and evaluated on the EchoNet-Dynamic dataset, comprising 10,030 echocardiogram videos. Our results demonstrate that modified 3D Inception architectures achieve the best overall performance, with a root mean squared error (RMSE) of 6.79%. Across architectures, we observe a tendency toward overfitting, with smaller and simpler models generally exhibiting improved generalization. Model performance was also found to be highly sensitive to hyperparameter choices, particularly convolutional kernel sizes and normalization strategies. While this study focuses on echocardiography-based LVEF estimation, the insights gained regarding architectural design and training strategies may be applicable to a broader range of medical and non-medical video analysis tasks.
Authors: Julia Ko\'nczal, Micha{\l} Balcerek, Krzysztof Burnecki
Abstract: In recent years, the growing frequency and severity of natural disasters have increased the need for effective tools to manage catastrophe risk. Catastrophe (CAT) bonds allow the transfer of part of this risk to investors, offering an alternative to traditional reinsurance. This paper examines the role of climate variability in CAT bond pricing and evaluates the predictive performance of various machine learning models in forecasting CAT bond coupons. We combine features typically used in the literature with a new set of climate indicators, including Oceanic Ni{\~n}o Index, Arctic Oscillation, North Atlantic Oscillation, Outgoing Longwave Radiation, Pacific-North American pattern, Pacific Decadal Oscillation, Southern Oscillation Index, and sea surface temperatures. We compare the performance of linear regression with several machine learning algorithms, such as random forest, gradient boosting, extremely randomized trees, and extreme gradient boosting. Our results show that including climate-related variables improves predictive accuracy across all models, with extremely randomized trees achieving the lowest root mean squared error (RMSE). These findings suggest that large-scale climate variability has a measurable influence on CAT bond pricing and that machine learning methods can effectively capture these complex relationships.
Authors: Mert Ikinci, Luna Toma, Karin U. Loeffler, Leticia Ussem, Daniela S\"usskind, Julia M. Weller, Yousef Yeganeh, Martina C. Herwig-Carl, Shadi Albarqouni
Abstract: Accurate grading of Conjunctival Melanocytic Intraepithelial Lesions (CMIL) is essential for treatment and melanoma prediction but remains difficult due to subtle morphological cues and interrelated diagnostic criteria. We introduce INTERACT-CMIL, a multi-head deep learning framework that jointly predicts five histopathological axes; WHO4, WHO5, horizontal spread, vertical spread, and cytologic atypia, through Shared Feature Learning with Combinatorial Partial Supervision and an Inter-Dependence Loss enforcing cross-task consistency. Trained and evaluated on a newly curated, multi-center dataset of 486 expert-annotated conjunctival biopsy patches from three university hospitals, INTERACT-CMIL achieves consistent improvements over CNN and foundation-model (FM) baselines, with relative macro F1 gains up to 55.1% (WHO4) and 25.0% (vertical spread). The framework provides coherent, interpretable multi-criteria predictions aligned with expert grading, offering a reproducible computational benchmark for CMIL diagnosis and a step toward standardized digital ocular pathology.
Authors: Pere Martra
Abstract: Structured width pruning of GLU-MLP layers, guided by the Maximum Absolute Weight (MAW) criterion, reveals a systematic dichotomy in how reducing the expansion ratio affects different model capabilities. While performance on tasks relying on parametric knowledge (e.g., MMLU, GSM8K) and perplexity metrics degrades predictably, instruction-following capabilities improve substantially (+46% to +75% in IFEval for Llama-3.2-1B and 3B models), and multi-step reasoning remains robust (MUSR). This pattern challenges the prevailing assumption that pruning induces uniform degradation. We evaluated seven expansion ratio configurations using comprehensive benchmarks assessing factual knowledge, mathematical reasoning, language comprehension, instruction-following, and truthfulness. Our analysis identifies the expansion ratio as a critical architectural parameter that selectively modulates cognitive capabilities, rather than merely serving as a compression metric. We provide the first systematic characterization of this selective preservation phenomenon. Notably, we document a robust inverse correlation (r = -0.864, p = 0.012 in Llama-3B) between factual knowledge capacity (MMLU) and truthfulness metrics (TruthfulQA-MC2): as knowledge degrades, the model's ability to discriminate misconceptions improves consistently. This connects two previously distinct research areas, demonstrating that MAW-guided width pruning acts as a selective filter, reducing parametric knowledge while preserving or enhancing behavioral alignment. Additionally, we quantify context-dependent efficiency trade-offs: pruned configurations achieve up to 23% reduction in energy consumption (J/token) but incur penalties in single-request latency, whereas batch processing workloads benefit uniformly.
Authors: Salvador Rodriguez-Sanz, Monica Hernandez
Abstract: This work proposes a multimodal diffeomorphic registration method using Neural Ordinary Differential Equations (Neural ODEs). Nonrigid registration algorithms exhibit tradeoffs between their accuracy, the computational complexity of their deformation model, and its proper regularization. In addition, they also assume intensity correlation in anatomically homologous regions of interest among image pairs, limiting their applicability to the monomodal setting. Unlike learning-based models, we propose an instance-specific framework that is not subject to high scan requirements for training and does not suffer performance degradation at inference time on modalities unseen during training. Our method exploits the potential of continuous-depth networks in the Neural ODE paradigm with structural descriptors, widely adopted as modality-agnostic metric models which exploit self-similarities on parameterized neighborhood geometries. We propose three different variants that integrate image-based or feature-based structural descriptors and nonstructural image similarities computed by local mutual information. We conduct extensive evaluations on different experiments formed by scan dataset combinations and show surpassing qualitative and quantitative results compared to state-of-the-art baselines adequate for large or small deformations, and specific of multimodal registration. Lastly, we also demonstrate the underlying robustness of the proposed framework to varying levels of explicit regularization while maintaining low error, its suitability for registration at varying scales, and its efficiency with respect to other methods targeted to large-deformation registration.
Authors: Ahmed Abdullah, Sana Fatima, Haroon Mahmood
Abstract: Hope speech has been relatively underrepresented in Natural Language Processing (NLP). Current studies are largely focused on English, which has resulted in a lack of resources for low-resource languages such as Urdu. As a result, the creation of tools that facilitate positive online communication remains limited. Although transformer-based architectures have proven to be effective in detecting hate and offensive speech, little has been done to apply them to hope speech or, more generally, to test them across a variety of linguistic settings. This paper presents a multilingual framework for hope speech detection with a focus on Urdu. Using pretrained transformer models such as XLM-RoBERTa, mBERT, EuroBERT, and UrduBERT, we apply simple preprocessing and train classifiers for improved results. Evaluations on the PolyHope-M 2025 benchmark demonstrate strong performance, achieving F1-scores of 95.2% for Urdu binary classification and 65.2% for Urdu multi-class classification, with similarly competitive results in Spanish, German, and English. These results highlight the possibility of implementing existing multilingual models in low-resource environments, thus making it easier to identify hope speech and helping to build a more constructive digital discourse.
Authors: Jun Wang
Abstract: We propose a theoretical framework for continual and experiential learning in large language model agents that integrates episodic memory with reinforcement learning. The framework identifies reflection as the key mechanism that enables agents to adapt through interaction without back propagation or model fine tuning, thereby relaxing the conventional separation between training and deployment.To formalise this process, we introduce the Stateful Reflective Decision Process, which models reflective learning as a two stage read write interaction with episodic memory. Writing stores interaction outcomes and corresponds to policy evaluation, while reading retrieves relevant past cases and corresponds to policy improvement. We show that this process induces an equivalent Markov decision process over augmented state memory representations, allowing the use of classical tools from dynamic programming and reinforcement learning. We further instantiate the framework using entropy regularised policy iteration and establish convergence guarantees. As episodic memory grows and achieves sufficient coverage of the state space, the resulting policy converges to the optimal solution. This work provides a principled foundation for memory augmented and retrieval based language model agents capable of continual adaptation without parameter updates.
Authors: Md Badsha Biswas
Abstract: Infant mortality remains a significant public health concern in the United States, with birth defects identified as a leading cause. Despite ongoing efforts to understand the causes of negative pregnancy outcomes like miscarriage, stillbirths, birth defects, and premature birth, there is still a need for more comprehensive research and strategies for intervention. This paper introduces a novel approach that uses publicly available social media data, especially from platforms like Twitter, to enhance current datasets for studying negative pregnancy outcomes through observational research. The inherent challenges in utilizing social media data, including imbalance, noise, and lack of structure, necessitate robust preprocessing techniques and data augmentation strategies. By constructing a natural language processing (NLP) pipeline, we aim to automatically identify women sharing their pregnancy experiences, categorizing them based on reported outcomes. Women reporting full gestation and normal birth weight will be classified as positive cases, while those reporting negative pregnancy outcomes will be identified as negative cases. Furthermore, this study offers potential applications in assessing the causal impact of specific interventions, treatments, or prenatal exposures on maternal and fetal health outcomes. Additionally, it provides a framework for future health studies involving pregnant cohorts and comparator groups. In a broader context, our research showcases the viability of social media data as an adjunctive resource in epidemiological investigations about pregnancy outcomes.
Authors: Zheng Qiu, Chih-Yuan Chiu, Glen Chou
Abstract: We present an iterative active constraint learning (ACL) algorithm, within the learning from demonstrations (LfD) paradigm, which intelligently solicits informative demonstration trajectories for inferring an unknown constraint in the demonstrator's environment. Our approach iteratively trains a Gaussian process (GP) on the available demonstration dataset to represent the unknown constraints, uses the resulting GP posterior to query start/goal states, and generates informative demonstrations which are added to the dataset. Across simulation and hardware experiments using high-dimensional nonlinear dynamics and unknown nonlinear constraints, our method outperforms a baseline, random-sampling based method at accurately performing constraint inference from an iteratively generated set of sparse but informative demonstrations.
Authors: Kiarash Vaziri, Lucine L. Oganesian, HyeongChan Jo, Roberto M. C. Vera, Charles Y. Liu, Brian Lee, Maryam M. Shanechi
Abstract: Dynamical modeling of multisite human intracranial neural recordings is essential for developing neurotechnologies such as brain-computer interfaces (BCIs). Linear dynamical models are widely used for this purpose due to their interpretability and their suitability for BCIs. In particular, these models enable flexible real-time inference, even in the presence of missing neural samples, which often occur in wireless BCIs. However, neural activity can exhibit nonlinear structure that is not captured by linear models. Furthermore, while recurrent neural network models can capture nonlinearity, their inference does not directly address handling missing observations. To address this gap, recent work introduced DFINE, a deep learning framework that integrates neural networks with linear state-space models to capture nonlinearities while enabling flexible inference. However, DFINE was developed for intracortical recordings that measure localized neuronal populations. Here we extend DFINE to modeling of multisite human intracranial electroencephalography (iEEG) recordings. We find that DFINE significantly outperforms linear state-space models (LSSMs) in forecasting future neural activity. Furthermore, DFINE matches or exceeds the accuracy of a gated recurrent unit (GRU) model in neural forecasting, indicating that a linear dynamical backbone, when paired and jointly trained with nonlinear neural networks, can effectively describe the dynamics of iEEG signals while also enabling flexible inference. Additionally, DFINE handles missing observations more robustly than the baselines, demonstrating its flexible inference and utility for BCIs. Finally, DFINE's advantage over LSSM is more pronounced in high gamma spectral bands. Taken together, these findings highlight DFINE as a strong and flexible framework for modeling human iEEG dynamics, with potential applications in next-generation BCIs.
Authors: Risha Surana, Adrian Law, Sunwoo Kim, Rishab Sridhar, Angxiao Han, Peiyu Hong
Abstract: Clinical notes are often stored in unstructured or semi-structured formats after extraction from electronic medical record (EMR) systems, which complicates their use for secondary analysis and downstream clinical applications. Reliable identification of section boundaries is a key step toward structuring these notes, as sections such as history of present illness, medications, and discharge instructions each provide distinct clinical contexts. In this work, we evaluate rule-based baselines, domain-specific transformer models, and large language models for clinical note segmentation using a curated dataset of 1,000 notes from MIMIC-IV. Our experiments show that large API-based models achieve the best overall performance, with GPT-5-mini reaching a best average F1 of 72.4 across sentence-level and freetext segmentation. Lightweight baselines remain competitive on structured sentence-level tasks but falter on unstructured freetext. Our results provide guidance for method selection and lay the groundwork for downstream tasks such as information extraction, cohort identification, and automated summarization.
Authors: Masahiro Kato
Abstract: This study proposes an end-to-end algorithm for policy learning in causal inference. We observe data consisting of covariates, treatment assignments, and outcomes, where only the outcome corresponding to the assigned treatment is observed. The goal of policy learning is to train a policy from the observed data, where a policy is a function that recommends an optimal treatment for each individual, to maximize the policy value. In this study, we first show that maximizing the policy value is equivalent to minimizing the mean squared error for the conditional average treatment effect (CATE) under $\{-1, 1\}$ restricted regression models. Based on this finding, we modify the causal forest, an end-to-end CATE estimation algorithm, for policy learning. We refer to our algorithm as the causal-policy forest. Our algorithm has three advantages. First, it is a simple modification of an existing, widely used CATE estimation method, therefore, it helps bridge the gap between policy learning and CATE estimation in practice. Second, while existing studies typically estimate nuisance parameters for policy learning as a separate task, our algorithm trains the policy in a more end-to-end manner. Third, as in standard decision trees and random forests, we train the models efficiently, avoiding computational intractability.
Authors: Bangya Liu, Xinyu Gong, Zelin Zhao, Ziyang Song, Yulei Lu, Suhui Wu, Jun Zhang, Suman Banerjee, Hao Zhang
Abstract: Human-object interaction (HOI) video generation has garnered increasing attention due to its promising applications in digital humans, e-commerce, advertising, and robotics imitation learning. However, existing methods face two critical limitations: (1) a lack of effective mechanisms to inject multi-view information of the object into the model, leading to poor cross-view consistency, and (2) heavy reliance on fine-grained hand mesh annotations for modeling interaction occlusions. To address these challenges, we introduce ByteLoom, a Diffusion Transformer (DiT)-based framework that generates realistic HOI videos with geometrically consistent object illustration, using simplified human conditioning and 3D object inputs. We first propose an RCM-cache mechanism that leverages Relative Coordinate Maps (RCM) as a universal representation to maintain object's geometry consistency and precisely control 6-DoF object transformations in the meantime. To compensate HOI dataset scarcity and leverage existing datasets, we further design a training curriculum that enhances model capabilities in a progressive style and relaxes the demand of hand mesh. Extensive experiments demonstrate that our method faithfully preserves human identity and the object's multi-view geometry, while maintaining smooth motion and object manipulation.
Authors: Soham Padia, Dhananjay Vaidya, Ramchandra Mangrulkar
Abstract: Securing blockchain-enabled IoT networks against sophisticated adversarial attacks remains a critical challenge. This paper presents a trust-based delegated consensus framework integrating Fully Homomorphic Encryption (FHE) with Attribute-Based Access Control (ABAC) for privacy-preserving policy evaluation, combined with learning-based defense mechanisms. We systematically compare three reinforcement learning approaches -- tabular Q-learning (RL), Deep RL with Dueling Double DQN (DRL), and Multi-Agent RL (MARL) -- against five distinct attack families: Naive Malicious Attack (NMA), Collusive Rumor Attack (CRA), Adaptive Adversarial Attack (AAA), Byzantine Fault Injection (BFI), and Time-Delayed Poisoning (TDP). Experimental results on a 16-node simulated IoT network reveal significant performance variations: MARL achieves superior detection under collusive attacks (F1=0.85 vs. DRL's 0.68 and RL's 0.50), while DRL and MARL both attain perfect detection (F1=1.00) against adaptive attacks where RL fails (F1=0.50). All agents successfully defend against Byzantine attacks (F1=1.00). Most critically, the Time-Delayed Poisoning attack proves catastrophic for all agents, with F1 scores dropping to 0.11-0.16 after sleeper activation, demonstrating the severe threat posed by trust-building adversaries. Our findings indicate that coordinated multi-agent learning provides measurable advantages for defending against sophisticated trust manipulation attacks in blockchain IoT environments.
Authors: Maksim Kryzhanovskiy, Svetlana Glazyrina, Roman Ischenko, Konstantin Vorontsov
Abstract: Modern AI systems often comprise multiple learnable components that can be naturally organized as graphs. A central challenge is the end-to-end training of such systems without restrictive architectural or training assumptions. Such tasks fit the theory and approaches of the collaborative Multi-Agent Reinforcement Learning (MARL) field. We introduce Reinforcement Networks, a general framework for MARL that organizes agents as vertices in a directed acyclic graph (DAG). This structure extends hierarchical RL to arbitrary DAGs, enabling flexible credit assignment and scalable coordination while avoiding strict topologies, fully centralized training, and other limitations of current approaches. We formalize training and inference methods for the Reinforcement Networks framework and connect it to the LevelEnv concept to support reproducible construction, training, and evaluation. We demonstrate the effectiveness of our approach on several collaborative MARL setups by developing several Reinforcement Networks models that achieve improved performance over standard MARL baselines. Beyond empirical gains, Reinforcement Networks unify hierarchical, modular, and graph-structured views of MARL, opening a principled path toward designing and training complex multi-agent systems. We conclude with theoretical and practical directions - richer graph morphologies, compositional curricula, and graph-aware exploration. That positions Reinforcement Networks as a foundation for a new line of research in scalable, structured MARL.
Authors: Si-Yu Xiao, Xin-Di Zhao, Xiang-Zhan Wang, Tian-Hao Mao, Ying-Kai Liao, Xing-Yu Liao, Yu-Qiao Chen, Jun-Jie Wang, Shuang Liu, Tu-Pei Chen, Yang Liu
Abstract: Accurate downhole positioning is critical in oil and gas operations but is often compromised by signal degradation in traditional surface-based Casing Collar Locator (CCL) monitoring. To address this, we present an in-situ, real-time collar recognition system using embedded neural network. We introduce lightweight "Collar Recognition Nets" (CRNs) optimized for resource-constrained ARM Cortex-M7 microprocessors. By leveraging temporal and depthwise separable convolutions, our most compact model reduces computational complexity to just 8,208 MACs while maintaining an F1 score of 0.972. Hardware validation confirms an average inference latency of 343.2 {\mu}s, demonstrating that robust, autonomous signal processing is feasible within the severe power and space limitations of downhole instrumentation.
Authors: Zhaosong Lu, Sanyou Mei
Abstract: In this paper we study a nonconvex-strongly-concave constrained minimax problem. Specifically, we propose a first-order augmented Lagrangian method for solving it, whose subproblems are nonconvex-strongly-concave unconstrained minimax problems and suitably solved by a first-order method developed in this paper that leverages the strong concavity structure. Under suitable assumptions, the proposed method achieves an \emph{operation complexity} of $O(\varepsilon^{-3.5}\log\varepsilon^{-1})$, measured in terms of its fundamental operations, for finding an $\varepsilon$-KKT solution of the constrained minimax problem, which improves the previous best-known operation complexity by a factor of $\varepsilon^{-0.5}$.
Authors: Ling Xin, Mojtaba Nayyeri, Zahra Makki Nayeri, Steffen Staab
Abstract: Structural knowledge graph foundation models aim to generalize reasoning to completely new graphs with unseen entities and relations. A key limitation of existing approaches like Ultra is their reliance on a single relational transformation (e.g., element-wise multiplication) in message passing, which can constrain expressiveness and fail to capture diverse relational and structural patterns exhibited on diverse graphs. In this paper, we propose Gamma, a novel foundation model that introduces multi-head geometric attention to knowledge graph reasoning. Gamma replaces the single relational transformation with multiple parallel ones, including real, complex, split-complex, and dual number based transformations, each designed to model different relational structures. A relational conditioned attention fusion mechanism then adaptively fuses them at link level via a lightweight gating with entropy regularization, allowing the model to robustly emphasize the most appropriate relational bias for each triple pattern. We present a full formalization of these algebraic message functions and discuss how their combination increases expressiveness beyond any single space. Comprehensive experiments on 56 diverse knowledge graphs demonstrate that Gamma consistently outperforms Ultra in zero-shot inductive link prediction, with a 5.5% improvement in mean reciprocal rank on the inductive benchmarks and a 4.4% improvement across all benchmarks, highlighting benefits from complementary geometric representations.
Authors: Mathieu Lauri\`ere, Mehdi Talbi
Abstract: This paper presents a novel deep learning framework for solving multiple optimal stopping problems in high dimensions. While deep learning has recently shown promise for single stopping problems, the multiple exercise case involves complex recursive dependencies that remain challenging. We address this by combining the Dynamic Programming Principle with neural network approximation of the value function. Unlike policy-search methods, our algorithm explicitly learns the value surface. We first consider the discrete-time problem and analyze neural network training error. We then turn to continuous problems and analyze the additional error due to the discretization of the underlying stochastic processes. Numerical experiments on high-dimensional American basket options and nonlinear utility maximization demonstrate that our method provides an efficient and scalable method for the multiple optimal stopping problem.
Authors: Siyi Wang, Zifan Wang, Karl H. Johansson
Abstract: In safety-critical decision-making, the environment may evolve over time, and the learner adjusts its risk level accordingly. This work investigates risk-averse online optimization in dynamic environments with varying risk levels, employing Conditional Value-at-Risk (CVaR) as the risk measure. To capture the dynamics of the environment and risk levels, we employ the function variation metric and introduce a novel risk-level variation metric. Two information settings are considered: a first-order scenario, where the learner observes both function values and their gradients; and a zeroth-order scenario, where only function evaluations are available. For both cases, we develop risk-averse learning algorithms with a limited sampling budget and analyze their dynamic regret bounds in terms of function variation, risk-level variation, and the total number of samples. The regret analysis demonstrates the adaptability of the algorithms in non-stationary and risk-sensitive settings. Finally, numerical experiments are presented to demonstrate the efficacy of the methods.
Authors: Niels Bracher, Lars K\"uhmichel, Desi R. Ivanova, Xavier Intes, Paul-Christian B\"urkner, Stefan T. Radev
Abstract: We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain. To this end, we introduce JADAI, a framework that jointly amortizes Bayesian adaptive design and inference by training a policy, a history network, and an inference network end-to-end. The networks minimize a generic loss that aggregates incremental reductions in posterior error along experimental sequences. Inference networks are instantiated with diffusion-based posterior estimators that can approximate high-dimensional and multimodal posteriors at every experimental step. Across standard adaptive design benchmarks, JADAI achieves superior or competitive performance.
Authors: Kerem Zaman, Shashank Srivastava
Abstract: Recent work, using the Biasing Features metric, labels a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction. We argue this metric confuses unfaithfulness with incompleteness, the lossy compression needed to turn distributed transformer computation into a linear natural language narrative. On multi-hop reasoning tasks with Llama-3 and Gemma-3, many CoTs flagged as unfaithful by Biasing Features are judged faithful by other metrics, exceeding 50% in some models. With a new faithful@k metric, we show that larger inference-time token budgets greatly increase hint verbalization (up to 90% in some settings), suggesting much apparent unfaithfulness is due to tight token limits. Using Causal Mediation Analysis, we further show that even non-verbalized hints can causally mediate prediction changes through the CoT. We therefore caution against relying solely on hint-based evaluations and advocate a broader interpretability toolkit, including causal mediation and corruption-based metrics.
Authors: Fady Rezk, Yuangang Pan, Chuan-Sheng Foo, Xun Xu, Nancy Chen, Henry Gouk, Timothy Hospedales
Abstract: Personalized alignment from preference data has focused primarily on improving reward model (RM) accuracy, with the implicit assumption that better preference ranking translates to better personalized behavior. However, in deployment, computational constraints necessitate inference-time adaptation via reward-guided decoding (RGD) rather than per-user policy fine-tuning. This creates a critical but overlooked requirement: reward models must not only rank preferences accurately but also effectively guide token-level generation decisions. We demonstrate that standard RM accuracy fails catastrophically as a selection criterion for deployment-ready personalized alignment. Through systematic evaluation across three datasets, we introduce policy accuracy, a metric quantifying whether RGD scoring functions correctly discriminate between preferred and dispreferred responses. We show that RM accuracy correlates only weakly with this policy-level discrimination ability (Kendall's tau = 0.08--0.31). More critically, we introduce Pref-LaMP, the first personalized alignment benchmark with ground-truth user completions, enabling direct behavioral evaluation without circular reward-based metrics. On Pref-LaMP, we expose a complete decoupling between discrimination and generation: methods with 20-point RM accuracy differences produce almost identical output quality, and even methods achieving high discrimination fail to generate behaviorally aligned responses. Finally, simple in-context learning (ICL) dominates all reward-guided methods for models > 3B parameters, achieving 3-5 point ROUGE-1 gains over the best reward method at 7B scale. These findings show that the field optimizes proxy metrics that fail to predict deployment performance and do not translate preferences into real behavioral adaptation under deployment constraints.
Authors: Krishna Harsha Kovelakuntla Huthasana, Alireza Olama, Andreas Lundell
Abstract: Federated Learning (FL) is a distributed machine learning setting that requires multiple clients to collaborate on training a model while maintaining data privacy. The unaddressed inherent sparsity in data and models often results in overly dense models and poor generalizability under data and client participation heterogeneity. We propose FL with an L0 constraint on the density of non-zero parameters, achieved through a reparameterization using probabilistic gates and their continuous relaxation: originally proposed for sparsity in centralized machine learning. We show that the objective for L0 constrained stochastic minimization naturally arises from an entropy maximization problem of the stochastic gates and propose an algorithm based on federated stochastic gradient descent for distributed learning. We demonstrate that the target density (rho) of parameters can be achieved in FL, under data and client participation heterogeneity, with minimal loss in statistical performance for linear and non-linear models: Linear regression (LR), Logistic regression (LG), Softmax multi-class classification (MC), Multi-label classification with logistic units (MLC), Convolution Neural Network (CNN) for multi-class classification (MC). We compare the results with a magnitude pruning-based thresholding algorithm for sparsity in FL. Experiments on synthetic data with target density down to rho = 0.05 and publicly available RCV1, MNIST, and EMNIST datasets with target density down to rho = 0.005 demonstrate that our approach is communication-efficient and consistently better in statistical performance.
Authors: Jianping Mei, Michael Moses, Jan Waelty, Yucheng Yang
Abstract: We study how deep learning can improve valuation in the art market by incorporating the visual content of artworks into predictive models. Using a large repeated-sales dataset from major auction houses, we benchmark classical hedonic regressions and tree-based methods against modern deep architectures, including multi-modal models that fuse tabular and image data. We find that while artist identity and prior transaction history dominate overall predictive power, visual embeddings provide a distinct and economically meaningful contribution for fresh-to-market works where historical anchors are absent. Interpretability analyses using Grad-CAM and embedding visualizations show that models attend to compositional and stylistic cues. Our findings demonstrate that multi-modal deep learning delivers significant value precisely when valuation is hardest, namely first-time sales, and thus offers new insights for both academic research and practice in art market valuation.
Authors: Tim C. Pearce, Ahmed Ibrahim
Abstract: The discovery of novel odorant molecules is key for the fragrance and flavor industries, yet efficiently navigating the vast chemical space to identify structures with desirable olfactory properties remains a significant challenge. Generative artificial intelligence offers a promising approach for \textit{de novo} molecular design but typically requires large sets of molecules to learn from. To address this problem, we present a framework combining a variational autoencoder (VAE) with a quantitative structure-activity relationship (QSAR) model to generate novel odorants from limited training sets of odor molecules. The self-supervised learning capabilities of the VAE allow it to learn SMILES grammar from ChemBL database, while its training objective is augmented with a loss term derived from an external QSAR model to structure the latent representation according to odor probability. While the VAE demonstrated high internal consistency in learning the QSAR supervision signal, validation against an external, unseen ground truth dataset (Unique Good Scents) confirms the model generates syntactically valid structures (100\% validity achieved via rejection sampling) and 94.8\% unique structures. The latent space is effectively structured by odor likelihood, evidenced by a Fr\'echet ChemNet Distance (FCD) of $\approx$ 6.96 between generated molecules and known odorants, compared to $\approx$ 21.6 for the ChemBL baseline. Structural analysis via Bemis-Murcko scaffolds reveals that 74.4\% of candidates possess novel core frameworks distinct from the training data, indicating the model performs extensive chemical space exploration beyond simple derivatization of known odorants. Generated candidates display physicochemical properties ....
Authors: Armin Berger, Manuela Bergau, Helen Schneider, Saad Ahmad, Tom Anglim Lagones, Gianluca Brugnara, Martha Foltyn-Dumitru, Kai Schlamp, Philipp Vollmuth, Rafet Sifa
Abstract: Recent Reinforcement Learning (RL) advances for Large Language Models (LLMs) have improved reasoning tasks, yet their resource-constrained application to medical imaging remains underexplored. We introduce ChexReason, a vision-language model trained via R1-style methodology (SFT followed by GRPO) using only 2,000 SFT samples, 1,000 RL samples, and a single A100 GPU. Evaluations on CheXpert and NIH benchmarks reveal a fundamental tension: GRPO recovers in-distribution performance (23% improvement on CheXpert, macro-F1 = 0.346) but degrades cross-dataset transferability (19% drop on NIH). This mirrors high-resource models like NV-Reason-CXR-3B, suggesting the issue stems from the RL paradigm rather than scale. We identify a generalization paradox where the SFT checkpoint uniquely improves on NIH before optimization, indicating teacher-guided reasoning captures more institution-agnostic features. Furthermore, cross-model comparisons show structured reasoning scaffolds benefit general-purpose VLMs but offer minimal gain for medically pre-trained models. Consequently, curated supervised fine-tuning may outperform aggressive RL for clinical deployment requiring robustness across diverse populations.
Authors: Yu Li, Tian Lan, Zhengling Qi
Abstract: Direct Preference Optimization (DPO) and its variants have become standard for aligning Large Language Models due to their simplicity and offline stability. However, we identify two fundamental limitations. First, the optimal policy depends on arbitrary modeling choices (scalarization function, reference policy), yielding behavior reflecting parameterization artifacts rather than true preferences. Second, treating response generation in isolation fails to leverage comparative information in pairwise data, leaving the model's capacity for intrinsic self-reflection untapped. To address it, we propose Intrinsic Self-reflective Preference Optimization (\q), deriving a globally optimal policy conditioning on both context and alternative responses. We prove this formulation superior to DPO/RLHF while guaranteeing invariance to scalarization and reference choices. \q~serves as a plug-and-play enhancement without architectural changes or inference overhead. Experiments demonstrate consistent improvements in win rates and length-controlled metrics, validating that unlocking self-reflection yields more robust, human-aligned LLMs.
Authors: Armstrong Foundjem, Lionel Nganyewou Tidjon, Leuson Da Silva, Foutse Khomh
Abstract: Machine learning (ML) underpins foundation models in finance, healthcare, and critical infrastructure, making them targets for data poisoning, model extraction, prompt injection, automated jailbreaking, and preference-guided black-box attacks that exploit model comparisons. Larger models can be more vulnerable to introspection-driven jailbreaks and cross-modal manipulation. Traditional cybersecurity lacks ML-specific threat modeling for foundation, multimodal, and RAG systems. Objective: Characterize ML security risks by identifying dominant TTPs, vulnerabilities, and targeted lifecycle stages. Methods: We extract 93 threats from MITRE ATLAS (26), AI Incident Database (12), and literature (55), and analyze 854 GitHub/Python repositories. A multi-agent RAG system (ChatGPT-4o, temp 0.4) mines 300+ articles to build an ontology-driven threat graph linking TTPs, vulnerabilities, and stages. Results: We identify unreported threats including commercial LLM API model stealing, parameter memorization leakage, and preference-guided text-only jailbreaks. Dominant TTPs include MASTERKEY-style jailbreaking, federated poisoning, diffusion backdoors, and preference optimization leakage, mainly impacting pre-training and inference. Graph analysis reveals dense vulnerability clusters in libraries with poor patch propagation. Conclusion: Adaptive, ML-specific security frameworks, combining dependency hygiene, threat intelligence, and monitoring, are essential to mitigate supply-chain and inference risks across the ML lifecycle.
Authors: Yuan-Sen Ting
Abstract: Attenuation bias -- the systematic underestimation of regression coefficients due to measurement errors in input variables -- affects astronomical data-driven models. For linear regression, this problem was solved by treating the true input values as latent variables to be estimated alongside model parameters. In this paper, we show that neural networks suffer from the same attenuation bias and that the latent variable solution generalizes directly to neural networks. We introduce LatentNN, a method that jointly optimizes network parameters and latent input values by maximizing the joint likelihood of observing both inputs and outputs. We demonstrate the correction on one-dimensional regression, multivariate inputs with correlated features, and stellar spectroscopy applications. LatentNN reduces attenuation bias across a range of signal-to-noise ratios where standard neural networks show large bias. This provides a framework for improved neural network inference in the low signal-to-noise regime characteristic of astronomical data. This bias correction is most effective when measurement errors are less than roughly half the intrinsic data range; in the regime of very low signal-to-noise and few informative features. Code is available at https://github.com/tingyuansen/LatentNN.
Authors: Wendyam Eric Lionel Ilboudo, Saori C Tanaka
Abstract: Decision paralysis, i.e. hesitation, freezing, or failure to act despite full knowledge and motivation, poses a challenge for choice models that assume options are already specified and readily comparable. Drawing on qualitative reports in autism research that are especially salient, we propose a computational account in which paralysis arises from convergence failure in a hierarchical decision process. We separate intent selection (what to pursue) from affordance selection (how to pursue the goal) and formalize commitment as inference under a mixture of reverse- and forward-Kullback-Leibler (KL) objectives. Reverse KL is mode-seeking and promotes rapid commitment, whereas forward KL is mode-covering and preserves multiple plausible goals or actions. In static and dynamic (drift-diffusion) models, forward-KL-biased inference yields slow, heavy-tailed response times and two distinct failure modes, intent saturation and affordance saturation, when values are similar. Simulations in multi-option tasks reproduce key features of decision inertia and shutdown, treating autism as an extreme regime of a general, inference-based, decision-making continuum.
Authors: Yifan Zhang, Giridhar Ganapavarapu, Srideepika Jayaraman, Bhavna Agrawal, Dhaval Patel, Achille Fokoue
Abstract: Large Language Models (LLMs) often falter at complex planning tasks that require exploration and self-correction, as their linear reasoning process struggles to recover from early mistakes. While search algorithms like Monte Carlo Tree Search (MCTS) can explore alternatives, they are often ineffective when guided by sparse rewards and fail to leverage the rich semantic capabilities of LLMs. We introduce SPIRAL (Symbolic LLM Planning via Grounded and Reflective Search), a novel framework that embeds a cognitive architecture of three specialized LLM agents into an MCTS loop. SPIRAL's key contribution is its integrated planning pipeline where a Planner proposes creative next steps, a Simulator grounds the search by predicting realistic outcomes, and a Critic provides dense reward signals through reflection. This synergy transforms MCTS from a brute-force search into a guided, self-correcting reasoning process. On the DailyLifeAPIs and HuggingFace datasets, SPIRAL consistently outperforms the default Chain-of-Thought planning method and other state-of-the-art agents. More importantly, it substantially surpasses other state-of-the-art agents; for example, SPIRAL achieves 83.6% overall accuracy on DailyLifeAPIs, an improvement of over 16 percentage points against the next-best search framework, while also demonstrating superior token efficiency. Our work demonstrates that structuring LLM reasoning as a guided, reflective, and grounded search process yields more robust and efficient autonomous planners. The source code, full appendices, and all experimental data are available for reproducibility at the official project repository.
Authors: Yu Jiang, Xindi Tong, Ziyao Liu, Xiaoxi Zhang, Kwok-Yan Lam, Chee Wei Tan
Abstract: Federated unlearning has become an attractive approach to address privacy concerns in collaborative machine learning, for situations when sensitive data is remembered by AI models during the machine learning process. It enables the removal of specific data influences from trained models, aligning with the growing emphasis on the "right to be forgotten." While extensively studied in horizontal federated learning, unlearning in vertical federated learning (VFL) remains challenging due to the distributed feature architecture. VFL unlearning includes sample unlearning that removes specific data points' influence and label unlearning that removes entire classes. Since different parties hold complementary features of the same samples, unlearning tasks require cross-party coordination, creating computational overhead and complexities from feature interdependencies. To address such challenges, we propose FedORA (Federated Optimization for data Removal via primal-dual Algorithm), designed for sample and label unlearning in VFL. FedORA formulates the removal of certain samples or labels as a constrained optimization problem solved using a primal-dual framework. Our approach introduces a new unlearning loss function that promotes classification uncertainty rather than misclassification. An adaptive step size enhances stability, while an asymmetric batch design, considering the prior influence of the remaining data on the model, handles unlearning and retained data differently to efficiently reduce computational costs. We provide theoretical analysis proving that the model difference between FedORA and Train-from-scratch is bounded, establishing guarantees for unlearning effectiveness. Experiments on tabular and image datasets demonstrate that FedORA achieves unlearning effectiveness and utility preservation comparable to Train-from-scratch with reduced computation and communication overhead.
Authors: Zijian Liu
Abstract: Optimization under heavy-tailed noise has become popular recently, since it better fits many modern machine learning tasks, as captured by empirical observations. Concretely, instead of a finite second moment on gradient noise, a bounded ${\frak p}$-th moment where ${\frak p}\in(1,2]$ has been recognized to be more realistic (say being upper bounded by $\sigma_{\frak l}^{\frak p}$ for some $\sigma_{\frak l}\ge0$). A simple yet effective operation, gradient clipping, is known to handle this new challenge successfully. Specifically, Clipped Stochastic Gradient Descent (Clipped SGD) guarantees a high-probability rate ${\cal O}(\sigma_{\frak l}\ln(1/\delta)T^{1/{\frak p}-1})$ (resp. ${\cal O}(\sigma_{\frak l}^2\ln^2(1/\delta)T^{2/{\frak p}-2})$) for nonsmooth convex (resp. strongly convex) problems, where $\delta\in(0,1]$ is the failure probability and $T\in\mathbb{N}$ is the time horizon. In this work, we provide a refined analysis for Clipped SGD and offer two faster rates, ${\cal O}(\sigma_{\frak l}d_{\rm eff}^{-1/2{\frak p}}\ln^{1-1/{\frak p}}(1/\delta)T^{1/{\frak p}-1})$ and ${\cal O}(\sigma_{\frak l}^2d_{\rm eff}^{-1/{\frak p}}\ln^{2-2/{\frak p}}(1/\delta)T^{2/{\frak p}-2})$, than the aforementioned best results, where $d_{\rm eff}\ge1$ is a quantity we call the $\textit{generalized effective dimension}$. Our analysis improves upon the existing approach on two sides: better utilization of Freedman's inequality and finer bounds for clipping error under heavy-tailed noise. In addition, we extend the refined analysis to convergence in expectation and obtain new rates that break the known lower bounds. Lastly, to complement the study, we establish new lower bounds for both high-probability and in-expectation convergence. Notably, the in-expectation lower bounds match our new upper bounds, indicating the optimality of our refined analysis for convergence in expectation.
Authors: Saif Khalfan Saif Al Mazrouei
Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation, yet they exhibit systematic errors on complex, multi-step programming tasks. We hypothesize that these errors stem from the flexibility of general-purpose languages, which permits multiple valid approaches and requires implicit state management. To test this hypothesis, we introduce Anka, a domain-specific language (DSL) for data transformation pipelines designed with explicit, constrained syntax that reduces ambiguity in code generation. Despite having zero prior training exposure to Anka, Claude 3.5 Haiku achieves 99.9% parse success and 95.8% overall task accuracy across 100 benchmark problems. Critically, Anka demonstrates a 40 percentage point accuracy advantage over Python on multi-step pipeline tasks (100% vs. 60%), where Python's flexible syntax leads to frequent errors in operation sequencing and variable management. Cross-model validation with GPT-4o-mini confirms this advantage (+26.7 percentage points on multi-step tasks). Our results demonstrate that: (1) LLMs can learn novel DSLs entirely from in-context prompts, achieving near-native accuracy; (2) constrained syntax significantly reduces errors on complex tasks; and (3) domain-specific languages purposefully designed for LLM generation can outperform general-purpose languages on which the LLM has extensive training. We release the complete language implementation, benchmark suite, and evaluation framework to facilitate further research.
Authors: Dianyun Wang, Qingsen Ma, Yuhu Shang, Zhifeng Lu, Lechen Ning, Zhenbo Xu, Huijia Wu, Zhaofeng He
Abstract: Parameter-efficient fine-tuning has become the dominant paradigm for adapting large language models to downstream tasks. Low-rank adaptation methods such as LoRA operate under the assumption that task-relevant weight updates reside in a low-rank subspace, yet this subspace is learned implicitly from data in a black-box manner, offering no interpretability or direct control. We hypothesize that this difficulty stems from polysemanticity--individual dimensions encoding multiple entangled concepts. To address this, we leverage pre-trained Sparse Autoencoders (SAEs) to identify task-relevant features in a disentangled feature space, then construct an explicit, interpretable low-rank subspace to guide adapter initialization. We provide theoretical analysis proving that under monosemanticity assumptions, SAE-based subspace identification achieves arbitrarily small recovery error, while direct identification in polysemantic space suffers an irreducible error floor. On safety alignment, our method achieves up to 99.6% safety rate--exceeding full fine-tuning by 7.4 percentage points and approaching RLHF-based methods--while updating only 0.19-0.24% of parameters. Crucially, our method provides interpretable insights into the learned alignment subspace through the semantic grounding of SAE features. Our work demonstrates that incorporating mechanistic interpretability into the fine-tuning process can simultaneously improve both performance and transparency.
Authors: Mahdi Kchaou, Francesco Contino, Diederik Coppitters
Abstract: Given the central role of green e-molecule imports in the European energy transition, many studies optimize import pathways and identify a single cost-optimal solution. However, cost optimality is fragile, as real-world implementation depends on regulatory, spatial, and stakeholder constraints that are difficult to represent in optimization models and can render cost-optimal designs infeasible. To address this limitation, we generate a diverse set of near-cost-optimal alternatives within an acceptable cost margin using Modeling to Generate Alternatives, accounting for unmodeled uncertainties. Interpretable machine learning is then applied to extract insights from the resulting solution space. The approach is applied to hydrogen import pathways considering hydrogen, ammonia, methane, and methanol as carriers. Results reveal a broad near-optimal space with great flexibility: solar, wind, and storage are not strictly required to remain within 10% of the cost optimum. Wind constraints favor solar-storage methanol pathways, while limited storage favors wind-based ammonia or methane pathways.
Authors: Yoonpyo Lee, Kazuma Kobayashi, Sai Puppala, Sajedul Talukder, Seid Koric, Souvik Chakraborty, Syed Bahauddin Alam
Abstract: The prevailing paradigm in AI for physical systems, scaling general-purpose foundation models toward universal multimodal reasoning, confronts a fundamental barrier at the control interface. Recent benchmarks show that even frontier vision-language models achieve only 50-53% accuracy on basic quantitative physics tasks, behaving as approximate guessers that preserve semantic plausibility while violating physical constraints. This input unfaithfulness is not a scaling deficiency but a structural limitation. Perception-centric architectures optimize parameter-space imitation, whereas safety-critical control demands outcome-space guarantees over executed actions. Here, we present a fundamentally different pathway toward domain-specific foundation models by introducing compact language models operating as Agentic Physical AI, in which policy optimization is driven by physics-based validation rather than perceptual inference. We train a 360-million-parameter model on synthetic reactor control scenarios, scaling the dataset from 10^3 to 10^5 examples. This induces a sharp phase transition absent in general-purpose models. Small-scale systems exhibit high-variance imitation with catastrophic tail risk, while large-scale models undergo variance collapse exceeding 500x reduction, stabilizing execution-level behavior. Despite balanced exposure to four actuation families, the model autonomously rejects approximately 70% of the training distribution and concentrates 95% of runtime execution on a single-bank strategy. Learned representations transfer across distinct physics and continuous input modalities without architectural modification.
Authors: Xiu Li
Abstract: We study visual representation learning from a structural and topological perspective. We begin from a single hypothesis: that visual understanding presupposes a semantic language for vision, in which many perceptual observations correspond to a small number of discrete semantic states. Together with widely assumed premises on transferability and abstraction in representation learning, this hypothesis implies that the visual observation space must be organized in a fiber bundle like structure, where nuisance variation populates fibers and semantics correspond to a quotient base space. From this structure we derive two theoretical consequences. First, the semantic quotient $X/G$ is not a submanifold of $X$ and cannot be obtained through smooth deformation alone, semantic invariance requires a non-homeomorphic, discriminative target, for example, supervision via labels, cross instance identification, or multimodal alignment that supplies explicit semantic equivalence. Second, we show that approximating the quotient also places structural demands on the model architecture. Semantic abstraction requires not only an external semantic target, but a representation mechanism capable of supporting topology change: an expand-and-snap process in which the manifold is first geometrically expanded to separate structure and then collapsed to form discrete semantic regions. We emphasize that these results are interpretive rather than prescriptive: the framework provides a topological lens that aligns with empirical regularities observed in large-scale discriminative and multimodal models, and with classical principles in statistical learning theory.
Authors: Sel\c{c}uk Kayacan
Abstract: We propose a functorial framework for persistent homology based on finite topological spaces and their associated posets. Starting from a finite metric space, we associate a filtration of finite topologies whose structure maps are continuous identity maps. By passing functorially to posets and to simplicial complexes via crosscut constructions, we obtain persistence modules without requiring inclusion relations between the resulting complexes. We show that standard poset-level simplifications preserve persistent invariants and prove stability of the resulting persistence diagrams under perturbations of the input metric in a density-based instantiation.
Authors: Abd Ullah Khan, Uman Khalid, Muhammad Tanveer, Trung Q. Duong, Hyundong Shin
Abstract: A beyond-diagonal reconfigurable intelligent surface (BD-RIS) is an innovative type of reconfigurable intelligent surface (RIS) that has recently been proposed and is considered a revolutionary advancement in wave manipulation. Unlike the mutually disconnected arrangement of elements in traditional RISs, BD-RIS creates cost-effective and simple inter-element connections, allowing for greater freedom in configuring the amplitude and phase of impinging waves. However, there are numerous underlying challenges in realizing the advantages associated with BD-RIS, prompting the research community to actively investigate cutting-edge schemes and algorithms in this direction. Particularly, the passive beamforming design for BD-RIS under specific environmental conditions has become a major focus in this research area. In this article, we provide a systematic introduction to BD-RIS, elaborating on its functional principles concerning architectural design, promising advantages, and classification. Subsequently, we present recent advances and identify a series of challenges and opportunities. Additionally, we consider a specific case study where beamforming is designed using four different algorithms, and we analyze their performance with respect to sum rate and computation cost. To augment the beamforming capabilities in 6G BD-RIS with quantum enhancement, we analyze various hybrid quantum-classical machine learning (ML) models to improve beam prediction performance, employing real-world communication Scenario 8 from the DeepSense 6G dataset. Consequently, we derive useful insights about the practical implications of BD-RIS.
Authors: Bruno Mlodozeniec, David Krueger, Richard E. Turner
Abstract: Causal inference is a key research area in machine learning, yet confusion reigns over the tools needed to tackle it. There are prevalent claims in the machine learning literature that you need a bespoke causal framework or notation to answer causal questions. In this paper, we want to make it clear that you \emph{can} answer any causal inference question within the realm of probabilistic modelling and inference, without causal-specific tools or notation. Through concrete examples, we demonstrate how causal questions can be tackled by writing down the probability of everything. Lastly, we reinterpret causal tools as emerging from standard probabilistic modelling and inference, elucidating their necessity and utility.
Authors: Jinye Du, Quan Yuan, Zuyao Zhang, Yanzhi Yi, Jiahui Hu, Wangyi Chen, Yiyang Zhu, Qishui Zheng, Wenxiang Zou, Xiangyu Chang, Zuohe Zheng, Zichun Ye, Chao Liu, Shanni Li, Renwei Zhang, Yiping Deng, Xinwei Hu, Xuefeng Jin, Jie Zhao
Abstract: Modern AI models demand high-performance computation kernels. The growing complexity of LLMs, multimodal architectures, and recommendation systems, combined with techniques like sparsity and quantization, creates significant computational challenges. Moreover, frequent hardware updates and diverse chip architectures further complicate this landscape, requiring tailored kernel implementations for each platform. However, manual optimization cannot keep pace with these demands, creating a critical bottleneck in AI system development. Recent advances in LLM code generation capabilities have opened new possibilities for automating kernel development. In this work, we propose AKG kernel agent (AI-driven Kernel Generator), a multi-agent system that automates kernel generation, migration, and performance tuning. AKG kernel agent is designed to support multiple domain-specific languages (DSLs), including Triton, TileLang, CPP, and CUDA-C, enabling it to target different hardware backends while maintaining correctness and portability. The system's modular design allows rapid integration of new DSLs and hardware targets. When evaluated on KernelBench using Triton DSL across GPU and NPU backends, AKG kernel agent achieves an average speedup of 1.46$\times$ over PyTorch Eager baselines implementations, demonstrating its effectiveness in accelerating kernel development for modern AI workloads.
Authors: William Kengne, Modou Wade
Abstract: This paper develops a general approach for deep learning for a setting that includes nonparametric regression and classification. We perform a framework from data that fulfills a generalized Bernstein-type inequality, including independent, $\phi$-mixing, strongly mixing and $\mathcal{C}$-mixing observations. Two estimators are proposed: a non-penalized deep neural network estimator (NPDNN) and a sparse-penalized deep neural network estimator (SPDNN). For each of these estimators, bounds of the expected excess risk on the class of H\"older smooth functions and composition H\"older functions are established. Applications to independent data, as well as to $\phi$-mixing, strongly mixing, $\mathcal{C}$-mixing processes are considered. For each of these examples, the upper bounds of the expected excess risk of the proposed NPDNN and SPDNN predictors are derived. It is shown that both the NPDNN and SPDNN estimators are minimax optimal (up to a logarithmic factor) in many classical settings.
Authors: Jesse Brouwers, Xiaoyan Xing, Alexander Timans
Abstract: Foundation models for segmentation such as the Segment Anything Model (SAM) family exhibit strong zero-shot performance, but remain vulnerable in shifted or limited-knowledge domains. This work investigates whether uncertainty quantification can mitigate such challenges and enhance model generalisability in a domain-agnostic manner. To this end, we (1) curate UncertSAM, a benchmark comprising eight datasets designed to stress-test SAM under challenging segmentation conditions including shadows, transparency, and camouflage; (2) evaluate a suite of lightweight, post-hoc uncertainty estimation methods; and (3) assess a preliminary uncertainty-guided prediction refinement step. Among evaluated approaches, a last-layer Laplace approximation yields uncertainty estimates that correlate well with segmentation errors, indicating a meaningful signal. While refinement benefits are preliminary, our findings underscore the potential of incorporating uncertainty into segmentation models to support robust, domain-agnostic performance. Our benchmark and code are made publicly available.
Authors: Jiada Huang, Hao Ma, Zhibin Shen, Yizhou Qiao, Haiyang Li
Abstract: Local high strain in solid rocket motor grains is a primary cause of structural failure. However, traditional numerical simulations are computationally expensive, and existing surrogate models cannot explicitly establish geometric models and accurately capture high-strain regions. Therefore, this paper proposes an adaptive graph network, GrainGNet, which employs an adaptive pooling dynamic node selection mechanism to effectively preserve the key mechanical features of structurally critical regions, while concurrently utilising feature fusion to transmit deep features and enhance the model's representational capacity. In the joint prediction task involving four sequential conditions--curing and cooling, storage, overloading, and ignition--GrainGNet reduces the mean squared error by 62.8% compared to the baseline graph U-Net model, with only a 5.2% increase in parameter count and an approximately sevenfold improvement in training efficiency. Furthermore, in the high-strain regions of debonding seams, the prediction error is further reduced by 33% compared to the second-best method, offering a computationally efficient and high-fidelity approach to evaluate motor structural safety.
Authors: Manuel Franco-Vivo
Abstract: As autonomous vehicle technology advances, ensuring the safety and reliability of these systems becomes paramount. Consequently, comprehensive testing methodologies are essential to evaluate the performance of autonomous vehicles in diverse and complex real-world scenarios. This study focuses on the behaviour coverage analysis of a multi-agent system simulation designed for autonomous vehicle testing, and provides a systematic approach to measure and assess behaviour coverage within the simulation environment. By defining a set of driving scenarios, and agent interactions, we evaluate the extent to which the simulation encompasses a broad range of behaviours relevant to autonomous driving. Our findings highlight the importance of behaviour coverage in validating the effectiveness and robustness of autonomous vehicle systems. Through the analysis of behaviour coverage metrics and coverage-based testing, we identify key areas for improvement and optimization in the simulation framework. Thus, a Model Predictive Control (MPC) pedestrian agent is proposed, where its objective function is formulated to encourage \textit{interesting} tests while promoting a more realistic behaviour than other previously studied pedestrian agents. This research contributes to advancing the field of autonomous vehicle testing by providing insights into the comprehensive evaluation of system behaviour in simulated environments. The results offer valuable implications for enhancing the safety, reliability, and performance of autonomous vehicles through rigorous testing methodologies.
Authors: Ang Lv, Jin Ma, Yiyuan Ma, Siyuan Qiao
Abstract: Mixture-of-Experts (MoE) models lack explicit constraints to ensure the router's decisions align well with the experts' capabilities, which ultimately limits model performance. To address this, we propose expert-router coupling (ERC) loss, a lightweight auxiliary loss that tightly couples the router's decisions with expert capabilities. Our approach treats each expert's router embedding as a proxy token for the tokens assigned to that expert, and feeds perturbed router embeddings through the experts to obtain internal activations. The ERC loss enforces two constraints on these activations: (1) Each expert must exhibit higher activation for its own proxy token than for the proxy tokens of any other expert. (2) Each proxy token must elicit stronger activation from its corresponding expert than from any other expert. These constraints jointly ensure that each router embedding faithfully represents its corresponding expert's capability, while each expert specializes in processing the tokens actually routed to it. The ERC loss is computationally efficient, operating only on n^2 activations, where n is the number of experts. This represents a fixed cost independent of batch size, unlike prior coupling methods that scale with the number of tokens (often millions per batch). Through pre-training MoE-LLMs ranging from 3B to 15B parameters and extensive analysis on trillions of tokens, we demonstrate the effectiveness of the ERC loss. Moreover, the ERC loss offers flexible control and quantitative tracking of expert specialization levels during training, providing valuable insights into MoEs.
Authors: Kongcheng Zhang, Qi Yao, Shunyu Liu, Wenjian Zhang, Min Cen, Yang Zhou, Wenkai Fang, Yiru Zhao, Baisheng Lai, Mingli Song
Abstract: Reinforcement Learning (RL) has shown promise for aligning Large Language Models (LLMs) to follow instructions with various constraints. Despite the encouraging results, RL improvement inevitably relies on sampling successful, high-quality responses; however, the initial model often struggles to generate responses that satisfy all constraints due to its limited capabilities, yielding sparse or indistinguishable rewards that impede learning. In this work, we propose Hindsight instruction Replay (HiR), a novel sample-efficient RL framework for complex instruction following tasks, which employs a select-then-rewrite strategy to replay failed attempts as successes based on the constraints that have been satisfied in hindsight. We perform RL on these replayed samples as well as the original ones, theoretically framing the objective as dual-preference learning at both the instruction- and response-level to enable efficient optimization using only a binary reward signal. Extensive experiments demonstrate that the proposed HiR yields promising results across different instruction following tasks, while requiring less computational budget. Our code and dataset is available at https://github.com/sastpg/HIR.
Authors: Zuoyou Jiang, Li Zhao, Rui Sun, Ruohan Sun, Zhongjian Li, Jing Li, Daxin Jiang, Zuo Bai, Cheng Hua
Abstract: Signal decay and regime shifts pose recurring challenges for data-driven investment strategies in non-stationary markets. Conventional time-series and machine learning approaches, which rely primarily on historical correlations, often struggle to generalize when the economic environment changes. While large language models (LLMs) offer strong capabilities for processing unstructured information, their potential to support quantitative factor screening through explicit economic reasoning remains underexplored. Existing factor-based methods typically reduce alphas to numerical time series, overlooking the semantic rationale that determines when a factor is economically relevant. We propose Alpha-R1, an 8B-parameter reasoning model trained via reinforcement learning for context-aware alpha screening. Alpha-R1 reasons over factor logic and real-time news to evaluate alpha relevance under changing market conditions, selectively activating or deactivating factors based on contextual consistency. Empirical results across multiple asset pools show that Alpha-R1 consistently outperforms benchmark strategies and exhibits improved robustness to alpha decay. The full implementation and resources are available at https://github.com/FinStep-AI/Alpha-R1.
Authors: Dimitra Maoutsa
Abstract: How can we learn the laws underlying the dynamics of stochastic systems when their trajectories are sampled sparsely in time? Existing methods either require temporally resolved high-frequency observations, or rely on geometric arguments that apply only to conservative systems, limiting the range of dynamics they can recover. Here, we present a new framework that reconciles these two perspectives by reformulating inference as a stochastic control problem. Our method uses geometry-driven path augmentation, guided by the geometry in the system's invariant density to reconstruct likely trajectories and infer the underlying dynamics without assuming specific parametric models. Applied to overdamped Langevin systems, our approach accurately recovers stochastic dynamics even from extremely undersampled data, outperforming existing methods in synthetic benchmarks. This work demonstrates the effectiveness of incorporating geometric inductive biases into stochastic system identification methods.
Authors: Agostino Capponi, Chengpiao Huang, J. Antonio Sidaoui, Kaizheng Wang, Jiacheng Zou
Abstract: We investigate machine learning models for stock return prediction in non-stationary environments, revealing a fundamental nonstationarity-complexity tradeoff: complex models reduce misspecification error but require longer training windows that introduce stronger non-stationarity. We resolve this tension with a novel model selection method that jointly optimizes model class and training window size using a tournament procedure that adaptively evaluates candidates on non-stationary validation data. Our theoretical analysis demonstrates that this approach balances misspecification error, estimation variance, and non-stationarity, performing close to the best model in hindsight. Applying our method to 17 industry portfolio returns, we consistently outperform standard rolling-window benchmarks, improving out-of-sample $R^2$ by 14-23% on average. During NBER-designated recessions, improvements are substantial: our method achieves positive $R^2$ during the Gulf War recession while benchmarks are negative, and improves $R^2$ in absolute terms by at least 80bps during the 2001 recession as well as superior performance during the 2008 Financial Crisis. Economically, a trading strategy based on our selected model generates 31% higher cumulative returns averaged across the industries.
Authors: Federico Baldo, Charles K. Assaad
Abstract: Most causal discovery methods recover a completed partially directed acyclic graph representing a Markov equivalence class from observational data. Recent work has extended these methods to federated settings to address data decentralization and privacy constraints, but often under idealized assumptions that all clients share the same causal model. Such assumptions are unrealistic in practice, as client-specific policies or protocols, for example, across hospitals, naturally induce heterogeneous and unknown interventions. In this work, we address federated causal discovery under unknown client-level interventions. We propose I-PERI, a novel federated algorithm that first recovers the CPDAG of the union of client graphs and then orients additional edges by exploiting structural differences induced by interventions across clients. This yields a tighter equivalence class, which we call the $\mathbf{\Phi}$-Markov Equivalence Class, represented by the $\mathbf{\Phi}$-CPDAG. We provide theoretical guarantees on the convergence of I-PERI, as well as on its privacy-preserving properties, and present empirical evaluations on synthetic data demonstrating the effectiveness of the proposed algorithm.
Authors: Shu Pu, Boya Zeng, Kaichen Zhou, Mengyu Wang, Zhuang Liu
Abstract: Generative models are increasingly used in 3D vision to synthesize novel shapes, yet it remains unclear whether their generation relies on memorizing training shapes. Understanding their memorization could help prevent training data leakage and improve the diversity of generated results. In this paper, we design an evaluation framework to quantify memorization in 3D generative models and study the influence of different data and modeling designs on memorization. We first apply our framework to quantify memorization in existing methods. Next, through controlled experiments with a latent vector-set (Vecset) diffusion model, we find that, on the data side, memorization depends on data modality, and increases with data diversity and finer-grained conditioning; on the modeling side, it peaks at a moderate guidance scale and can be mitigated by longer Vecsets and simple rotation augmentation. Together, our framework and analysis provide an empirical understanding of memorization in 3D generative models and suggest simple yet effective strategies to reduce it without degrading generation quality. Our code is available at https://github.com/zlab-princeton/3d_mem.
Authors: LearnLM Team, Eedi, :, Albert Wang, Aliya Rysbek, Andrea Huber, Anjali Nambiar, Anna Kenolty, Ben Caulfield, Beth Lilley-Draper, Bibi Groot, Brian Veprek, Chelsea Burdett, Claire Willis, Craig Barton, Digory Smith, George Mu, Harriet Walters, Irina Jurenka, Iris Hulls, James Stalley-Moores, Jonathan Caton, Julia Wilkowski, Kaiz Alarakyia, Kevin R. McKee, Liam McCafferty, Lucy Dalton, Markus Kunesch, Pauline Malubay, Rachel Kidson, Rich Wells, Sam Wheeler, Sara Wiltberger, Shakir Mohamed, Simon Woodhead, Vasco Braz\~ao
Abstract: One-to-one tutoring is widely considered the gold standard for personalized education, yet it remains prohibitively expensive to scale. To evaluate whether generative AI might help expand access to this resource, we conducted an exploratory randomized controlled trial (RCT) with $N = 165$ students across five UK secondary schools. We integrated LearnLM -- a generative AI model fine-tuned for pedagogy -- into chat-based tutoring sessions on the Eedi mathematics platform. In the RCT, expert tutors directly supervised LearnLM, with the remit to revise each message it drafted until they would be satisfied sending it themselves. LearnLM proved to be a reliable source of pedagogical instruction, with supervising tutors approving 76.4% of its drafted messages making zero or minimal edits (i.e., changing only one or two characters). This translated into effective tutoring support: students guided by LearnLM performed at least as well as students chatting with human tutors on each learning outcome we measured. In fact, students who received support from LearnLM were 5.5 percentage points more likely to solve novel problems on subsequent topics (with a success rate of 66.2%) than those who received tutoring from human tutors alone (rate of 60.7%). In interviews, tutors highlighted LearnLM's strength at drafting Socratic questions that encouraged deeper reflection from students, with multiple tutors even reporting that they learned new pedagogical practices from the model. Overall, our results suggest that pedagogically fine-tuned AI tutoring systems may play a promising role in delivering effective, individualized learning support at scale.
Authors: Konstantin Yakovlev, Nikita Puchkin
Abstract: We present a theory for simultaneous approximation of the score function and its derivatives, enabling the handling of data distributions with low-dimensional structure and unbounded support. Our approximation error bounds match those in the literature while relying on assumptions that relax the usual bounded support requirement. Crucially, our bounds are free from the curse of dimensionality. Moreover, we establish approximation guarantees for derivatives of any prescribed order, extending beyond the commonly considered first-order setting.
Authors: Tiffany Ding, Isaac Gibbs, Ryan J. Tibshirani
Abstract: We present an online method for guaranteeing calibration of quantile forecasts at multiple quantile levels simultaneously. A sequence of $\alpha$-level quantile forecasts is calibrated if the forecasts are larger than the target value at an $\alpha$-fraction of time steps. We introduce a lightweight method called Multi-Level Quantile Tracker (MultiQT) that wraps around any existing point or quantile forecaster to produce corrected forecasts guaranteed to achieve calibration, even against adversarial distribution shifts, while ensuring that the forecasts are ordered -- e.g., the 0.5-level quantile forecast is never larger than the 0.6-level forecast. Furthermore, the method comes with a no-regret guarantee that implies it will not worsen the performance of an existing forecaster, asymptotically, with respect to the quantile loss. In experiments, we find that MultiQT significantly improves the calibration of real forecasters in epidemic and energy forecasting problems.
Authors: Lars van der Laan, Nathan Kallus
Abstract: We introduce Iterated Bellman Calibration, a simple, model-agnostic, post-hoc procedure for calibrating off-policy value predictions in infinite-horizon Markov decision processes. Bellman calibration requires that states with similar predicted long-term returns exhibit one-step returns consistent with the Bellman equation under the target policy. We adapt classical histogram and isotonic calibration to the dynamic, counterfactual setting by repeatedly regressing fitted Bellman targets onto a model's predictions, using a doubly robust pseudo-outcome to handle off-policy data. This yields a one-dimensional fitted value iteration scheme that can be applied to any value estimator. Our analysis provides finite-sample guarantees for both calibration and prediction under weak assumptions, and critically, without requiring Bellman completeness or realizability.
Authors: Jing Huang, Shujian Zhang, Lun Wang, Andrew Hard, Rajiv Mathews, John Lambert
Abstract: Identifying specific and often complex behaviors from large language models (LLMs) in conversational settings is crucial for their evaluation. Recent work proposes novel techniques to find natural language prompts that induce specific behaviors from a target model, yet they are mainly studied in single-turn settings. In this work, we study behavior elicitation in the context of multi-turn conversations. We first offer an analytical framework that categorizes existing methods into three families based on their interactions with the target model: those that use only prior knowledge, those that use offline interactions, and those that learn from online interactions. We then introduce a generalized multi-turn formulation of the online method, unifying single-turn and multi-turn elicitation. We evaluate all three families of methods on automatically generating multi-turn test cases. We investigate the efficiency of these approaches by analyzing the trade-off between the query budget, i.e., the number of interactions with the target model, and the success rate, i.e., the discovery rate of behavior-eliciting inputs. We find that online methods can achieve an average success rate of 45/19/77% with just a few thousand queries over three tasks where static methods from existing multi-turn conversation benchmarks find few or even no failure cases. Our work highlights a novel application of behavior elicitation methods in multi-turn conversation evaluation and the need for the community to move towards dynamic benchmarks.
Authors: Nisar Ahmed, Hafiz Muhammad Shahzad Asif, Gulshan Saleem, Muhammad Usman Younus
Abstract: Crop yield is affected by various soil and environmental parameters and can vary significantly. Therefore, a crop yield estimation model which can predict pre-harvest yield is required for food security. The study is conducted on tea forms operating under National Tea Research Institute, Pakistan. The data is recorded on monthly basis for ten years period. The parameters collected are minimum and maximum temperature, humidity, rainfall, PH level of the soil, usage of pesticide and labor expertise. The design of model incorporated all of these parameters and identified the parameters which are most crucial for yield predictions. Feature transformation is performed to obtain better performing model. The designed model is based on an ensemble of neural networks and provided an R-squared of 0.9461 and RMSE of 0.1204 indicating the usability of the proposed model in yield forecasting based on surface and environmental parameters.
Authors: Vahan Arsenyan, Antoine Grosnit, Haitham Bou-Ammar, Arnak Dalalyan
Abstract: We introduce a unified framework for contextual and causal Bayesian optimisation, which aims to design intervention policies maximising the expectation of a target variable. Our approach leverages both observed contextual information and known causal graph structures to guide the search. Within this framework, we propose a novel algorithm that jointly optimises over policies and the sets of variables on which these policies are defined. This thereby extends and unifies two previously distinct approaches: Causal Bayesian Optimisation and Contextual Bayesian Optimisation, while also addressing their limitations in scenarios that yield suboptimal results. We derive worst-case and instance-dependent high-probability regret bounds for our algorithm. We report experimental results across diverse environments, corroborating that our approach achieves sublinear regret and reduces sample complexity in high-dimensional settings.
Authors: Wanda Hou, Miao Li, Yi-Zhuang You
Abstract: Generative models aim to learn the probability distributions underlying data, enabling the generation of new, realistic samples. Quantum inspired generative models, such as Born machines based on the matrix product state framework, have demonstrated remarkable capabilities in unsupervised learning tasks. This study advances the Born machine paradigm by introducing trainable token embeddings through positive operator valued measurements, replacing the traditional approach of static tensor indices. Key technical innovations include encoding tokens as quantum measurement operators with trainable parameters and leveraging QR decomposition to adjust the physical dimensions of the MPS. This approach maximizes the utilization of operator space and enhances the model's expressiveness. Empirical results on RNA data demonstrate that the proposed method significantly reduces negative log likelihood compared to one hot embeddings, with higher physical dimensions further enhancing single site probabilities and multi site correlations. The model also outperforms GPT2 in single site estimation and achieves competitive correlation modeling, showcasing the potential of trainable POVM embeddings for complex data correlations in quantum inspired sequence modeling.
Authors: Zijian Liu, Zhengyuan Zhou
Abstract: In the past several years, the last-iterate convergence of the Stochastic Gradient Descent (SGD) algorithm has triggered people's interest due to its good performance in practice but lack of theoretical understanding. For Lipschitz convex functions, different works have established the optimal $O(\log(1/\delta)\log T/\sqrt{T})$ or $O(\sqrt{\log(1/\delta)/T})$ high-probability convergence rates for the final iterate, where T is the time horizon and \delta is the failure probability. However, to prove these bounds, all the existing works are either limited to compact domains or require almost surely bounded noise. It is natural to ask whether the last iterate of SGD can still guarantee the optimal convergence rate but without these two restrictive assumptions. Besides this important question, there are still lots of theoretical problems lacking an answer. For example, compared with the last-iterate convergence of SGD for non-smooth problems, only few results for smooth optimization have yet been developed. Additionally, the existing results are all limited to a non-composite objective and the standard Euclidean norm. It still remains unclear whether the last-iterate convergence can be provably extended to wider composite optimization and non-Euclidean norms. In this work, to address the issues mentioned above, we revisit the last-iterate convergence of stochastic gradient methods and provide the first unified way to prove the convergence rates both in expectation and in high probability to accommodate general domains, composite objectives, non-Euclidean norms, Lipschitz conditions, smoothness, and (strong) convexity simultaneously. Additionally, we extend our analysis to obtain the last-iterate convergence under heavy-tailed noise.
Authors: Timo Kaufmann, Paul Weng, Viktor Bengs, Eyke H\"ullermeier
Abstract: Reinforcement learning from human feedback (RLHF) is a variant of reinforcement learning (RL) that learns from human feedback instead of relying on an engineered reward function. Building on prior work on the related setting of preference-based reinforcement learning (PbRL), it stands at the intersection of artificial intelligence and human-computer interaction. This positioning provides a promising approach to enhance the performance and adaptability of intelligent systems while also improving the alignment of their objectives with human values. The success in training large language models (LLMs) has impressively demonstrated this potential in recent years, where RLHF has played a decisive role in directing the model's capabilities towards human objectives. This article provides an overview of the fundamentals of RLHF, exploring how RL agents interact with human feedback. While recent focus has been on RLHF for LLMs, our survey covers the technique across multiple domains. We provide our most comprehensive coverage in control and robotics, where many fundamental techniques originate, alongside a dedicated LLM section. We examine the core principles that underpin RLHF, how algorithms and human feedback work together, and the main research trends in the field. Our goal is to give researchers and practitioners a clear understanding of this rapidly growing field.
Authors: Barak Or
Abstract: Velocity estimation is a core component of state estimation and sensor fusion pipelines in mobile robotics and autonomous ground systems, directly affecting navigation accuracy, control stability, and operational safety. In conventional systems, velocity is obtained through wheel encoders, inertial navigation units, or tightly coupled multi-sensor fusion architectures. However, these sensing configurations are not always available or reliable, particularly in low-cost, redundancy-constrained, or degraded operational scenarios where sensors may fail, drift, or become temporarily unavailable. This paper investigates the feasibility of estimating vehicle speed using only a single low-cost inertial sensor: a three-axis accelerometer embedded in a commodity smartphone. We present CarSpeedNet, a learning-based inertial estimation framework designed to infer speed directly from raw accelerometer measurements, without access to gyroscopes, wheel odometry, vehicle bus data, or external positioning during inference. From a sensor fusion perspective, this setting represents an extreme case of sensing sparsity, in which classical integration-based or filter-based approaches become unstable due to bias accumulation and partial observability. Rather than explicitly estimating physical states such as orientation or sensor bias, the proposed approach performs implicit latent-state approximation from temporal accelerometer data.
Authors: Jianing He, Qi Zhang, Weiping Ding, Duoqian Miao, Jun Zhao, Liang Hu, Longbing Cao
Abstract: Early exiting has demonstrated its effectiveness in accelerating the inference of pre-trained language models like BERT by dynamically adjusting the number of layers executed. However, most existing early exiting methods only consider local information from an individual test sample to determine their exiting indicators, failing to leverage the global information offered by sample population. This leads to suboptimal estimation of prediction correctness, resulting in erroneous exiting decisions. To bridge the gap, we explore the necessity of effectively combining both local and global information to ensure reliable early exiting during inference. Purposefully, we leverage prototypical networks to learn class prototypes and devise a distance metric between samples and class prototypes. This enables us to utilize global information for estimating the correctness of early predictions. On this basis, we propose a novel Distance-Enhanced Early Exiting framework for BERT (DE$^3$-BERT). DE$^3$-BERT implements a hybrid exiting strategy that supplements classic entropy-based local information with distance-based global information to enhance the estimation of prediction correctness for more reliable early exiting decisions. Extensive experiments on the GLUE benchmark demonstrate that DE$^3$-BERT consistently outperforms state-of-the-art models under different speed-up ratios with minimal storage or computational overhead, yielding a better trade-off between model performance and inference efficiency. Additionally, an in-depth analysis further validates the generality and interpretability of our method.
Authors: David Rolnick, Alan Aspuru-Guzik, Sara Beery, Bistra Dilkina, Priya L. Donti, Marzyeh Ghassemi, Hannah Kerner, Claire Monteleoni, Esther Rolf, Milind Tambe, Adam White
Abstract: In this position paper, we argue that application-driven research has been systemically under-valued in the machine learning community. As applications of machine learning proliferate, innovative algorithms inspired by specific real-world challenges have become increasingly important. Such work offers the potential for significant impact not merely in domains of application but also in machine learning itself. In this paper, we describe the paradigm of application-driven research in machine learning, contrasting it with the more standard paradigm of methods-driven research. We illustrate the benefits of application-driven machine learning and how this approach can productively synergize with methods-driven work. Despite these benefits, we find that reviewing, hiring, and teaching practices in machine learning often hold back application-driven innovation. We outline how these processes may be improved.
Authors: Adam Jelley, Yuhan Cao, Dave Bignell, Amos Storkey, Sam Devlin, Tabish Rashid
Abstract: Training agents to act competently in complex 3D environments from high-dimensional visual information is challenging. Reinforcement learning is conventionally used to train such agents, but requires a carefully designed reward function, and is difficult to scale to obtain robust agents that generalize to new tasks. In contrast, Large Language Models (LLMs) demonstrate impressively general capabilities resulting from large-scale pre-training and post-training alignment, but struggle to act in complex environments. This position paper draws explicit analogies between decision-making agents and LLMs, and argues that agents should be trained like LLMs to achieve more general, robust, and aligned behaviors. We provide a proof-of-concept to demonstrate how the procedure for training LLMs can be used to train an agent in a 3D video game environment from pixels. We investigate the importance of each stage of the LLM training pipeline, while providing guidance and insights for successfully applying this approach to agents. Our paper provides an alternative perspective to contemporary LLM Agents on how recent progress in LLMs can be leveraged for decision-making agents, and we hope will illuminate a path towards developing more generally capable agents for video games and beyond. Project summary and videos: https://adamjelley.github.io/aligning-agents-like-llms .
URLs: https://adamjelley.github.io/aligning-agents-like-llms
Authors: Adam Jelley, Trevor McInroe, Sam Devlin, Amos Storkey
Abstract: Supervised imitation-based approaches are often favored over off-policy reinforcement learning approaches for learning policies offline, since their straightforward optimization objective makes them computationally efficient and stable to train. However, their performance is fundamentally limited by the behavior policy that collected the dataset. Off-policy reinforcement learning provides a promising approach for improving on the behavior policy, but training is often computationally inefficient and unstable due to temporal-difference bootstrapping. In this paper, we propose a best-of-both approach by pre-training with supervised learning before improving performance with off-policy reinforcement learning. Specifically, we demonstrate improved efficiency by pre-training an actor with behavior cloning and a critic with a supervised Monte-Carlo value error. We find that we are able to substantially improve the training time of popular off-policy algorithms on standard benchmarks, and also achieve greater stability. Code is available at: https://github.com/AdamJelley/EfficientOfflineRL
Authors: Anqi Mao, Mehryar Mohri, Yutao Zhong
Abstract: Recent research has introduced a key notion of $H$-consistency bounds for surrogate losses. These bounds offer finite-sample guarantees, quantifying the relationship between the zero-one estimation error (or other target loss) and the surrogate loss estimation error for a specific hypothesis set. However, previous bounds were derived under the condition that a lower bound of the surrogate loss conditional regret is given as a convex function of the target conditional regret, without non-constant factors depending on the predictor or input instance. Can we derive finer and more favorable $H$-consistency bounds? In this work, we relax this condition and present a general framework for establishing enhanced $H$-consistency bounds based on more general inequalities relating conditional regrets. Our theorems not only subsume existing results as special cases but also enable the derivation of more favorable bounds in various scenarios. These include standard multi-class classification, binary and multi-class classification under Tsybakov noise conditions, and bipartite ranking.
Authors: Ali Royat, Seyed Mohamad Moghadas, Lesley De Cruz, Adrian Munteanu
Abstract: Deep neural networks (DNNs) have demonstrated remarkable performance across various domains, but their inherent complexity makes them challenging to interpret. This is especially true for temporal graph regression tasks due to the complex underlying spatio-temporal patterns in the graph. While interpretability concerns in Graph Neural Networks (GNNs) mirror those of DNNs, no notable work has addressed the interpretability of temporal GNNs to the best of our knowledge. Innovative methods, such as prototypes, aim to make DNN models more interpretable. However, a combined approach based on prototype-based methods and Information Bottleneck (IB) principles has not yet been developed for temporal GNNs. Our research introduces a novel approach that uniquely integrates these techniques to enhance the interpretability of temporal graph regression models. The key contributions of our work are threefold: We introduce the Graph INterpretability in Temporal Regression task using Information bottleneck and Prototype (GINTRIP) framework, the first combined application of IB and prototype-based methods for interpretable temporal graph tasks. We derive a novel theoretical bound on mutual information (MI), extending the applicability of IB principles to graph regression tasks. We incorporate an unsupervised auxiliary classification head, fostering diverse concept representation using multi-task learning, which enhances the model's interpretability. Our model is evaluated on real-world datasets like traffic and crime, outperforming existing methods in both forecasting accuracy and interpretability-related metrics such as MAE, RMSE, MAPE, and fidelity.
Authors: Qiang Ye
Abstract: Accelerated training algorithms, such as adaptive learning rates (or preconditioning) and various normalization methods, are widely used but not fully understood. When regularization is introduced, standard optimizers like adaptive learning rates may not perform effectively. This raises the need for alternative regularization approaches such as AdamW and the question of how to properly combine regularization with preconditioning. In this paper, we address these challenges using the theory of preconditioning as follows: (1) We explain how AdaGrad, RMSProp, and Adam accelerates training through improving Hessian conditioning; (2) We explore the interaction between $L_2$-regularization and preconditioning, demonstrating that AdamW amounts to selecting the underlying intrinsic parameters for regularization, and we derive a generalization for the $L_1$-regularization; and (3) We demonstrate how various normalization methods such as input data normalization, batch normalization, and layer normalization accelerate training by improving Hessian conditioning. Our analysis offers a unified mathematical framework for understanding various acceleration techniques or deriving appropriate regularization schemes.
Authors: Jaeyoung Park, Minsu Kim, Steven Euijong Whang
Abstract: Model fairness is becoming important in class-incremental learning for Trustworthy AI. While accuracy has been a central focus in class-incremental learning, fairness has been relatively understudied. However, naively using all the samples of the current task for training results in unfair catastrophic forgetting for certain sensitive groups including classes. We theoretically analyze that forgetting occurs if the average gradient vector of the current task data is in an "opposite direction" compared to the average gradient vector of a sensitive group, which means their inner products are negative. We then propose a fair class-incremental learning framework that adjusts the training weights of current task samples to change the direction of the average gradient vector and thus reduce the forgetting of underperforming groups and achieve fairness. For various group fairness measures, we formulate optimization problems to minimize the overall losses of sensitive groups while minimizing the disparities among them. We also show the problems can be solved with linear programming and propose an efficient Fairness-aware Sample Weighting (FSW) algorithm. Experiments show that FSW achieves better accuracy-fairness tradeoff results than state-of-the-art approaches on real datasets.
Authors: Zhan-Lun Chang, Dong-Jun Han, Seyyedali Hosseinalipour, Mung Chiang, Christopher G. Brinton
Abstract: Most federated learning (FL) approaches assume a fixed device set. However, real-world scenarios often involve devices dynamically joining or leaving the system, driven by, e.g., user mobility patterns or handovers across cell boundaries. This dynamic setting introduces unique challenges: (1) the optimization objective evolves with the active device set, unlike traditional FL's static objective; and (2) the current global model may no longer serve as an effective initialization for subsequent rounds, potentially hindering adaptation, delaying convergence, and reducing resource efficiency. To address these challenges, we first provide a convergence analysis for FL under a dynamic device set, accounting for factors such as gradient noise, local training iterations, and data heterogeneity. Building on this analysis, we propose a model initialization algorithm that enables rapid adaptation whenever devices join or leave the network. Our key idea is to compute a weighted average of previous global models, guided by gradient similarity, to prioritize models trained on data distributions that closely align with the current device set, thereby accelerating recovery from distribution shifts in fewer training rounds. This plug-and-play algorithm is designed to integrate seamlessly with existing FL methods, offering broad applicability. Experiments demonstrate that our approach achieves convergence speedups typically an order of magnitude or more compared to baselines, which we show drastically reduces energy consumption to reach a target accuracy.
Authors: Yawen Li, Yan Li, Junping Du, Yingxia Shao, Meiyu Liang, Guanhua Ye
Abstract: Personalized collaborative learning in federated settings faces a critical trade-off between customization and participant trust. Existing approaches typically rely on centralized coordinators or trusted peer groups, limiting their applicability in open, trust-averse environments. While recent decentralized methods explore anonymous knowledge sharing, they often lack global scalability and robust mechanisms against malicious peers. To bridge this gap, we propose TPFed, a \textit{Trust-free Personalized Decentralized Federated Learning} framework. TPFed replaces central aggregators with a blockchain-based bulletin board, enabling participants to dynamically select global communication partners based on Locality-Sensitive Hashing (LSH) and peer ranking. Crucially, we introduce an ``all-in-one'' knowledge distillation protocol that simultaneously handles knowledge transfer, model quality evaluation, and similarity verification via a public reference dataset. This design ensures secure, globally personalized collaboration without exposing local models or data. Extensive experiments demonstrate that TPFed significantly outperforms traditional federated baselines in both learning accuracy and system robustness against adversarial attacks.
Authors: Xu Han, Yuancheng Sun, Kai Chen, Yuxuan Ren, Kang Liu, Qiwei Ye
Abstract: Coarse-grained (CG) molecular dynamics simulations enable efficient exploration of protein conformational ensembles. However, reconstructing atomic details from CG structures (backmapping) remains a challenging problem. Current approaches face an inherent trade-off between maintaining atomistic accuracy and exploring diverse conformations, often necessitating complex constraint handling or extensive refinement steps. To address these challenges, we introduce a novel two-stage framework, named CODLAD (COnstraint Decoupled LAtent Diffusion). This framework first compresses atomic structures into discrete latent representations, explicitly embedding structural constraints, thereby decoupling constraint handling from generation. Subsequently, it performs efficient denoising diffusion in this latent space to produce structurally valid and diverse all-atom conformations. Comprehensive evaluations on diverse protein datasets demonstrate that CODLAD achieves state-of-the-art performance in atomistic accuracy, conformational diversity, and computational efficiency while exhibiting strong generalization across different protein systems. Code is available at https://github.com/xiaoxiaokuye/CODLAD.
Authors: Zhaoxian Wu, Quan Xiao, Tayfun Gokmen, Hsinyu Tsai, Kaoutar El Maghraoui, Tianyi Chen
Abstract: Aiming to accelerate the training of large deep neural networks (DNN) in an energy-efficient way, analog in-memory computing (AIMC) emerges as a solution with immense potential. AIMC accelerator keeps model weights in memory without moving them from memory to processors during training, reducing overhead dramatically. Despite its efficiency, scaling up AIMC systems presents significant challenges. Since weight copying is expensive and inaccurate, data parallelism is less efficient on AIMC accelerators. It necessitates the exploration of pipeline parallelism, particularly asynchronous pipeline parallelism, which utilizes all available accelerators during the training process. This paper examines the convergence theory of stochastic gradient descent on AIMC hardware with an asynchronous pipeline (Analog-SGD-AP). Although there is empirical exploration of AIMC accelerators, the theoretical understanding of how analog hardware imperfections in weight updates affect the training of multi-layer DNN models remains underexplored. Furthermore, the asynchronous pipeline parallelism results in stale weights issues, which render the update signals no longer valid gradients. To close the gap, this paper investigates the convergence properties of Analog-SGD-AP on multi-layer DNN training. We show that the Analog-SGD-AP converges with iteration complexity $O(\varepsilon^{-2}+\varepsilon^{-1})$ despite the aforementioned issues, which matches the complexities of digital SGD and Analog SGD with synchronous pipeline, except the non-dominant term $O(\varepsilon^{-1})$. It implies that AIMC training benefits from asynchronous pipelining almost for free compared with the synchronous pipeline by overlapping computation.
Authors: Amartya Hatua, Trung T. Nguyen, Filip Cano, Andrew H. Sung
Abstract: Modern computer systems store vast amounts of personal data, enabling advances in AI and ML but risking user privacy and trust. For privacy reasons, it is sometimes desired for an ML model to forget part of the data it was trained on. In this paper, we introduce a novel unlearning approach based on Forgetting Neural Networks (FNNs), a neuroscience-inspired architecture that explicitly encodes forgetting through multiplicative decay factors. While FNNs had previously been studied as a theoretical construct, we provide the first concrete implementation and demonstrate their effectiveness for targeted unlearning. We propose several variants with per-neuron forgetting factors, including rank-based assignments guided by activation levels, and evaluate them on MNIST and Fashion-MNIST benchmarks. Our method systematically removes information associated with forget sets while preserving performance on retained data. Membership inference attacks confirm the effectiveness of FNN-based unlearning in erasing information about the training data from the neural network. These results establish FNNs as a promising foundation for efficient and interpretable unlearning.
Authors: Yufan Zheng, Wei Jiang, Tong Chen, Alexander Zhou, Nguyen Quoc Viet Hung, Choujun Zhan, Hongzhi Yin
Abstract: Among various spatio-temporal prediction tasks, epidemic forecasting plays a critical role in public health management. Recent studies have demonstrated the strong potential of spatio-temporal graph neural networks (STGNNs) in extracting heterogeneous spatio-temporal patterns for epidemic forecasting. However, most of these methods bear an over-simplified assumption that two locations (e.g., cities) with similar observed features in previous time steps will develop similar infection numbers in the future. In fact, for any epidemic disease, there exists strong heterogeneity of its intrinsic evolution mechanisms across geolocation and time, which can eventually lead to diverged infection numbers in two ``similar'' locations. However, such mechanistic heterogeneity is non-trivial to be captured due to the existence of numerous influencing factors like medical resource accessibility, virus mutations, mobility patterns, etc., most of which are spatio-temporal yet unreachable or even unobservable. To address this challenge, we propose a Heterogeneous Epidemic-Aware Transmission Graph Neural Network (HeatGNN), a novel epidemic forecasting framework. By binding the epidemiology mechanistic model into a GNN, HeatGNN learns epidemiology-informed location embeddings of different locations that reflect their own transmission mechanisms over time. With the time-varying mechanistic affinity graphs computed with the epidemiology-informed location embeddings, a heterogeneous transmission graph network is designed to encode the mechanistic heterogeneity among locations, providing additional predictive signals to facilitate accurate forecasting. Experiments on three benchmark datasets have revealed that HeatGNN outperforms various strong baselines. Moreover, our efficiency analysis verifies the real-world practicality of HeatGNN on datasets of different sizes.
Authors: Xiao Zhu, Srinivasan S. Iyengar
Abstract: Computing high-dimensional potential energy surfaces for molecular systems and materials is considered to be a great challenge in computational chemistry with potential impact in a range of areas including the fundamental prediction of reaction rates. In this paper, we design and discuss an algorithm that has similarities to large language models in generative AI and natural language processing. Specifically, we represent a molecular system as a graph which contains a set of nodes, edges, faces, etc. Interactions between these sets, which represent molecular subsystems in our case, are used to construct the potential energy surface for a reasonably sized chemical system with 51 nuclear dimensions. For this purpose, a family of neural networks that pertain to the graph-theoretically obtained subsystems get the job done for this 51 nuclear dimensional system. We then ask if this same family of lower-dimensional graph-based neural networks can be transformed to provide accurate predictions for a 186-dimensional potential energy surface. We find that our algorithm does provide accurate results for this larger-dimensional problem with sub-kcal/mol accuracy for the higher-dimensional potential energy surface problem. Indeed, as a result of these developments, here we produce the first efforts towards a full-dimensional potential energy surface for the protonated 21-water cluster (186 nuclear dimensions) at CCSD level accuracy.
Authors: Sicheng Zhu, Brandon Amos, Yuandong Tian, Chuan Guo, Ivan Evtimov
Abstract: Many jailbreak attacks on large language models (LLMs) rely on a common objective: making the model respond with the prefix ``Sure, here is (harmful request)''. While straightforward, this objective has two limitations: limited control over model behaviors, yielding incomplete or unrealistic jailbroken responses, and a rigid format that hinders optimization. We introduce AdvPrefix, a plug-and-play prefix-forcing objective that selects one or more model-dependent prefixes by combining two criteria: high prefilling attack success rates and low negative log-likelihood. AdvPrefix integrates seamlessly into existing jailbreak attacks to mitigate the previous limitations for free. For example, replacing GCG's default prefixes on Llama-3 improves nuanced attack success rates from 14% to 80%, revealing that current safety alignment fails to generalize to new prefixes. Code and selected prefixes are released at github.com/facebookresearch/jailbreak-objectives.
Authors: Michael Bezick, Blake A. Wilson, Vaishnavi Iyer, Yuheng Chen, Vladimir M. Shalaev, Sabre Kais, Alexander V. Kildishev, Alexandra Boltasseva, Brad Lackey
Abstract: PearSAN is a machine learning-assisted optimization algorithm applicable to inverse design problems with large design spaces, where traditional optimizers struggle. The algorithm leverages the latent space of a generative model for rapid sampling and employs a Pearson correlated surrogate model to predict the figure of merit of the true design metric. As a showcase example, PearSAN is applied to thermophotovoltaic (TPV) metasurface design by matching the working bands between a thermal radiator and a photovoltaic cell. PearSAN can work with any pretrained generative model with a discretized latent space, making it easy to integrate with VQ-VAEs and binary autoencoders. Its novel Pearson correlational loss can be used as both a latent regularization method, similar to batch and layer normalization, and as a surrogate training loss. We compare both to previous energy matching losses, which are shown to enforce poor regularization and performance, even with upgraded affine parameters. PearSAN achieves a state-of-the-art maximum design efficiency of 97%, and is at least an order of magnitude faster than previous methods, with an improved maximum figure-of-merit gain.
Authors: Arseniy Andreyev, Pierfrancesco Beneventano
Abstract: Recent findings by Cohen et al., 2021, demonstrate that when training neural networks using full-batch gradient descent with a step size of $\eta$, the largest eigenvalue $\lambda_{\max}$ of the full-batch Hessian consistently stabilizes around $2/\eta$. These results have significant implications for convergence and generalization. This, however, is not the case for mini-batch optimization algorithms, limiting the broader applicabilityof the consequences of these findings. We show mini-batch Stochastic Gradient Descent (SGD) trains in a different regime we term Edge of Stochastic Stability (EoSS). In this regime, what stabilizes at $2/\eta$ is Batch Sharpness: the expected directional curvature of mini-batch Hessians along their corresponding stochastic gradients. As a consequence $\lambda_{\max}$ -- which is generally smaller than Batch Sharpness -- is suppressed, aligning with the long-standing empirical observation that smaller batches and larger step sizes favor flatter minima. We further discuss implications for mathematical modeling of SGD trajectories.
Authors: Christopher Burger
Abstract: Adversarial attacks challenge the reliability of Explainable AI (XAI) by altering explanations while the model's output remains unchanged. The success of these attacks on text-based XAI is often judged using standard information retrieval metrics. We argue these measures are poorly suited in the evaluation of trustworthiness, as they treat all word perturbations equally while ignoring synonymity, which can misrepresent an attack's true impact. To address this, we apply synonymity weighting, a method that amends these measures by incorporating the semantic similarity of perturbed words. This produces more accurate vulnerability assessments and provides an important tool for assessing the robustness of AI systems. Our approach prevents the overestimation of attack success, leading to a more faithful understanding of an XAI system's true resilience against adversarial manipulation.
Authors: Xiaohong Yang, Minghui Liwang, Xianbin Wang, Zhipeng Cheng, Seyyedali Hosseinalipour, Huaiyu Dai, Zhenzhen Jiao
Abstract: The rapid growth of AI-enabled Internet of Vehicles (IoV) calls for efficient machine learning (ML) solutions that can handle high vehicular mobility and decentralized data. This has motivated the emergence of Hierarchical Federated Learning over vehicle-edge-cloud architectures (VEC-HFL). Nevertheless, one aspect which is underexplored in the literature on VEC-HFL is that vehicles often need to execute multiple ML tasks simultaneously, where this multi-model training environment introduces crucial challenges. First, improper aggregation rules can lead to model obsolescence and prolonged training times. Second, vehicular mobility may result in inefficient data utilization by preventing the vehicles from returning their models to the network edge. Third, achieving a balanced resource allocation across diverse tasks becomes of paramount importance as it majorly affects the effectiveness of collaborative training. We take one of the first steps towards addressing these challenges via proposing a framework for multi-model training in dynamic VEC-HFL with the goal of minimizing global training latency while ensuring balanced training across various tasks-a problem that turns out to be NP-hard. To facilitate timely model training, we introduce a hybrid synchronous-asynchronous aggregation rule. Building on this, we present a novel method called Hybrid Evolutionary And gReedy allocaTion (HEART). The framework operates in two stages: first, it achieves balanced task scheduling through a hybrid heuristic approach that combines improved Particle Swarm Optimization (PSO) and Genetic Algorithms (GA); second, it employs a low-complexity greedy algorithm to determine the training priority of assigned tasks on vehicles. Experiments on real-world datasets demonstrate the superiority of HEART over existing methods.
Authors: Akash Kumar
Abstract: The success of deep networks is crucially attributed to their ability to capture latent features within a representation space. In this work, we investigate whether the underlying learned features of a model can be efficiently retrieved through feedback from an agent, such as a large language model (LLM), in the form of relative \tt{triplet comparisons}. These features may represent various constructs, including dictionaries in LLMs or a covariance matrix of Mahalanobis distances. We analyze the feedback complexity associated with learning a feature matrix in sparse settings. Our results establish tight bounds when the agent is permitted to construct activations and demonstrate strong upper bounds in sparse scenarios when the agent's feedback is limited to distributional information. We validate our theoretical findings through experiments on two distinct applications: feature recovery from Recursive Feature Machines and dictionary extraction from sparse autoencoders trained on Large Language Models.
Authors: Shizhou Xu, Thomas Strohmer
Abstract: How can we effectively remove or ''unlearn'' undesirable information, such as specific features or the influence of individual data points, from a learning outcome while minimizing utility loss and ensuring rigorous guarantees? We introduce a unified mathematical framework based on information-theoretic regularization to address both data point unlearning and feature unlearning. For data point unlearning, we introduce the $\textit{Marginal Unlearning Principle}$, an auditable and provable framework inspired by memory suppression studies in neuroscience. Moreover, we provide formal information-theoretic unlearning definition based on the proposed principle, named marginal unlearning, and provable guarantees on sufficiency and necessity of marginal unlearning to the existing approximate unlearning definitions. We then show the proposed framework provide natural solution to the marginal unlearning problems. For feature unlearning, the framework applies to deep learning with arbitrary training objectives. By combining flexibility in learning objectives with simplicity in regularization design, our approach is highly adaptable and practical for a wide range of machine learning and AI applications. From a mathematical perspective, we provide an unified analytic solution to the optimal feature unlearning problem with a variety of information-theoretic training objectives. Our theoretical analysis reveals intriguing connections between machine unlearning, information theory, optimal transport, and extremal sigma algebras. Numerical simulations support our theoretical finding.
Authors: Corinna Cortes, Anqi Mao, Mehryar Mohri, Yutao Zhong
Abstract: Class imbalance remains a major challenge in machine learning, especially in multi-class problems with long-tailed distributions. Existing methods, such as data resampling, cost-sensitive techniques, and logistic loss modifications, though popular and often effective, lack solid theoretical foundations. As an example, we demonstrate that cost-sensitive methods are not Bayes-consistent. This paper introduces a novel theoretical framework for analyzing generalization in imbalanced classification. We propose a new class-imbalanced margin loss function for both binary and multi-class settings, prove its strong $H$-consistency, and derive corresponding learning guarantees based on empirical loss and a new notion of class-sensitive Rademacher complexity. Leveraging these theoretical results, we devise novel and general learning algorithms, IMMAX (Imbalanced Margin Maximization), which incorporate confidence margins and are applicable to various hypothesis sets. While our focus is theoretical, we also present extensive empirical results demonstrating the effectiveness of our algorithms compared to existing baselines.
Authors: Maximilian Beck, Korbinian P\"oppel, Phillip Lippe, Sepp Hochreiter
Abstract: Linear RNNs with gating recently demonstrated competitive performance compared to Transformers in language modeling. Although their linear compute scaling in sequence length offers theoretical runtime advantages over Transformers, realizing these benefits in practice requires optimized custom kernels, as Transformers rely on the highly efficient Flash Attention kernels (Dao, 2024). Leveraging the chunkwise-parallel formulation of linear RNNs, Flash Linear Attention (FLA) (Yang & Zhang, 2024) shows that linear RNN kernels are faster than Flash Attention, by parallelizing over chunks of the input sequence. However, since the chunk size of FLA is limited, many intermediate states must be materialized in GPU memory. This leads to low arithmetic intensity and causes high memory consumption and IO cost, especially for long-context pre-training. In this work, we present Tiled Flash Linear Attention (TFLA), a novel kernel algorithm for linear RNNs, that enables arbitrary large chunk sizes and high arithmetic intensity by introducing an additional level of sequence parallelization within each chunk. First, we apply TFLA to the xLSTM with matrix memory, the mLSTM (Beck et al., 2024). Second, we propose an mLSTM variant with sigmoid input gate and reduced computation for even faster kernel runtimes at equal language modeling performance. In our speed benchmarks, we show that our new mLSTM kernels based on TFLA outperform highly optimized Flash Attention, Linear Attention and Mamba kernels, setting a new state of the art for efficient long-context sequence modeling primitives.
Authors: Enjun Du, Siyi Liu, Yongqi Zhang
Abstract: Knowledge graph reasoning in the fully-inductive setting, where both entities and relations at test time are unseen during training, remains an open challenge. In this work, we introduce GraphOracle, a novel framework that achieves robust fully-inductive reasoning by transforming each knowledge graph into a Relation-Dependency Graph (RDG). The RDG encodes directed precedence links between relations, capturing essential compositional patterns while drastically reducing graph density. Conditioned on a query relation, a multi-head attention mechanism propagates information over the RDG to produce context-aware relation embeddings. These embeddings then guide a second GNN to perform inductive message passing over the original knowledge graph, enabling prediction on entirely new entities and relations. Comprehensive experiments on 60 benchmarks demonstrate that GraphOracle outperforms prior methods by up to 25% in fully-inductive and 28% in cross-domain scenarios. Our analysis further confirms that the compact RDG structure and attention-based propagation are key to efficient and accurate generalization.
Authors: Vanessa Kosoy
Abstract: We propose a new variant of online learning that we call "ambiguous online learning". In this setting, the learner is allowed to produce multiple predicted labels. Such an "ambiguous prediction" is considered correct when at least one of the labels is correct, and none of the labels are "predictably wrong". The definition of "predictably wrong" comes from a hypothesis class in which hypotheses are also multi-valued. Thus, a prediction is "predictably wrong" if it's not allowed by the (unknown) true hypothesis. In particular, this setting is natural in the context of multivalued dynamical systems, recommendation algorithms and lossless compression. It is also strongly related to so-called "apple tasting". We show that in this setting, there is a trichotomy of mistake bounds: up to logarithmic factors, any hypothesis class has an optimal mistake bound of either Theta(1), Theta(sqrt(N)) or N.
Authors: Anqi Mao, Mehryar Mohri, Yutao Zhong
Abstract: The problem of learning to defer with multiple experts consists of optimally assigning input instances to experts, balancing the trade-off between their accuracy and computational cost. This is a critical challenge in natural language generation, but also in other fields such as image processing, and medical diagnostics. Recent studies have proposed surrogate loss functions to optimize deferral, but challenges remain in ensuring their consistency properties. This paper introduces novel surrogate loss functions and efficient algorithms with strong theoretical learning guarantees. We address open questions regarding realizable $H$-consistency, $H$-consistency bounds, and Bayes-consistency for both single-stage (jointly learning predictor and deferral function) and two-stage (learning only the deferral function with a fixed expert) learning scenarios. For single-stage deferral, we introduce a family of new realizable $H$-consistent surrogate losses and further prove $H$-consistency for a selected member. For two-stage deferral, we derive new surrogate losses that achieve realizable $H$-consistency, $H$-consistency bounds, and Bayes-consistency for the two-expert scenario and, under natural assumptions, multiple-expert scenario. Additionally, we provide enhanced theoretical guarantees under low-noise assumptions for both scenarios. Finally, we report the results of experiments using our proposed surrogate losses, comparing their performance against existing baselines.
Authors: Aho Yapi, Pierre Latouche, Arnaud Guillin, Yan Bailly
Abstract: We propose a model-agnostic framework for short-term occupational accident forecasting that leverages safety inspections and models accident occurrences as binary time series. The approach generates daily predictions, which are then aggregated into weekly safety assessments for better decision making. To ensure the reliability and operational applicability of the forecasts, we apply a sliding-window cross-validation procedure specifically designed for time series data, combined with an evaluation based on aggregated period-level metrics. Several machine learning algorithms, including logistic regression, tree-based models, and neural networks, are trained and systematically compared within this framework. Across all tested algorithms, the proposed framework reliably identifies upcoming high-risk periods and delivers robust period-level performance, demonstrating that converting safety inspections into binary time series yields actionable, short-term risk signals. The proposed methodology converts routine safety inspection data into clear weekly and daily risk scores, detecting the periods when accidents are most likely to occur. Decision-makers can integrate these scores into their planning tools to classify inspection priorities, schedule targeted interventions, and funnel resources to the sites or shifts classified as highest risk, stepping in before incidents occur and getting the greatest return on safety investments.
Authors: Keyon Vafa, Peter G. Chang, Ashesh Rambachan, Sendhil Mullainathan
Abstract: Foundation models are premised on the idea that sequence prediction can uncover deeper domain understanding, much like how Kepler's predictions of planetary motion later led to the discovery of Newtonian mechanics. However, evaluating whether these models truly capture deeper structure remains a challenge. We develop a technique for evaluating foundation models that examines how they adapt to synthetic datasets generated from some postulated world model. Our technique measures whether the foundation model's inductive bias aligns with the world model, and so we refer to it as an inductive bias probe. Across multiple domains, we find that foundation models can excel at their training tasks yet fail to develop inductive biases towards the underlying world model when adapted to new tasks. We particularly find that foundation models trained on orbital trajectories consistently fail to apply Newtonian mechanics when adapted to new physics tasks. Further analysis reveals that these models behave as if they develop task-specific heuristics that fail to generalize.
Authors: James Chapman, Kedar Karhadkar, Guido Montufar
Abstract: Deep reinforcement learning (DRL) has achieved remarkable success across multiple domains, including competitive games, natural language processing, and robotics. Despite these advancements, policies trained via DRL often struggle to generalize to evaluation environments with different parameters. This challenge is typically addressed by training with multiple contexts and/or by leveraging additional structure in the problem. However, obtaining sufficient training data across diverse contexts can be impractical in real-world applications. In this work, we consider contextual Markov decision processes (CMDPs) with transition and reward functions that exhibit regularity in context parameters. We introduce the context-enhanced Bellman equation (CEBE) to improve generalization when training on a single context. We prove both analytically and empirically that the CEBE yields a first-order approximation to the Q-function trained across multiple contexts. We then derive context sample enhancement (CSE) as an efficient data augmentation method for approximating the CEBE in deterministic control environments. We numerically validate the performance of CSE in simulation environments, showcasing its potential to improve generalization in DRL.
Authors: Youssef Tawfilis, Hossam Amer, Minar El-Aasser, Tallal Elshabrawy
Abstract: Federated Learning has gained increasing attention for its ability to enable multiple nodes to collaboratively train machine learning models without sharing their raw data. At the same time, Generative AI -- particularly Generative Adversarial Networks (GANs) -- have achieved remarkable success across a wide range of domains, such as healthcare, security, and Image Generation. However, training generative models typically requires large datasets and significant computational resources, which are often unavailable in real-world settings. Acquiring such resources can be costly and inefficient, especially when many underutilized devices -- such as IoT devices and edge devices -- with varying capabilities remain idle. Moreover, obtaining large datasets is challenging due to privacy concerns and copyright restrictions, as most devices are unwilling to share their data. To address these challenges, we propose a novel approach for decentralized GAN training that enables the utilization of distributed data and underutilized, low-capability devices while not sharing data in its raw form. Our approach is designed to tackle key challenges in decentralized environments, combining KLD-weighted Clustered Federated Learning to address the issues of data heterogeneity and multi-domain datasets, with Heterogeneous U-Shaped split learning to tackle the challenge of device heterogeneity under strict data sharing constraints -- ensuring that no labels or raw data, whether real or synthetic, are ever shared between nodes. Experiments show that our approach demonstrates significant improvements across key metrics, where it achieves an average 10% boost in classification metrics (up to 60% in multi-domain non-IID settings), 1.1x -- 3x higher image generation scores for the MNIST family datasets, and 2x -- 70x lower FID scores for higher resolution datasets. Find our code at https://github.com/youssefga28/HuSCF-GAN.
Authors: He Xiao, Qingyao Yang, Dirui Xie, Wendong Xu, Zunhai Su, Runming yang, Wenyong Zhou, Haobo Liu, Zhengwu Liu, Ngai Wong
Abstract: Large language models with billions of parameters are often over-provisioned: many layers contribute little unique information yet dominate the memory and energy footprint during inference. We present LieQ Layer-wise information effectiveness Quantization, a hardware-native, metric-driven post-training quantization framework that addresses the critical challenge of maintaining accuracy in sub-8B models, model parameters less than 8B, under extreme low-bit compression. LieQ keeps uniform bit-width within each layer while mixing precision across layers, preserving standard multiplication kernels and avoiding irregular memory access, codebooks, or irregular formats at inference time. Our method uncovers a strong correlation between layer-wise functional saliency and representational compactness, revealing that layers with higher training-induced energy concentration are functionally irreplaceable. Leveraging this insight, we propose a purely geometry-driven sensitivity proxy that enables automatic bit-width allocation under a target average-bit budget without expensive gradient updates or inference-based perplexity probing. At sub 2-bit, LieQ consistently reduces the large accuracy gap typically observed for naive 2-bit baselines on Qwen3 and LLaMA3.x families, while retaining standard-kernel efficiency. These properties make LieQ a practical path toward deploying small language models on resource-constrained edge devices. Code will available here: https://github.com/HeXiao-55/LieQ-official.git.
Authors: Zichun Ye, Runqi Wang, Xutong Liu, Shuai Li
Abstract: The combinatorial multi-armed bandit (CMAB) is a cornerstone of sequential decision-making framework, dominated by two algorithmic families: UCB-based and adversarial methods such as follow the regularized leader (FTRL) and online mirror descent (OMD). However, prominent UCB-based approaches like CUCB suffer from additional regret factor $\log T$ that is detrimental over long horizons, while adversarial methods such as EXP3.M and HYBRID impose significant computational overhead. To resolve this trade-off, we introduce the Combinatorial Minimax Optimal Strategy in the Stochastic setting (CMOSS). CMOSS is a computationally efficient algorithm that achieves an instance-independent regret of $O\big( (\log k)\sqrt{kmT}\big )$ when $k\leq \frac{m}{2}$ and $O\big((m-k)\sqrt{\log k\log(m-k)T}\big )$ when $k>\frac{m}{2}$ under semi-bandit feedback, where $m$ is the number of arms and $k$ is the maximum cardinality of a feasible action. Crucially, this result eliminates the dependency on $\log T$ and matches the established lower bounds of $\Omega\big(\sqrt{kmT}\big)$ when $k\leq \frac{m}{2}$ and $\Omega\big((m-k)\sqrt{\log (\frac{m}{m-k}) T}\big)$ when $k>\frac{m}{2}$ up to logarithmic terms of $k$ and $m$. We then extend our analysis to show that CMOSS is also applicable to cascading feedback. Experiments on synthetic and real-world datasets validate that CMOSS consistently outperforms benchmark algorithms in both regret and runtime efficiency.
Authors: Dong Xue, Ziyao Shao, Zhaoyang Duan, Fangzhou Liu, Bing Li, Zhongheng Zhang
Abstract: Large multimodal models (LMMs) have demonstrated significant potential in providing innovative solutions for various biomedical tasks, including pathology analysis, radiology report generation, and biomedical assistance. However, the existing multimodal biomedical AI is typically based on foundation LLMs, thus hindering the understanding of intricate medical concepts with limited medical training data. Moreover, recent LLaVA-induced medical LMMs struggle to effectively capture the intricate relationship between the texts and the images. Therefore, we introduce Doctor Sun, a large multimodal generative model specialized in medicine, developed to encode, integrate, and interpret diverse biomedical data modalities such as text and images. In particular, Doctor Sun integrates a pre-trained vision encoder with a medical LLM and conducts two-stage training on various medical datasets, focusing on feature alignment and instruction tuning. Moreover, we release SunMed-VL, a wide-range bilingual medical multimodal dataset, along with all associated models, code, and resources, to freely support the advancement of biomedical multimodal research.
Authors: Elon Litman
Abstract: The scaled-dot-product attention (SDPA) mechanism is a core component of modern deep learning, but its mathematical form is often motivated by heuristics. This work provides a first-principles justification for SDPA. We first show that the attention forward pass is the exact solution to a degenerate, one-sided Entropic Optimal Transport (EOT) problem, which seeks a distribution that maximizes similarity while being maximally entropic. This optimization perspective has a direct consequence for the backward pass. We prove that the standard gradient computed via backpropagation is mathematically identical to an advantage-based policy gradient, a variance-reduced update rule from reinforcement learning. Crucially, we demonstrate that the EOT formulation of the forward pass induces a specific information geometry on the space of attention distributions. It is this geometry, characterized by the Fisher Information Matrix, that dictates the precise form of the learning gradient, revealing the advantage-based update as a natural consequence of the optimization problem being solved. This unified view reveals SDPA as a principled mechanism where the forward pass performs optimal inference and the backward pass implements a rational, manifold-aware learning update.
Authors: Quercus Hernandez, Max Win, Thomas C. O'Connor, Paulo E. Arratia, Nathaniel Trask
Abstract: Multiscale systems are ubiquitous in science and technology, but are notoriously challenging to simulate as short spatiotemporal scales must be appropriately linked to emergent bulk physics. When expensive high-dimensional dynamical systems are coarse-grained into low-dimensional models, the entropic loss of information leads to emergent physics which are dissipative, history-dependent, and stochastic. To machine learn coarse-grained dynamics from time-series observations of particle trajectories, we propose a framework using the metriplectic bracket formalism that preserves these properties by construction; most notably, the framework guarantees discrete notions of the first and second laws of thermodynamics, conservation of momentum, and a discrete fluctuation-dissipation balance crucial for capturing non-equilibrium statistics. We introduce the mathematical framework abstractly before specializing to a particle discretization. As labels are generally unavailable for entropic state variables, we introduce a novel self-supervised learning strategy to identify emergent structural variables. We validate the method on benchmark systems and demonstrate its utility on two challenging examples: (1) coarse-graining star polymers at challenging levels of coarse-graining while preserving non-equilibrium statistics, and (2) learning models from high-speed video of colloidal suspensions that capture coupling between local rearrangement events and emergent stochastic dynamics. We provide open-source implementations in both PyTorch and LAMMPS, enabling large-scale inference and extensibility to diverse particle-based systems.
Authors: Chao Yu, Yuanqing Wang, Zhen Guo, Hao Lin, Si Xu, Hongzhi Zang, Quanlu Zhang, Yongji Wu, Chunyang Zhu, Junhao Hu, Zixiao Huang, Mingjie Wei, Yuqing Xie, Ke Yang, Bo Dai, Zhexuan Xu, Jiakun Du, Xiangyuan Wang, Xu Fu, Letong Shi, Zhihao Liu, Kang Chen, Weilin Liu, Gang Liu, Boxun Li, Jianlei Yang, Zhi Yang, Guohao Dai, Yu Wang
Abstract: Reinforcement learning (RL) has demonstrated immense potential in advancing artificial general intelligence, agentic intelligence, and embodied intelligence. However, the inherent heterogeneity and dynamicity of RL workflows often lead to low hardware utilization and slow training on existing systems. In this paper, we present RLinf, a high-performance RL training system based on our key observation that the major roadblock to efficient RL training lies in system flexibility. To maximize flexibility and efficiency, RLinf is built atop a novel RL system design paradigm called macro-to-micro flow transformation (M2Flow), which automatically breaks down high-level, easy-to-compose RL workflows at both the temporal and spatial dimensions, and recomposes them into optimized execution flows. Supported by RLinf worker's adaptive communication capability, we devise context switching and elastic pipelining to realize M2Flow transformation, and a profiling-guided scheduling policy to generate optimal execution plans. Extensive evaluations on both reasoning RL and embodied RL tasks demonstrate that RLinf consistently outperforms state-of-the-art systems, achieving $1.07\times-2.43\times$ speedup in end-to-end training throughput.
Authors: Kristina P. Sinaga
Abstract: This paper proposes a personalized federated learning framework integrating heat-kernel enhanced tensorized multi-view fuzzy c-means clustering with tensor decomposition techniques. The approach combines heat-kernel coefficients adapted from quantum field theory with PARAFAC2 and Tucker decomposition to transform distance metrics and efficiently represent high-dimensional multi-view structures. Two main algorithms, FedHK-PARAFAC2 and FedHK-Tucker, are developed to extract shared and view-specific features while preserving inter-view relationships. The framework addresses data heterogeneity, privacy preservation, and communication efficiency challenges in federated learning environments. Theoretical analysis provides convergence guarantees, privacy bounds, and complexity analysis. The integration of heat-kernel methods with tensor decomposition in a federated setting offers a novel approach for effective multi-view data analysis while ensuring data privacy.
Authors: Xiang Zhang, Kun Wei, Xu Yang, Chenghao Xu, Su Yan, Cheng Deng
Abstract: As Large Language Models (LLMs) become increasingly prevalent, their security vulnerabilities have already drawn attention. Machine unlearning is introduced to seek to mitigate these risks by removing the influence of undesirable data. However, existing methods not only rely on the retained dataset to preserve model utility, but also suffer from cumulative catastrophic utility loss under continuous unlearning requests. To solve this dilemma, we propose a novel method, called Rotation Control Unlearning (RCU), which leverages the rotational salience weight of RCU to quantify and control the unlearning degree in the continuous unlearning process. The skew symmetric loss is designed to construct the existence of the cognitive rotation space, where the changes of rotational angle can simulate the continuous unlearning process. Furthermore, we design an orthogonal rotation axes regularization to enforce mutually perpendicular rotation directions for continuous unlearning requests, effectively minimizing interference and addressing cumulative catastrophic utility loss. Experiments on multiple datasets confirm that our method without retained dataset achieves SOTA performance.
Authors: Kevin Rojas, Jiahe Lin, Kashif Rasul, Anderson Schneider, Yuriy Nevmyvaka, Molei Tao, Wei Deng
Abstract: Diffusion language models (DLMs) enable parallel, order-agnostic generation with iterative refinement, offering a flexible alternative to autoregressive large language models (LLMs). However, adapting reinforcement learning (RL) fine-tuning to DLMs remains an open challenge because of the intractable likelihood. Pioneering work such as diffu-GRPO estimated token-level likelihoods via one-step unmasking. While computationally efficient, this approach is severely biased. A more principled foundation lies in sequence-level likelihoods, where the evidence lower bound (ELBO) serves as a surrogate. Yet, despite this clean mathematical connection, ELBO-based methods have seen limited adoption due to the prohibitive cost of likelihood evaluation. In this work, we revisit ELBO estimation and disentangle its sources of variance. This decomposition motivates reducing variance through fast, deterministic integral approximations along a few pivotal dimensions. Building on this insight, we introduce Group Diffusion Policy Optimization (GDPO), a new RL algorithm tailored for DLMs. GDPO leverages simple yet effective Semi-deterministic Monte Carlo schemes to mitigate the variance explosion of ELBO estimators under vanilla double Monte Carlo sampling, yielding a provably lower-variance estimator under tight evaluation budgets. Empirically, GDPO achieves consistent gains over pretrained checkpoints and outperforms diffu-GRPO, one of the state-of-the-art baselines, on the majority of math, reasoning, and coding benchmarks.
Authors: Jing He, Hua Jiang, Cheng Li, Siqian Xin, Shuzhen Yang
Abstract: This work, based on Random Matrix Theory (RMT), introduces a novel early-stopping strategy for Transformer training dynamics. Utilizing the Power Law (PL) fit to tansformer attention matrices as a probe, we demarcate training into three stages: structural exploration, heavy-tailed structure stabilization, and convergence saturation. Empirically, we observe that the spectral density of the shallow self-attention matrix $V$ consistently evolves into a heavy-tailed distribution. Crucially, we propose two consistent and validation-set-free criteria: a quantitative metric for heavy-tailed dynamics and a novel spectral signature indicative of convergence. The strong alignment between these criteria highlights the utility of RMT for monitoring and diagnosing the progression of Transformer model training.
Authors: Hongliang Lu, Yuhang Wen, Pengyu Cheng, Ruijin Ding, Jiaqi Guo, Haotian Xu, Chutian Wang, Haonan Chen, Xiaoxi Jiang, Guanjun Jiang
Abstract: Reinforcement learning with verifiable rewards (RLVR) has become the mainstream technique for training LLM agents. However, RLVR highly depends on well-crafted task queries and corresponding ground-truth answers to provide accurate rewards, which requires significant human effort and hinders the scaling of RL processes, especially in agentic scenarios. Although a few recent works explore task synthesis methods, the difficulty of generated agentic tasks can hardly be controlled to provide effective RL training advantages. To achieve agentic RLVR with higher scalability, we explore self-play training for deep search agents, in which the learning LLM utilizes multi-turn search engine calling and acts simultaneously as both a task proposer and a problem solver. The task proposer aims to generate deep search queries with well-defined ground-truth answers and increasing task difficulty. The problem solver tries to handle the generated search queries and output the correct answer predictions. To ensure that each generated search query has accurate ground truth, we collect all the searching results from the proposer's trajectory as external knowledge, then conduct retrieval-augmentation generation (RAG) to test whether the proposed query can be correctly answered with all necessary search documents provided. In this search self-play (SSP) game, the proposer and the solver co-evolve their agent capabilities through both competition and cooperation. With substantial experimental results, we find that SSP can significantly improve search agents' performance uniformly on various benchmarks without any supervision under both from-scratch and continuous RL training setups. The code is at https://github.com/Qwen-Applications/SSP.
Authors: Manuel Iglesias-Alonso, Felix Schur, Julius von K\"ugelgen, Jonas Peters
Abstract: We consider the problem of estimating a causal effect in a multi-domain setting. The causal effect of interest is confounded by an unobserved confounder and can change between the different domains. We assume that we have access to a proxy of the hidden confounder and that all variables are discrete or categorical. We propose methodology to estimate the causal effect in the target domain, where we assume to observe only the proxy variable. Under these conditions, we prove identifiability (even when treatment and response variables are continuous). We introduce two estimation techniques, prove consistency, and derive confidence intervals. The theoretical results are supported by simulation studies and a real-world example studying the causal effect of website rankings on consumer choices.
Authors: Xiaoyu Fan, Lin Guo, Ruizhen Jia, Yang Tian, Zhihao Yang, Boxue Tian
Abstract: Artificial Intelligence (AI)-aided drug discovery is an active research field, yet AI models often exhibit poor accuracy in regression tasks for molecular property prediction, and perform catastrophically poorly for out-of-distribution (OOD) molecules. Here, we present MolRuleLoss, a substructure-substitution-rule-informed framework that improves the accuracy and generalizability of multiple molecular property regression models (MPRMs) such as GEM and UniMol for diverse molecular property prediction tasks. MolRuleLoss incorporates partial derivative constraints for substructure substitution rules (SSRs) into an MPRM's loss function. When using GEM models for predicting lipophilicity, water solubility, and solvation-free energy (using lipophilicity, ESOL, and freeSolv datasets from MoleculeNet), the root mean squared error (RMSE) values with and without MolRuleLoss were 0.587 vs. 0.660, 0.777 vs. 0.798, and 1.252 vs. 1.877, respectively, representing 2.6-33.3% performance improvements. We show that both the number and the quality of SSRs contribute to the magnitude of prediction accuracy gains obtained upon adding MolRuleLoss to an MPRM. MolRuleLoss improved the generalizability of MPRMs for "activity cliff" molecules in a lipophilicity prediction task and improved the generalizability of MPRMs for OOD molecules in a melting point prediction task. In a molecular weight prediction task for OOD molecules, MolRuleLoss reduced the RMSE value of a GEM model from 29.507 to 0.007. We also provide a formal demonstration that the upper bound of the variation for property change of SSRs is positively correlated with an MPRM's error. Together, we show that using the MolRuleLoss framework as a bolt-on boosts the prediction accuracy and generalizability of multiple MPRMs, supporting diverse applications in areas like cheminformatics and AI-aided drug discovery.
Authors: Kabir Khan, Manju Sarkar, Anita Kar, Suresh Ghosh
Abstract: Large generative models (for example, language and diffusion models) enable high-quality text and image synthesis but are hard to train or adapt in cross-device federated settings due to heavy computation and communication and statistical/system heterogeneity. We propose FedGen-Edge, a framework that decouples a frozen, pre-trained global backbone from lightweight client-side adapters and federates only the adapters. Using Low-Rank Adaptation (LoRA) constrains client updates to a compact subspace, which reduces uplink traffic by more than 99 percent versus full-model FedAvg, stabilizes aggregation under non-IID data, and naturally supports personalization because each client can keep a locally tuned adapter. On language modeling (PTB) and image generation (CIFAR-10), FedGen-Edge achieves lower perplexity/FID and faster convergence than strong baselines while retaining a simple FedAvg-style server. A brief ablation shows diminishing returns beyond moderate LoRA rank and a trade-off between local epochs and client drift. FedGen-Edge offers a practical path toward privacy-preserving, resource-aware, and personalized generative AI on heterogeneous edge devices.
Authors: Minghui Chen, Hrad Ghoukasian, Ruinan Jin, Zehua Wang, Sai Praneeth Karimireddy, Xiaoxiao Li
Abstract: Federated Learning (FL) enables decentralized, privacy-preserving model training but struggles to balance global generalization and local personalization due to non-identical data distributions across clients. Personalized Fine-Tuning (PFT), a popular post-hoc solution, fine-tunes the final global model locally but often overfits to skewed client distributions or fails under domain shifts. We propose adapting Linear Probing followed by full Fine-Tuning (LP-FT), a principled centralized strategy for alleviating feature distortion (Kumar et al., 2022), to the FL setting. Through systematic evaluation across seven datasets and six PFT variants, we demonstrate LP-FT's superiority in balancing personalization and generalization. Our analysis uncovers federated feature distortion, a phenomenon where local fine-tuning destabilizes globally learned features, and theoretically characterizes how LP-FT mitigates this via phased parameter updates. We further establish conditions (e.g., partial feature overlap, covariate-concept shift) under which LP-FT outperforms standard fine-tuning, offering actionable guidelines for deploying robust personalization in FL.
Authors: Daniel Furelos-Blanco, Charles Pert, Frederik Kelbel, Alex F. Spies, Alessandra Russo, Michael Dennis
Abstract: Training general agents to follow complex instructions (tasks) in intricate environments (levels) remains a core challenge in reinforcement learning. Random sampling of task-level pairs often produces unsolvable combinations, highlighting the need to co-design tasks and levels. While unsupervised environment design (UED) has proven effective at automatically designing level curricula, prior work has only considered a fixed task. We present ATLAS (Aligning Tasks and Levels for Autocurricula of Specifications), a novel method that generates joint autocurricula over tasks and levels. Our approach builds upon UED to automatically produce solvable yet challenging task-level pairs for policy training. To evaluate ATLAS and drive progress in the field, we introduce an evaluation suite that models tasks as reward machines in Minigrid levels. Experiments demonstrate that ATLAS vastly outperforms random sampling approaches, particularly when sampling solvable pairs is unlikely. We further show that mutations leveraging the structure of both tasks and levels accelerate convergence to performant policies.
Authors: Omar Adalat, Francesco Belardinelli
Abstract: Specifying informative and dense reward functions remains a pivotal challenge in Reinforcement Learning, as it directly affects the efficiency of agent training. In this work, we harness the expressive power of quantitative Linear Temporal Logic on finite traces (($\text{LTL}_f[\mathcal{F}]$)) to synthesize reward monitors that generate a dense stream of rewards for runtime-observable state trajectories. By providing nuanced feedback during training, these monitors guide agents toward optimal behaviour and help mitigate the well-known issue of sparse rewards under long-horizon decision making, which arises under the Boolean semantics dominating the current literature. Our framework is algorithm-agnostic and only relies on a state labelling function, and naturally accommodates specifying non-Markovian properties. Empirical results show that our quantitative monitors consistently subsume and, depending on the environment, outperform Boolean monitors in maximizing a quantitative measure of task completion and in reducing convergence time.
Authors: Kushal Agrawal, Frank Xiao, Guido Bergman, Asa Cooper Stickland
Abstract: The deployment of Large Language Models (LLMs) as tool-using agents causes their alignment training to manifest in new ways. Recent work finds that language models can use tools in ways that contradict the interests or explicit instructions of the user. We study LLM whistleblowing: a subset of this behavior where models disclose suspected misconduct to parties beyond the dialog boundary (e.g., regulatory agencies) without user instruction or knowledge. We introduce an evaluation suite of diverse and realistic staged misconduct scenarios to assess agents for this behavior. Across models and settings, we find that: (1) the frequency of whistleblowing varies widely across model families, (2) increasing the complexity of the task the agent is instructed to complete lowers whistleblowing tendencies, (3) nudging the agent in the system prompt to act morally substantially raises whistleblowing rates, and (4) giving the model more obvious avenues for non-whistleblowing behavior, by providing more tools and a detailed workflow to follow, decreases whistleblowing rates. Additionally, we verify the robustness of our dataset by testing for model evaluation awareness, and find that both black-box methods and probes on model activations show lower evaluation awareness in our settings than in comparable previous work.
Authors: Lei Wang, Zikun Ye, Jinglong Zhao
Abstract: Driven by recent advances in artificial intelligence (AI), a growing literature has demonstrated the potential for using large language models (LLMs) as scalable surrogates to generate human-like responses in many business applications. Two common approaches to improve the performance of LLMs include: fine-tuning, which aligns LLMs more closely with human responses, and rectification, which corrects biases in LLM outputs. In this paper, we develop a two-stage framework that combines fine-tuning and rectification, and optimally allocates limited labeled samples across the two stages. Unlike the conventional objective that minimizes the mean squared prediction errors, we propose to minimize the variance of the prediction errors as the fine-tuning objective, which is optimal for the downstream rectification stage. Building on this insight, we leverage the scaling law of fine-tuning to optimally allocate the limited labeled human data between the fine-tuning and rectification stages. Our empirical analysis validates the fine-tuning scaling law and confirms that our proposed optimal allocation rule reliably identifies the optimal sample allocation. We demonstrate substantial efficiency gains in estimation and inference performance relative to fine-tuning or rectification alone, or to employing the standard mean-squared error objective within the fine-tuning then rectification framework, resulting in significant cost savings for reliable business decisions.
Authors: Suzan Ece Ada, Georg Martius, Emre Ugur, Erhan Oztop
Abstract: Offline Reinforcement Learning (RL) provides a promising avenue for training policies from pre-collected datasets when gathering additional interaction data is infeasible. However, existing offline RL methods often assume stationarity or only consider synthetic perturbations at test time, assumptions that often fail in real-world scenarios characterized by abrupt, time-varying offsets. These offsets can lead to partial observability, causing agents to misperceive their true state and degrade performance. To overcome this challenge, we introduce Forecasting in Non-stationary Offline RL (FORL), a framework that unifies (i) conditional diffusion-based candidate state generation, trained without presupposing any specific pattern of future non-stationarity, and (ii) zero-shot time-series foundation models. FORL targets environments prone to unexpected, potentially non-Markovian offsets, requiring robust agent performance from the onset of each episode. Empirical evaluations on offline RL benchmarks, augmented with real-world time-series data to simulate realistic non-stationarity, demonstrate that FORL consistently improves performance compared to competitive baselines. By integrating zero-shot forecasting with the agent's experience, we aim to bridge the gap between offline RL and the complexities of real-world, non-stationary environments.
Authors: David Strnadel
Abstract: This paper presents an exploratory, simulator-based proof of concept investigating whether differentiable energy terms derived from parameterized quantum circuits can serve as auxiliary regularization signals in Generative Adversarial Networks (GANs). We augment the Auxiliary Classifier GAN (ACGAN) generator objective with a Variational Quantum Eigensolver (VQE)-inspired energy term computed from class-specific Ising Hamiltonians using Qiskit's EstimatorQNN and TorchConnector. All experiments are performed on a noiseless statevector simulator with only four qubits, using a deliberately simple Hamiltonian parameterization. On MNIST, the energy-regularized model initially achieves high external-classifier accuracy (99-100 percent) within five epochs compared to 87.8 percent for an earlier, unmatched ACGAN baseline. However, a rigorous, pre-registered ablation study demonstrates that these improvements are fully replicated by simple classical alternatives, including learned per-class biases, MLP-based surrogates, random noise, and even an unregularized baseline under matched training conditions. All classical variants reach approximately 99 percent accuracy. For sample quality as measured by FID, classical baselines are not merely equivalent but systematically superior to the VQE-based formulation. We therefore report a clear negative result. The VQE-inspired energy term provides no measurable causal benefit beyond trivial classical regularizers in this setting. The primary contribution of this work is methodological, demonstrating both the technical feasibility of differentiable VQE integration into GAN training and the necessity of rigorous ablation studies to avoid spurious claims of quantum-enhanced performance.
Authors: Yunhong Min, Juil Koo, Seungwoo Yoo, Minhyuk Sung
Abstract: We introduce B\'ezierFlow, a lightweight training approach for few-step generation with pretrained diffusion and flow models. B\'ezierFlow achieves a 2-3x performance improvement for sampling with $\leq$ 10 NFEs while requiring only 15 minutes of training. Recent lightweight training approaches have shown promise by learning optimal timesteps, but their scope remains restricted to ODE discretizations. To broaden this scope, we propose learning the optimal transformation of the sampling trajectory by parameterizing stochastic interpolant (SI) schedulers. The main challenge lies in designing a parameterization that satisfies critical desiderata, including boundary conditions, differentiability, and monotonicity of the SNR. To effectively meet these requirements, we represent scheduler functions as B\'ezier functions, where control points naturally enforce these properties. This reduces the problem to learning an ordered set of points in the time range, while the interpretation of the points changes from ODE timesteps to B\'ezier control points. Across a range of pretrained diffusion and flow models, B\'ezierFlow consistently outperforms prior timestep-learning methods, demonstrating the effectiveness of expanding the search space from discrete timesteps to B\'ezier-based trajectory transformations.
Authors: Xinjie He, Chenggong Zhang
Abstract: Partial Differential Equations (PDEs) are central to modeling complex systems across physical, biological, and engineering domains, yet traditional numerical methods often struggle with high-dimensional or complex problems. Physics-Informed Neural Networks (PINNs) have emerged as an efficient alternative by embedding physics-based constraints into deep learning frameworks, but they face challenges in achieving high accuracy and handling complex boundary conditions. In this work, we extend the Time-Evolving Natural Gradient (TENG) framework to address Dirichlet boundary conditions, integrating natural gradient optimization with numerical time-stepping schemes, including Euler and Heun methods, to ensure both stability and accuracy. By incorporating boundary condition penalty terms into the loss function, the proposed approach enables precise enforcement of Dirichlet constraints. Experiments on the heat equation demonstrate the superior accuracy of the Heun method due to its second-order corrections and the computational efficiency of the Euler method for simpler scenarios. This work establishes a foundation for extending the framework to Neumann and mixed boundary conditions, as well as broader classes of PDEs, advancing the applicability of neural network-based solvers for real-world problems.
Authors: Jiwoo Song, Daning Huang, John Harlim
Abstract: In this work, we propose a simple kernel ridge regression (KRR) framework with a dynamic-aware validation strategy for long-term prediction of complex dynamical systems. By employing a data-driven kernel derived from diffusion maps, the proposed Diffusion Maps Kernel Ridge Regression (DM-KRR) method implicitly adapts to the intrinsic geometry of the system's invariant set, without requiring explicit manifold reconstruction or attractor modeling, procedures that often limit predictive performance. Across a broad range of systems, including smooth manifolds, chaotic attractors, and high-dimensional spatiotemporal flows, DM-KRR consistently outperforms state-of-the-art random feature, neural-network and operator-learning methods in both accuracy and data efficiency. These findings underscore that long-term predictive skill depends not only on model expressiveness, but critically on respecting the geometric constraints encoded in the data through dynamically consistent model selection. Together, simplicity, geometry awareness, and strong empirical performance point to a promising path for reliable and efficient learning of complex dynamical systems.
Authors: Yidong Chai, Yi Liu, Mohammadreza Ebrahimi, Weifeng Li, Balaji Padmanabhan
Abstract: Social media platforms are plagued by harmful content such as hate speech, misinformation, and extremist rhetoric. Machine learning (ML) models are widely adopted to detect such content; however, they remain highly vulnerable to adversarial attacks, wherein malicious users subtly modify text to evade detection. Enhancing adversarial robustness is therefore essential, requiring detectors that can defend against diverse attacks (generalizability) while maintaining high overall accuracy. However, simultaneously achieving both optimal generalizability and accuracy is challenging. Following the computational design science paradigm, this study takes a sequential approach that first proposes a novel framework (Large Language Model-based Sample Generation and Aggregation, LLM-SGA) by identifying the key invariances of textual adversarial attacks and leveraging them to ensure that a detector instantiated within the framework has strong generalizability. Second, we instantiate our detector (Adversarially Robust Harmful Online Content Detector, ARHOCD) with three novel design components to improve detection accuracy: (1) an ensemble of multiple base detectors that exploits their complementary strengths; (2) a novel weight assignment method that dynamically adjusts weights based on each sample's predictability and each base detector's capability, with weights initialized using domain knowledge and updated via Bayesian inference; and (3) a novel adversarial training strategy that iteratively optimizes both the base detectors and the weight assignor. We addressed several limitations of existing adversarial robustness enhancement research and empirically evaluated ARHOCD across three datasets spanning hate speech, rumor, and extremist content. Results show that ARHOCD offers strong generalizability and improves detection accuracy under adversarial conditions.
Authors: Maxmillan Ries, Sohan Seth
Abstract: Survival analysis is a critical tool for modeling time-to-event data. Recent deep learning-based models have reduced various modeling assumptions including proportional hazard and linearity. However, a persistent challenge remains in incorporating longitudinal covariates, with prior work largely focusing on cross-sectional features, and in assessing calibration of these models, with research primarily focusing on discrimination during evaluation. We introduce TraCeR, a transformer-based survival analysis framework for incorporating longitudinal covariates. Based on a factorized self-attention architecture, TraCeR estimates the hazard function from a sequence of measurements, naturally capturing temporal covariate interactions without assumptions about the underlying data-generating process. The framework is inherently designed to handle censored data and competing events. Experiments on multiple real-world datasets demonstrate that TraCeR achieves substantial and statistically significant performance improvements over state-of-the-art methods. Furthermore, our evaluation extends beyond discrimination metrics and assesses model calibration, addressing a key oversight in literature.
Authors: Liping Sun, Yucheng Guo, Siliang Lu, Zhenzhen Li
Abstract: With the press of global climate change, extreme weather and sudden weather changes are becoming increasingly common. To maintain a comfortable indoor environment and minimize the contribution of the building to climate change as much as possible, higher requirements are placed on the operation and control of HVAC systems, e.g., more energy-efficient and flexible to response to the rapid change of weather. This places demands on the rapid modeling and prediction of zone air temperatures of buildings. Compared to the traditional simulation-based approach such as EnergyPlus and DOE2, a hybrid approach combined physics and data-driven is more suitable. Recently, the availability of high-quality datasets and algorithmic breakthroughs have driven a considerable amount of work in this field. However, in the niche of short- and long-term predictions, there are still some gaps in existing research. This paper aims to develop a time series forecast model to predict the zone air temperature in a building located in America on a 2-week horizon. The findings could be further improved to support intelligent control and operation of HVAC systems (i.e. demand flexibility) and could also be used as hybrid building energy modeling.
Authors: Yuxing Gan, Ziyu Lei
Abstract: Mixture-of-Experts (MoE) architectures achieve parameter efficiency through conditional computation, yet contemporary designs suffer from two fundamental limitations: structural parameter isolation that causes catastrophic forgetting, and instruction-overfitting that degrades performance in instruction-free scenarios. We propose CDSP-MoE (Conflict-Driven Subspace Pruning MoE), a framework that addresses these issues through a paradigm shift from isolated expert containers to dynamic expert instantiation within a shared physical subspace. Grounded in the Universal Weight Subspace Hypothesis, CDSP-MoE maintains a super-complete parameter backbone where logical experts are carved out via learnable topology masks. Unlike prior work that uses gradient conflict for token reassignment or optimization surgery, we leverage it as a structural supervisory signal: a Lagged Gradient Game penalizes interfering connections in the shared manifold, enabling the topology to spontaneously prune conflicting pathways and evolve interpretable modular structures. Experimental results demonstrate that CDSP-MoE achieves robust content-driven routing without human-defined task labels, maintaining semantic specialization even under strict blind inference protocols where explicit instructions are absent. Code is available at: https://github.com/konodiodaaaaa1/Conflict-Driven-Subspace-Pruning-Mixture-of-Experts
URLs: https://github.com/konodiodaaaaa1/Conflict-Driven-Subspace-Pruning-Mixture-of-Experts
Authors: Wenlong Tang
Abstract: This study proposes a multi-agent language framework that enables continual strategy evolution without fine-tuning the language model's parameters. The core idea is to liberate the latent vectors of abstract concepts from traditional static semantic representations, allowing them to be continuously updated through environmental interaction and reinforcement feedback. We construct a dual-loop architecture: the behavior loop adjusts action preferences based on environmental rewards, while the language loop updates the external latent vectors by reflecting on the semantic embeddings of generated text. Together, these mechanisms allow agents to develop stable and disentangled strategic styles over long-horizon multi-round interactions. Experiments show that agents' latent spaces exhibit clear convergence trajectories under reflection-driven updates, along with structured shifts at critical moments. Moreover, the system demonstrates an emergent ability to implicitly infer and continually adapt to emotional agents, even without shared rewards. These results indicate that, without modifying model parameters, an external latent space can provide language agents with a low-cost, scalable, and interpretable form of abstract strategic representation.
Authors: Gil Kur, Eli Putterman
Abstract: This work studies the computational aspects of multivariate convex regression in dimensions $d \ge 5$. Our results include the \emph{first} estimators that are minimax optimal (up to logarithmic factors) with polynomial runtime in the sample size for both $L$-Lipschitz convex regression, and $\Gamma$-bounded convex regression under polytopal support. Our analysis combines techniques from empirical process theory, stochastic geometry, and potential theory, and leverages recent algorithmic advances in mean estimation for random vectors and in distribution-free linear regression. These results provide the first efficient, minimax-optimal procedures for non-Donsker classes for which their corresponding least-squares estimator is provably minimax-suboptimal.
Authors: Jianing Ye, Chenghao Li, Yongqiang Dou, Jianhao Wang, Guangwen Yang, Chongjie Zhang
Abstract: Decentralized execution is one core demand in multi-agent reinforcement learning (MARL). Recently, most popular MARL algorithms have adopted decentralized policies to enable decentralized execution, and use gradient descent as the optimizer. However, there is hardly any theoretical analysis of these algorithms taking the optimization method into consideration, and we find that various popular MARL algorithms with decentralized policies are suboptimal in toy tasks when gradient descent is chosen as their optimization method. In this paper, we theoretically analyze two common classes of algorithms with decentralized policies -- multi-agent policy gradient methods and value-decomposition methods, and prove their suboptimality when gradient descent is used. To address the suboptimality issue, we propose the Transformation And Distillation (TAD) framework, which reformulates a multi-agent MDP as a special single-agent MDP with a sequential structure and enables decentralized execution by distilling the learned policy on the derived "single-agent" MDP. The approach is a two-stage learning paradigm that addresses the optimization problem in cooperative MARL, providing optimality guarantee with decent execution performance. Empirically, we implement TAD-PPO based on PPO, which can theoretically perform optimal policy learning in the finite multi-agent MDPs and shows significant outperformance on a large set of cooperative multi-agent tasks, from matrix game, hallway task, to StarCraft II, and football game.
Authors: G. H. Bustos, H. H. Segnorile
Abstract: We propose a feature-extraction procedure based on the statistical characterization of waveforms, applied as a fast pre-processing stage in a pattern recognition task using simple artificial neural network models. This procedure involves measuring a set of 30 parameters, including moments and cumulants obtained from the waveform, its derivative, and its integral. The technique is presented as an extension of the Statistical Signal Characterization method, which is already established in the literature, and we referred to it as ESSC. As a testing methodology, we employed a procedure to distinguish a pulse-like signal from different versions of itself with altered or deformed frequency spectra, under various signal-to-noise ratio (SNR) conditions of Gaussian white noise. The recognition task was performed by machine learning networks using the proposed ESSC feature extraction method. Additionally, we compared the results with those obtained using raw data inputs in deep learning networks. The algorithms were trained and tested on cases involving Sinc-, Gaussian-, and Chirp-pulse waveforms. We measure accuracy and execution time for the different algorithms solving these pattern-recognition cases, and evaluate the architectural complexity of building such networks. We conclude that a simple multi-layer perceptron network using ESSC can achieve an accuracy of around 90%, comparable to that of deep learning algorithms, when solving pattern recognition tasks in practical scenarios with SNR above 20dB. Additionally, this approach offers an execution time approximately 4 times shorter and significantly lower network construction complexity, enabling its use in low-resource computational systems.
Authors: Milad Abdollahzadeh, Guimeng Liu, Touba Malekzadeh, Christopher T. H. Teo, Keshigeyan Chandrasegaran, Ngai-Man Cheung
Abstract: Generative modeling in machine learning aims to synthesize new data samples that are statistically similar to those observed during training. While conventional generative models such as GANs and diffusion models typically assume access to large and diverse datasets, many real-world applications (e.g. in medicine, satellite imaging, and artistic domains) operate under limited data availability and strict constraints. In this survey, we examine Generative Modeling under Data Constraint (GM-DC), which includes limited-data, few-shot, and zero-shot settings. We present a unified perspective on the key challenges in GM-DC, including overfitting, frequency bias, and incompatible knowledge transfer, and discuss how these issues impact model performance. To systematically analyze this growing field, we introduce two novel taxonomies: one categorizing GM-DC tasks (e.g. unconditional vs. conditional generation, cross-domain adaptation, and subject-driven modeling), and another organizing methodological approaches (e.g. transfer learning, data augmentation, meta-learning, and frequency-aware modeling). Our study reviews over 230 papers, offering a comprehensive view across generative model types and constraint scenarios. We further analyze task-approach-method interactions using a Sankey diagram and highlight promising directions for future work, including adaptation of foundation models, holistic evaluation frameworks, and data-centric strategies for sample selection. This survey provides a timely and practical roadmap for researchers and practitioners aiming to advance generative modeling under limited data. Project website: https://sutd-visual-computing-group.github.io/gmdc-survey/.
URLs: https://sutd-visual-computing-group.github.io/gmdc-survey/.
Authors: Vishal Ramesh, Rui Zhao, Naman Goel
Abstract: Data on the Web has fueled much of the recent progress in AI. As more high-quality data becomes difficult to access, synthetic data is emerging as a promising solution for privacy-friendly data release and complementing real datasets in developing robust and safe AI. But there is limited work on decentralised, scalable and contributor-centric synthetic data generation systems. A recent proposal, called Libertas, allows data contributors to autonomously participate in joint computations over their Web data without relying on a trusted centre. Libertas uses Solid (Social Linked Data) and MPC (Secure Multi-Party Computation) to achieve this goal. Solid is a decentralised Web specification that lets anyone store their data securely in their personal decentralised data stores called Pods and control which applications have access to their data. MPC refers to the set of cryptographic methods for different parties to jointly compute a function over their inputs while keeping those inputs private. Thus, Libertas can also be used to generate synthetic data from otherwise inaccessible Web data in a responsible way, by ensuring contributor autonomy, decentralisation and privacy. However, the scalability of this system remains limited due to the high computation and communication costs in MPC. In this paper, we show how one can improve Libertas using secure enclaves (in addition to MPC) to address the scalability challenge. Secure enclaves such as Intel SGX rely on hardware based features for confidentiality and integrity of code and data. We discuss a principled approach for integrating SGX within the Libertas architecture for scalable differentially private synthetic data generation, and support our analysis with rigorous empirical results on simulated and real datasets and different synthetic data generation algorithms.
Authors: Xinrui Jiang, Yohan Jun, Jaejin Cho, Mengze Gao, Xingwang Yong, Berkin Bilgic
Abstract: Typical quantitative MRI (qMRI) methods estimate parameter maps in a two-step pipeline that first reconstructs images from undersampled k-space data and then performs model fitting, which is prone to biases and error propagation. We propose NLCG-Net, a model-based nonlinear conjugate gradient (NLCG) framework for joint T2/T1 estimation that incorporates a U-Net regularizer trained in a scan-specific, zero-shot fashion. The method directly estimates qMRI maps from undersampled k-space using mono-exponential signal modeling with scan-specific neural network regularization, enabling high-fidelity T1 and T2 mapping. Experimental results on T2 and T1 mapping demonstrate that NLCG-Net improves estimation quality over subspace reconstruction at high acceleration factors.
Authors: Marion Ullmo, Nabila Aghanim, Aur\'elien Decelle, Miguel Aragon-Calvo
Abstract: Predicting the nonlinear evolution of cosmic structure from initial conditions is typically approached using Lagrangian, particle-based methods. These techniques excel in terms of tracking individual trajectories, but they might not be suitable for applications where point-based information is unavailable or impractical. In this work, we explore an alternative, field-based approach using Eulerian inputs. Specifically, we developed an autoencoder architecture based on a generative adversarial network (GAN) and trained it to evolve density fields drawn from dark matter N-body simulations. We tested this method on both 2D and 3D data. We find that while predictions on 2D density maps perform well based on density alone, accurate 3D predictions require the inclusion of associated velocity fields. Our results demonstrate the potential of field-based representations to model cosmic structure evolution, offering a complementary path to Lagrangian methods in contexts where field-level data is more accessible.
Authors: Nico Baumgart, Markus Lange-Hegermann, Mike M\"ucke
Abstract: In industrial manufacturing, deploying deep learning models for visual inspection is mostly hindered by the high and often intractable cost of collecting and annotating large-scale training datasets. While image synthesis from 3D CAD models is a common solution, the individual techniques of domain and rendering randomization to create rich synthetic training datasets have been well studied mainly in simple domains. Hence, their effectiveness on complex industrial tasks with densely arranged and similar objects remains unclear. In this paper, we investigate the sim-to-real generalization performance of standard object detectors on the complex industrial application of terminal strip object detection, carefully combining randomization and domain knowledge. We describe step-by-step the creation of our image synthesis pipeline that achieves high realism with minimal implementation effort and explain how this approach could be transferred to other industrial settings. Moreover, we created a dataset comprising 30.000 synthetic images and 300 manually annotated real images of terminal strips, which is publicly available for reference and future research. To provide a baseline as a lower bound of the expectable performance in these challenging industrial parts detection tasks, we show the sim-to-real generalization performance of standard object detectors on our dataset based on a fully synthetic training. While all considered models behave similarly, the transformer-based DINO model achieves the best score with 98.40 % mean average precision on the real test set, demonstrating that our pipeline enables high quality detections in complex industrial environments from existing CAD data and with a manageable image synthesis effort.
Authors: Massinissa Merouani, Afif Boudaoud, Iheb Nassim Aouadj, Nassim Tchoulak, Islem Kara Bernou, Hamza Benyamina, Fatima Benbouzid-Si Tayeb, Karima Benatchba, Hugh Leather, Riyadh Baghdadi
Abstract: While polyhedral compilers have shown success in implementing advanced code transformations, they still face challenges in selecting the ones that lead to the most profitable speedups. This has motivated the use of machine learning based cost models to guide the search for polyhedral optimizations. State-of-the-art polyhedral compilers have demonstrated a viable proof-of-concept of such an approach. While promising, this approach still faces significant limitations. State-of-the-art polyhedral compilers that use a deep learning cost model only support a small subset of affine transformations, limiting their ability to explore complex code transformations. Furthermore, their applicability does not scale beyond simple programs, thus excluding many program classes from their scope, such as those with non-rectangular iteration domains or multiple loop nests. These limitations significantly impact the generality of such compilers and autoschedulers and put into question the whole approach. In this paper, we introduce LOOPer, the first polyhedral autoscheduler that uses a deep learning based cost model and covers a large space of affine transformations and programs. LOOPer allows the optimization of an extensive set of programs while being effective at applying complex sequences of polyhedral transformations. We implement and evaluate LOOPer and show that it achieves competitive speedups over the state-of-the-art. On the PolyBench benchmarks, LOOPer achieves a geometric mean speedup of 1.84x over Tiramisu and 1.42x over Pluto, two state-of-the-art polyhedral autoschedulers.
Authors: Shankhanil Mitra, Diptanu De, Shika Rao, Rajiv Soundararajan
Abstract: The design of image and video quality assessment (QA) algorithms is extremely important to benchmark and calibrate user experience in modern visual systems. A major drawback of the state-of-the-art QA methods is their limited ability to generalize across diverse image and video datasets with reasonable distribution shifts. In this work, we leverage the denoising process of diffusion models for generalized image QA (IQA) and video QA (VQA) by understanding the degree of alignment between learnable quality-aware text prompts and images or video frames. In particular, we learn cross-attention maps from intermediate layers of the denoiser of latent diffusion models (LDMs) to capture quality-aware representations of images or video frames. Since applying text-to-image LDMs for every video frame is computationally expensive for videos, we only estimate the quality of a frame-rate sub-sampled version of the original video. To compensate for the loss in motion information due to frame-rate sub-sampling, we propose a novel temporal quality modulator. Our extensive cross-database experiments across various user-generated, synthetic, low-light, frame-rate variation, ultra high definition, and streaming content-based databases show that our model can achieve superior generalization in both IQA and VQA.
Authors: Yu-Jui Huang, Zachariah Malik
Abstract: This paper approaches the unsupervised learning problem by minimizing the second-order Wasserstein loss (the $W_2$ loss) through a distribution-dependent ordinary differential equation (ODE), whose dynamics involves the Kantorovich potential associated with the true data distribution and a current estimate of it. A main result shows that the time-marginal laws of the ODE form a gradient flow for the $W_2$ loss, which converges exponentially to the true data distribution. An Euler scheme for the ODE is proposed and it is shown to recover the gradient flow for the $W_2$ loss in the limit. An algorithm is designed by following the scheme and applying persistent training, which naturally fits our gradient-flow approach. In both low- and high-dimensional experiments, our algorithm outperforms Wasserstein generative adversarial networks by increasing the level of persistent training appropriately.
Authors: Hang Yu, Qidi Fang, Shijie Fang, Reuben M. Aronson, Elaine Schaertl Short
Abstract: Enhancing the expressiveness of human teaching is vital for both improving robots' learning from humans and the human-teaching-robot experience. In this work, we characterize and test a little-used teaching signal: \textit{progress}, designed to represent the completion percentage of a task. We conducted two online studies with 76 crowd-sourced participants and one public space study with 40 non-expert participants to validate the capability of this progress signal. We find that progress indicates whether the task is successfully performed, reflects the degree of task completion, identifies unproductive but harmless behaviors, and is likely to be more consistent across participants. Furthermore, our results show that giving progress does not require extra workload and time. An additional contribution of our work is a dataset of 40 non-expert demonstrations from the public space study through an ice cream topping-adding task, which we observe to be multi-policy and sub-optimal, with sub-optimality not only from teleoperation errors but also from exploratory actions and attempts. The dataset is available at https://github.com/TeachingwithProgress/Non-Expert\_Demonstrations.
URLs: https://github.com/TeachingwithProgress/Non-Expert\_Demonstrations.
Authors: Harry Cheon, Anneke Wernerfelt, Sorelle A. Friedler, Berk Ustun
Abstract: Consumer protection rules require companies that deploy models to automate decisions in high-stakes settings to explain predictions to decision subjects. These rules are motivated, in part, by the belief that explanations can promote recourse by revealing information that decision subjects can use to contest or overturn their predictions. In practice, companies provide individuals with a list of principal reasons based on feature importance derived from methods like SHAP and LIME. In this work, we show how common practices can fail to provide recourse and propose to highlight features based on their responsiveness -- the probability that a decision subject can attain a target prediction through an arbitrary intervention on the feature. We develop efficient methods to compute responsiveness scores for any model and actionability constraints. We show that standard practices in lending can undermine decision subjects by highlighting unresponsive features and explaining predictions that are fixed.
Authors: Jiaqi Yang, Nitish Mehta, Xiaoling Hu, Chao Chen, Chia-Ling Tsai
Abstract: Accurate segmentation of Optical Coherence Tomography (OCT) images is crucial for diagnosing and monitoring retinal diseases. However, the labor-intensive nature of pixel-level annotation limits the scalability of supervised learning for large datasets. Weakly Supervised Semantic Segmentation (WSSS) offers a promising alternative by using weaker forms of supervision, such as image-level labels, to reduce the annotation burden. Despite its advantages, weak supervision inherently carries limited information. We propose a novel WSSS framework with only image-level labels for OCT lesion segmentation that integrates structural and text-driven guidance to produce high-quality, pixel-level pseudo labels. The framework employs two visual processing modules: one that processes the original OCT images and another that operates on layer segmentations augmented with anomalous signals, enabling the model to associate lesions with their corresponding anatomical layers. Complementing these visual cues, we leverage large-scale pretrained models to provide two forms of textual guidance: label-derived descriptions that encode local semantics, and domain-agnostic synthetic descriptions that, although expressed in natural image terms, capture spatial and relational semantics useful for generating globally consistent representations. By fusing these visual and textual features in a multi-modal framework, our method aligns semantic meaning with structural relevance, thereby improving lesion localization and segmentation performance. Experiments on three OCT datasets demonstrate state-of-the-art results, highlighting its potential to advance diagnostic accuracy and efficiency in medical imaging.
Authors: Ehsan Kabir, Jason D. Bakos, David Andrews, Miaoqing Huang
Abstract: Transformer neural networks (TNN) excel in natural language processing (NLP), machine translation, and computer vision (CV) without relying on recurrent or convolutional layers. However, they have high computational and memory demands, particularly on resource-constrained devices like FPGAs. Moreover, transformer models vary in processing time across applications, requiring custom models with specific parameters. Designing custom accelerators for each model is complex and time-intensive. Some custom accelerators exist with no runtime adaptability, and they often rely on sparse matrices to reduce latency. However, hardware designs become more challenging due to the need for application-specific sparsity patterns. This paper introduces ADAPTOR, a runtime-adaptive accelerator for dense matrix computations in transformer encoders and decoders on FPGAs. ADAPTOR enhances the utilization of processing elements and on-chip memory, enhancing parallelism and reducing latency. It incorporates efficient matrix tiling to distribute resources across FPGA platforms and is fully quantized for computational efficiency and portability. Evaluations on Xilinx Alveo U55C data center cards and embedded platforms like VC707 and ZCU102 show that our design is 1.2$\times$ and 2.87$\times$ more power efficient than the NVIDIA K80 GPU and the i7-8700K CPU respectively. Additionally, it achieves a speedup of 1.7 to 2.25$\times$ compared to some state-of-the-art FPGA-based accelerators.
Authors: Zhihao Sun, Haoran Jiang, Haoran Chen, Yixin Cao, Xipeng Qiu, Zuxuan Wu, Yu-Gang Jiang
Abstract: Multimodal large language models have unlocked new possibilities for various multimodal tasks. However, their potential in image manipulation detection remains unexplored. When directly applied to the IMD task, M-LLMs often produce reasoning texts that suffer from hallucinations and overthinking. To address this, we propose ForgerySleuth, which leverages M-LLMs to perform comprehensive clue fusion and generate segmentation outputs indicating specific regions that are tampered with. Moreover, we construct the ForgeryAnalysis dataset through the Chain-of-Clues prompt, which includes analysis and reasoning text to upgrade the image manipulation detection task. A data engine is also introduced to build a larger-scale dataset for the pre-training phase. Our extensive experiments demonstrate the effectiveness of ForgeryAnalysis and show that ForgerySleuth significantly outperforms existing methods in generalization, robustness, and explainability.
Authors: Yixiong Fang, Ziran Yang, Zhaorun Chen, Zhuokai Zhao, Jiawei Zhou
Abstract: Large vision-language models (LVLMs) excel at multimodal tasks but are prone to misinterpreting visual inputs, often resulting in hallucinations and unreliable outputs. We present DROPOUT DECODING, a novel inference-time approach that quantifies the uncertainty of visual tokens and selectively masks uncertain tokens to improve decoding. Our method measures the uncertainty of each visual token by projecting it onto the text space and decomposing it into aleatoric and epistemic components. Specifically, we focus on epistemic uncertainty, which captures perception-related errors more effectively. Inspired by dropout regularization, we introduce uncertainty-guided token dropout, which applies the dropout principle to input visual tokens instead of model parameters, and during inference rather than training. By aggregating predictions from an ensemble of masked decoding contexts, we can robustly mitigate errors arising from visual token misinterpretations. Evaluations on benchmarks including CHAIR, THRONE, and MMBench demonstrate that DROPOUT DECODING significantly reduces object hallucinations (OH) and enhances both reliability and quality of LVLM outputs across diverse visual contexts. Code is released at https://github.com/kigb/DropoutDecoding.
Authors: O. Deniz Akyildiz, Pierre Del Moral, Joaqu\'in Miguez
Abstract: Entropic optimal transport problems are regularized versions of optimal transport problems. These models play an increasingly important role in machine learning and generative modelling. For finite spaces, these problems are commonly solved using Sinkhorn algorithm (a.k.a. iterative proportional fitting procedure). However, in more general settings the Sinkhorn iterations are based on nonlinear conditional/conjugate transformations and exact finite-dimensional solutions cannot be computed. This article presents a finite-dimensional recursive formulation of the iterative proportional fitting procedure for general Gaussian multivariate models. As expected, this recursive formulation is closely related to the celebrated Kalman filter and related Riccati matrix difference equations, and it yields algorithms that can be implemented in practical settings without further approximations. We extend this filtering methodology to develop a refined and self-contained convergence analysis of Gaussian Sinkhorn algorithms, including closed form expressions of entropic transport maps and Schr\"odinger bridges.
Authors: Pratik Nag
Abstract: Music generation has emerged as a significant topic in artificial intelligence and machine learning. While recurrent neural networks (RNNs) have been widely employed for sequence generation, generative adversarial networks (GANs) remain relatively underexplored in this domain. This paper presents two systems based on adversarial networks for music generation. The first system learns a set of music pieces without differentiating between styles, while the second system focuses on learning and deviating from specific composers' styles to create innovative music. By extending the Creative Adversarial Networks (CAN) framework to the music domain, this work introduces unrolled CAN to address mode collapse, evaluating both GAN and CAN in terms of creativity and variation.
Authors: Xindi Wu, Mengzhou Xia, Rulin Shao, Zhiwei Deng, Pang Wei Koh, Olga Russakovsky
Abstract: Training vision-language models via instruction tuning relies on large data mixtures spanning diverse tasks and domains, yet these mixtures frequently include redundant information that increases computational costs without proportional gains. Existing methods typically rely on task-agnostic heuristics to estimate data importance, limiting their effectiveness across tasks. We introduce ICONS, a gradient-based Influence CONsensus approach for vision-language data Selection. Our method leverages first-order training dynamics to estimate each example's influence on validation performance, then aggregates these estimates across tasks via majority voting. This cross-task consensus identifies consistently valuable data points while mitigating score calibration and outlier sensitivity, enabling robust and scalable data selection for diverse multitask mixtures. Models trained on our selected 20% data subset from LLAVA-665K (respectively: from CAMBRIAN-7M, from VISION-FLAN-186K) retain 98.6% (respectively: 98.8%, 99.8%) of full-dataset performance. We demonstrate that our selected data generalizes to unseen tasks and model architectures, and release three compact subsets LLAVA-ICONS-133K, CAMBRIAN-ICONS-1.4M, and VISION-FLAN-ICONS-37K for efficient vision-language model development.
Authors: Terrance Yu-Hao Chen, Yulin Chen, Pontus Soederhaell, Sadrishya Agrawal, Kateryna Shapovalenko
Abstract: Decoding speech from non-invasive brain signals, such as electroencephalography (EEG), has the potential to advance brain-computer interfaces (BCIs), with applications in silent communication and assistive technologies for individuals with speech impairments. However, EEG-based speech decoding faces major challenges, such as noisy data, limited datasets, and poor performance on complex tasks like speech perception. This study attempts to address these challenges by employing variational autoencoders (VAEs) for EEG data augmentation to improve data quality and applying a state-of-the-art (SOTA) sequence-to-sequence deep learning architecture, originally successful in electromyography (EMG) tasks, to EEG-based speech decoding. Additionally, we adapt this architecture for word classification tasks. Using the Brennan dataset, which contains EEG recordings of subjects listening to narrated speech, we preprocess the data and evaluate both classification and sequence-to-sequence models for EEG-to-words/sentences tasks. Our experiments show that VAEs have the potential to reconstruct artificial EEG data for augmentation. Meanwhile, our sequence-to-sequence model achieves more promising performance in generating sentences compared to our classification model, though both remain challenging tasks. These findings lay the groundwork for future research on EEG speech perception decoding, with possible extensions to speech production tasks such as silent or imagined speech.
Authors: Diyana Muhammed, Giusy Giulia Tuccari, Gollam Rabby, S\"oren Auer, Sahar Vahdati
Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse applications, from open-domain question answering to scientific writing, medical decision support, and legal analysis. However, their tendency to generate incorrect or fabricated content, commonly known as hallucinations, represents a critical barrier to reliable deployment in high-stakes domains. Current hallucination detection benchmarks are limited in scope, focusing primarily on general-knowledge domains while neglecting specialised fields where accuracy is paramount. To address this gap, we introduce the AIME Math Hallucination dataset, the first comprehensive benchmark specifically designed for evaluating mathematical reasoning hallucinations. Additionally, we propose SelfCheck-Eval, a LLM-agnostic, black-box hallucination detection framework applicable to both open and closed-source LLMs. Our approach leverages a novel multi-module architecture that integrates three independent detection strategies: the Semantic module, the Specialised Detection module, and the Contextual Consistency module. Our evaluation reveals systematic performance disparities across domains: existing methods perform well on biographical content but struggle significantly with mathematical reasoning, a challenge that persists across NLI fine-tuning, preference learning, and process supervision approaches. These findings highlight the fundamental limitations of current detection methods in mathematical domains and underscore the critical need for specialised, black-box compatible approaches to ensure reliable LLM deployment.
Authors: Zhangsong Li
Abstract: In this paper, assuming the low-degree conjecture, we provide evidence of computational hardness for two problems: (1) the (partial) matching recovery problem in the sparse correlated Erd\H{o}s-R\'enyi graphs $\mathcal G(n,q;\rho)$ when the edge-density $q=n^{-1+o(1)}$ and the correlation $\rho<\sqrt{\alpha}$ lies below the Otter's threshold, this resolves a remaining problem in \cite{DDL23+}; (2) the detection problem between a pair of correlated sparse stochastic block models $\mathcal S(n,\tfrac{\lambda}{n};k,\epsilon;s)$ and a pair of independent stochastic block models $\mathcal S(n,\tfrac{\lambda s}{n};k,\epsilon)$ when $\epsilon^2 \lambda s<1$ lies below the Kesten-Stigum (KS) threshold and $s<\sqrt{\alpha}$ lies below the Otter's threshold, this resolves a remaining problem in \cite{CDGL24+}. One of the main ingredient in our proof is to derive certain forms of \emph{algorithmic contiguity} between two probability measures based on bounds on their low-degree advantage. To be more precise, consider the high-dimensional hypothesis testing problem between two probability measures $\mathbb{P}$ and $\mathbb{Q}$ based on the sample $\mathsf Y$. We show that if the low-degree advantage $\mathsf{Adv}_{\leq D} \big( \frac{\mathrm{d}\mathbb{P}}{\mathrm{d}\mathbb{Q}} \big)=O(1)$, then (assuming the low-degree conjecture) there is no efficient algorithm $\mathcal A$ such that $\mathbb{Q}(\mathcal A(\mathsf Y)=0)=1-o(1)$ and $\mathbb{P}(\mathcal A(\mathsf Y)=1)=\Omega(1)$. This framework provides a useful tool for performing reductions between different inference tasks, without requiring a strengthened version of the low-degree conjecture as in \cite{MW23+, DHSS25+}.
Authors: Fengwei Teng, Quan Shi, Zhaoyang Yu, Jiayi Zhang, Yuyu Luo, Chenglin Wu, Zhijiang Guo
Abstract: Large Language Models (LLMs) have achieved significant performance gains through test-time scaling methods. However, existing approaches often incur redundant computations due to the accumulation of historical dependency information during inference. To address this challenge, we leverage the memoryless property of Markov processes to minimize reliance on historical context and propose a Markovian reasoning process. This foundational Markov chain structure enables seamless integration with various test-time scaling methods, thereby improving their scaling efficiency. By further scaling up the Markovian reasoning chain through integration with techniques such as tree search and reflective refinement, we uncover an emergent atomic reasoning structure, where reasoning trajectories are decomposed into a series of self-contained, low-complexity atomic units. We name this design Atom of Thoughts (\our). Extensive experiments demonstrate that \our consistently outperforms existing baselines as computational budgets increase. Importantly, \our integrates seamlessly with existing reasoning frameworks and different LLMs (both reasoning and non-reasoning), facilitating scalable, high-performance inference.We submit our code alongside this paper and will make it publicly available to facilitate reproducibility and future research.
Authors: Aditi De
Abstract: Zero-Input AI (ZIA) introduces a novel framework for human-computer interaction by enabling proactive intent prediction without explicit user commands. It integrates gaze tracking, bio-signals (EEG, heart rate), and contextual data (time, location, usage history) into a multi-modal model for real-time inference, targeting <100 ms latency. The proposed architecture employs a transformer-based model with cross-modal attention, variational Bayesian inference for uncertainty estimation, and reinforcement learning for adaptive optimization. To support deployment on edge devices (CPUs, TPUs, NPUs), ZIA utilizes quantization, weight pruning, and linear attention to reduce complexity from quadratic to linear with sequence length. Theoretical analysis establishes an information-theoretic bound on prediction error and demonstrates how multi-modal fusion improves accuracy over single-modal approaches. Expected performance suggests 85-90% accuracy with EEG integration and 60-100 ms inference latency. ZIA provides a scalable, privacy-preserving framework for accessibility, healthcare, and consumer applications, advancing AI toward anticipatory intelligence.
Authors: Chang Sun, Jennifer Ngadiuba, Maurizio Pierini, Maria Spiropulu
Abstract: We explore the innovative use of MLP-Mixer models for real-time jet tagging and establish their feasibility on resource-constrained hardware like FPGAs. MLP-Mixers excel in processing sequences of jet constituents, achieving state-of-the-art performance on datasets mimicking Large Hadron Collider conditions. By using advanced optimization techniques such as High-Granularity Quantization and Distributed Arithmetic, we achieve unprecedented efficiency. These models match or surpass the accuracy of previous architectures, reduce hardware resource usage by up to 97%, double the throughput, and half the latency. Additionally, non-permutation-invariant architectures enable smart feature prioritization and efficient FPGA deployment, setting a new benchmark for machine learning in real-time data processing at particle colliders.
Authors: Wentai Wu, Ligang He, Saiqin Long, Ahmed M. Abdelmoniem, Yingliang Wu, Rui Mao, Keqin Li
Abstract: Increasing legislation and regulations on private and proprietary information results in scattered data sources also known as the "data islands". Although Federated Learning-based paradigms can enable privacy-preserving collaboration over decentralized data, they have inherent deficiencies in fairness, costs and reproducibility because of being learning-centric, which greatly limits the way how participants cooperate with each other. In light of this, we investigate the possibilities to shift from resource-intensive learning to task-agnostic collaboration especially when the participants have no interest in a common goal. We term this new scenario as Task-Agnostic Federation (TAF), and investigate several branches of research that serve as the technical building blocks. These techniques directly or indirectly embrace data-centric approaches that can operate independently of any learning task. In this article, we first describe the system architecture and problem setting for TAF. Then, we present a three-way roadmap and categorize recent studies in three directions: collaborative data expansion, collaborative data refinement, and collective data harmonization in the federation. Further, we highlight several challenges and open questions that deserve more attention from the community. With our investigation, we intend to offer new insights about how autonomic parties with varied motivation can cooperate over decentralized data beyond learning.
Authors: Antonin Schrab
Abstract: This paper provides a unifying view of optimal kernel hypothesis testing across the MMD two-sample, HSIC independence, and KSD goodness-of-fit frameworks. Minimax optimal separation rates in the kernel and $L^2$ metrics are presented, with two adaptive kernel selection methods (kernel pooling and aggregation), and under various testing constraints: computational efficiency, differential privacy, and robustness to data corruption. Intuition behind the derivation of the power results is provided in a unified way across the three frameworks, and open problems are highlighted.
Authors: Morgane Austern, Yuanchuan Guo, Zheng Tracy Ke, Tianle Liu
Abstract: Topic modeling is traditionally applied to word counts without accounting for the context in which words appear. Recent advancements in large language models (LLMs) offer contextualized word embeddings, which capture deeper meaning and relationships between words. We aim to leverage such embeddings to improve topic modeling. We use a pre-trained LLM to convert each document into a sequence of word embeddings. This sequence is then modeled as a Poisson point process, with its intensity measure expressed as a convex combination of $K$ base measures, each corresponding to a topic. To estimate these topics, we propose a flexible algorithm that integrates traditional topic modeling methods, enhanced by net-rounding applied before and kernel smoothing applied after. One advantage of this framework is that it treats the LLM as a black box, requiring no fine-tuning of its parameters. Another advantage is its ability to seamlessly integrate any traditional topic modeling approach as a plug-in module, without the need for modifications Assuming each topic is a $\beta$-H\"{o}lder smooth intensity measure on the embedded space, we establish the rate of convergence of our method. We also provide a minimax lower bound and show that the rate of our method matches with the lower bound when $\beta\leq 1$. Additionally, we apply our method to several datasets, providing evidence that it offers an advantage over traditional topic modeling approaches.
Authors: Haimeng Zhao, Yuzhen Zhang, John Preskill
Abstract: The energy cost of erasing quantum states depends on our knowledge of the states. We show that learning algorithms can acquire such knowledge to erase many copies of an unknown state at the optimal energy cost. This is proved by showing that learning can be made fully reversible and has no fundamental energy cost itself. With simple counting arguments, we relate the energy cost of erasing quantum states to their complexity, entanglement, and magic. We further show that the constructed erasure protocol is computationally efficient when learning is efficient. Conversely, under standard cryptographic assumptions, we prove that the optimal energy cost cannot be achieved efficiently in general. These results also enable efficient work extraction based on learning. Together, our results establish a concrete connection between quantum learning theory and thermodynamics, highlighting the physical significance of learning processes and enabling provably-efficient learning-based protocols for thermodynamic tasks.
Authors: Yizhuo Wu, Francesco Fioranelli, Chang Gao
Abstract: Radar-based Human Activity Recognition (HAR) is an attractive alternative to wearables and cameras because it preserves privacy, and is contactless and robust to occlusions. However, dominant Convolutional Neural Network (CNN)- and Recurrent Neural Network (RNN)-based solutions are computationally intensive at deployment, and recent lightweight Vision Transformer (ViT) and State Space Model (SSM) variants still exhibit substantial complexity. In this paper, we present RadMamba, a parameter-efficient, micro-Doppler-oriented Mamba SSM tailored to radar HAR under on-sensor compute, latency, and energy constraints typical of distributed radar systems. RadMamba combines (i) channel fusion with downsampling, (ii) Doppler-aligned segmentation that preserves the physical continuity of Doppler over time, and (iii) convolutional token projections that better capture Doppler-span variations, thereby retaining temporal-Doppler structure while reducing the number of Floating-point Operations per Inference (#FLOP/Inf.). Evaluated across three datasets with different radars and types of activities, RadMamba matches the prior best 99.8% accuracy of a recent SSM-based model on the Continuous Wave (CW) radar dataset, while requiring only 1/400 of its parameters. On a dataset of non-continuous activities with Frequency Modulated Continuous Wave (FMCW) radar, RadMamba remains competitive with leading 92.0% results using about 1/10 of the parameters, and on a continuous FMCW radar dataset it surpasses methods with far more parameters by at least 3%, using only 6.7k parameters. Code: https://github.com/lab-emi/AIRHAR.
Authors: Qinfeng Zhu, Yunxi Jiang, Lei Fan
Abstract: We propose a result-level category-specific fusion architecture called ClassWise-CRF. This architecture employs a two-stage process: first, it selects expert networks that perform well in specific categories from a pool of candidate networks using a greedy algorithm; second, it integrates the segmentation predictions of these selected networks by adaptively weighting their contributions based on their segmentation performance in each category. Inspired by Conditional Random Field (CRF), the ClassWise-CRF architecture treats the segmentation predictions from multiple networks as confidence vector fields. It leverages segmentation metrics (such as Intersection over Union) from the validation set as priors and employs an exponential weighting strategy to fuse the category-specific confidence scores predicted by each network. This fusion method dynamically adjusts the weights of each network for different categories, achieving category-specific optimization. Building on this, the architecture further optimizes the fused results using unary and pairwise potentials in CRF to ensure spatial consistency and boundary accuracy. To validate the effectiveness of ClassWise-CRF, we conducted experiments on two remote sensing datasets, LoveDA and Vaihingen, using eight classic and advanced semantic segmentation networks. The results show that the ClassWise-CRF architecture significantly improves segmentation performance: on the LoveDA dataset, the mean Intersection over Union (mIoU) metric increased by 1.00% on the validation set and by 0.68% on the test set; on the Vaihingen dataset, the mIoU improved by 0.87% on the validation set and by 0.91% on the test set. These results fully demonstrate the effectiveness and generality of the ClassWise-CRF architecture in semantic segmentation of remote sensing images. The full code is available at https://github.com/zhuqinfeng1999/ClassWise-CRF.
Authors: Changshuai Wei, Phuc Nguyen, Benjamin Zelditch, Joyce Chen
Abstract: The standard A/B testing approaches are mostly based on t-test in large scale industry applications. These standard approaches however suffers from low statistical power in business settings, due to nature of small sample-size or non-Gaussian distribution or return-on-investment (ROI) consideration. In this paper, we (i) show the statistical efficiency of using estimating equation and U statistics, which can address these issues separately; and (ii) propose a novel doubly robust generalized U that allows flexible definition of treatment effect, and can handles small samples, distribution robustness, ROI and confounding consideration in one framework. We provide theoretical results on asymptotics and efficiency bounds, together with insights on the efficiency gain from theoretical analysis. We further conduct comprehensive simulation studies, apply the methods to multiple real A/B tests at LinkedIn, and share results and learnings that are broadly useful.
Authors: Samuel Simko, Mrinmaya Sachan, Bernhard Sch\"olkopf, Zhijing Jin
Abstract: Large Language Models (LLMs) are powerful tools with profound societal impacts, yet their ability to generate responses to diverse and uncontrolled inputs leaves them vulnerable to adversarial attacks. While existing defenses often struggle to generalize across varying attack types, recent advancements in representation engineering offer promising alternatives. In this work, we propose a defense framework that formulates model defense as a contrastive representation learning (CRL) problem. Our method finetunes a model using a triplet-based loss combined with adversarial hard negative mining to encourage separation between benign and harmful representations. Our experimental results across multiple models demonstrate that our approach outperforms prior representation engineering-based defenses, improving robustness against both input-level and embedding-space attacks without compromising standard performance. Our code is available at https://github.com/samuelsimko/crl-llm-defense
Authors: Georgios Arampatzis, Stylianos Katsarakis, Charalambos Makridakis
Abstract: The integration of Scientific Machine Learning (SciML) techniques with uncertainty quantification (UQ) represents a rapidly evolving frontier in computational science. This work advances Physics-Informed Neural Networks (PINNs) by incorporating probabilistic frameworks to effectively model uncertainty in complex systems. Our approach enhances the representation of uncertainty in forward problems by combining generative modeling techniques with PINNs. This integration enables in a systematic fashion uncertainty control while maintaining the predictive accuracy of the model. We demonstrate the utility of this method through applications to random differential equations and random partial differential equations (PDEs).
Authors: Amirali Sajadi, Kostadin Damevski, Preetha Chatterjee
Abstract: Large language models (LLMs) and their agentic frameworks are increasingly adopted to perform development tasks such as automated program repair (APR). While prior work has identified security risks in LLM-generated code, most have focused on synthetic, simplified, or isolated tasks that lack the complexity of real-world program repair. In this study, we present the first large-scale security analysis of LLM-generated patches using 20,000+ GitHub issues. We evaluate patches proposed by developers, a standalone LLM (Llama 3.3 Instruct-70B), and three top-performing agentic frameworks (OpenHands, AutoCodeRover, HoneyComb). Finally, we analyze a wide range of code, issue, and project-level factors to understand the conditions under which generating insecure patches is more likely. Our findings reveal that Llama introduces many new vulnerabilities, exhibiting unique patterns not found in developers' code. Agentic workflows also generate a number of vulnerabilities, particularly when given more autonomy. We find that vulnerabilities in LLM-generated patches are associated with distinctive code characteristics and are commonly observed in issues missing specific types of information. These results suggest that contextual factors play a critical role in the security of the generated patches and point toward the need for proactive risk assessment methods that account for both issue and code-level information.
Authors: Agnideep Aich, Ashit Baran Aich, Dipak C. Jain
Abstract: We propose \textbf{Temporal Conformal Prediction (TCP)}, a distribution-free framework for constructing well-calibrated prediction intervals in nonstationary time series. TCP couples a modern quantile forecaster with a split-conformal calibration layer on a rolling window and, in its \textbf{TCP-RM} variant, augments the conformal threshold with a single online Robbins-Monro (RM) offset to steer coverage toward a target level in real time. We benchmark TCP against GARCH, Historical Simulation, a rolling tree-based Quantile Regression (QR) model, a classical linear quantile regression baseline (QR-Linear), and an adaptive conformal method (ACI) across equities (S\&P 500), cryptocurrency (Bitcoin), and commodities (Gold). Three results are consistent across assets. First, both QR and QR-Linear yield the sharpest intervals but are materially under-calibrated, and even ACI remains below the nominal 95\% target in our full-sample backtests. Second, TCP (and TCP-RM) achieves near-nominal coverage across assets, with intervals that are wider than Historical Simulation in this evaluation (e.g., S\&P 500: 5.21 vs.\ 5.06). Third, the RM update changes calibration and width only marginally at our default hyperparameters. Crisis-window visualizations around March 2020 show TCP/TCP-RM expanding and then contracting their interval bands promptly as volatility spikes and recedes, with \textbf{red dots} marking days where realized returns fall outside the reported 95\% interval (miscoverage). A sensitivity study confirms robustness to window size and step-size choices. Overall, TCP provides a practical, theoretically grounded solution to calibrated uncertainty quantification under distribution shift, bridging statistical inference and machine learning for risk forecasting.
Authors: Shravya Kanchi, Neal Mangaokar, Aravind Cheruvu, Sifat Muhammad Abdullah, Shirin Nilizadeh, Atul Prakash, Bimal Viswanath
Abstract: Machine learning-based supervised classifiers are widely used for security tasks, and their improvement has been largely focused on algorithmic advancements. We argue that data challenges that negatively impact the performance of these classifiers have received limited attention. We address the following research question: Can developments in Generative AI (GenAI) address these data challenges and improve classifier performance? We propose augmenting training datasets with synthetic data generated using GenAI techniques to improve classifier generalization. We evaluate this approach across 7 diverse security tasks using 6 state-of-the-art GenAI methods and introduce a novel GenAI scheme called Nimai that enables highly controlled data synthesis. We find that GenAI techniques can significantly improve the performance of security classifiers, achieving improvements of up to 32.6% even in severely data-constrained settings (only ~180 training samples). Furthermore, we demonstrate that GenAI can facilitate rapid adaptation to concept drift post-deployment, requiring minimal labeling in the adjustment process. Despite successes, our study finds that some GenAI schemes struggle to initialize (train and produce data) on certain security tasks. We also identify characteristics of specific tasks, such as noisy labels, overlapping class distributions, and sparse feature vectors, which hinder performance boost using GenAI. We believe that our study will drive the development of future GenAI tools designed for security tasks.
Authors: Chenyu Su, Weiwei Shang, Chen Qian, Fei Zhang, Shuang Cong
Abstract: Fine-grained robotic manipulation requires grounding natural language into appropriate affordance targets. However, most existing methods driven by foundation models often compress rich semantics into oversimplified affordances, preventing exploitation of implicit semantic information. To address these challenges, we present ReSemAct, a novel unified manipulation framework that introduces Semantic Structuring and Affordance Refinement (SSAR), powered by the automated synergistic reasoning between Multimodal Large Language Models (MLLMs) and Vision Foundation Models (VFMs). Specifically, the Semantic Structuring module derives a unified semantic affordance description from natural language and RGB observations, organizing affordance regions, implicit functional intent, and coarse affordance anchors into a structured representation for downstream refinement. Building upon this specification, the Affordance Refinement strategy instantiates two complementary flows that separately specialize geometry and position, yielding fine-grained affordance targets. These refined targets are then encoded as real-time joint-space optimization objectives, enabling reactive and robust manipulation in dynamic environments. Extensive simulation and real-world experiments are conducted in semantically rich household and sparse chemical lab environments. The results demonstrate that ReSemAct performs diverse tasks under zero-shot conditions, showcasing the robustness of SSAR with foundation models in fine-grained manipulation. Code and videos at https://github.com/scy-v/ReSemAct and https://resemact.github.io.
URLs: https://github.com/scy-v/ReSemAct, https://resemact.github.io.
Authors: Sacha Braun, Eug\`ene Berta, Michael I. Jordan, Francis Bach
Abstract: While achieving exact conditional coverage in conformal prediction is unattainable without making strong, untestable regularity assumptions, the promise of conformal prediction hinges on finding approximations to conditional guarantees that are realizable in practice. A promising direction for obtaining conditional dependence for conformal sets--in particular capturing heteroskedasticity--is through estimating the conditional density $\mathbb{P}_{Y|X}$ and conformalizing its level sets. Previous work in this vein has focused on nonconformity scores based on the empirical cumulative distribution function (CDF). Such scores are, however, computationally costly, typically requiring expensive sampling methods. To avoid the need for sampling, we observe that the CDF-based score reduces to a Mahalanobis distance in the case of Gaussian scores, yielding a closed-form expression that can be directly conformalized. Moreover, the use of a Gaussian-based score opens the door to a number of extensions of the basic conformal method; in particular, we show how to construct conformal sets with missing output values, refine conformal sets as partial information about $Y$ becomes available, and construct conformal sets on transformations of the output space. Finally, empirical results indicate that our approach produces conformal sets that more closely approximate conditional coverage in multivariate settings compared to alternative methods.
Authors: Agnideep Aich, Ashit Baran Aich
Abstract: We study a simple statistic for benchmarking how well a sample preserves a known bivariate dependence structure. Given a target copula family (Clayton or Gumbel) and parameter $\theta_P$, the Copula Discrepancy (CD) compares the target Kendall's tau $\tau(\theta_P)$ with the Kendall's tau implied by a parameter $\hat{\theta}$ fitted to the sample within the target family, i.e., $|\tau(\theta_P)-\tau(\hat{\theta})|$. We develop a moment-based version, prove consistency, asymptotic normality, and robustness results under i.i.d.\ sampling, and use an MLE-based version empirically for greater power against tail-structure misspecification. Building on this, we define two information-theoretic copula summaries, a copula KL divergence (CKL) and a copula entropy gap (CED), and establish basic consistency and central limit results for their plug-in estimators. In controlled experiments, CD reliably separates on-target and off-target copulas with matched Kendall's $\tau$, provides a dependence-aware signal for tuning SGLD step sizes where Effective Sample Size favors overly aggressive (and biased) settings, and remains stably nonzero under deliberate tail-dependence mismatch where a naive $\tau$-based diagnostic fails; CKL and CED offer a complementary Shannon-style view that echoes these findings. Timing benchmarks show that both CD variants incur only millisecond-level overhead over the tested range and exhibit near-linear empirical scaling in sample size, providing a lightweight, dependence-focused complement to quadratic-cost omnibus discrepancies such as the Kernel Stein Discrepancy (KSD).
Authors: Jiayi Wang, Jacqueline Jil Vallon, Nikhil V. Kotha, Neil Panjwani, Xi Ling, Margaret Redfield, Sushmita Vij, Sandy Srinivas, John Leppert, Mark K. Buyyounouski, Mohsen Bayati
Abstract: Developing accurate clinical prediction models is often bottlenecked by the difficulty of deriving meaningful structured features from unstructured EHR notes, a process that traditionally requires manual, unscalable clinical abstraction. In this study, we first established a rigorous patient-level Clinician Feature Generation (CFG) protocol, in which domain experts manually reviewed notes to define and extract nuanced features for a cohort of 147 patients with prostate cancer. As a high-fidelity ground truth, this labor-intensive process provided the blueprint for SNOW (Scalable Note-to-Outcome Workflow), a transparent multi-agent large language model (LLM) system designed to autonomously mimic the iterative reasoning and validation workflow of clinical experts. On 5-year cancer recurrence prediction, SNOW (AUC-ROC 0.767) achieved performance comparable to manual CFG (0.762) and outperformed structured baselines, clinician-guided LLM extraction, and six representational feature generation (RFG) approaches. Once configured, SNOW produced the full patient-level feature table in 12 hours with 5 hours of clinician oversight, reducing human expert effort by approximately 48-fold versus manual CFG. To test scalability where manual CFG is infeasible, we deployed SNOW on an external heart failure with preserved ejection fraction (HFpEF) cohort from MIMIC-IV (n=2,084); without task-specific tuning, SNOW generated prognostic features that outperformed baseline and RFG methods for 30-day (SNOW: 0.851) and 1-year (SNOW: 0.763) mortality prediction. These results demonstrate that a modular LLM agent-based system can scale expert-level feature generation from clinical notes, while enabling interpretable use of unstructured EHR text in outcome prediction and preserving generalizability across a variety of settings and conditions.
Authors: G\'erard Ben Arous, Murat A. Erdogdu, Nuri Mert Vural, Denny Wu
Abstract: We study the optimization and sample complexity of gradient-based training of a two-layer neural network with quadratic activation function in the high-dimensional regime, where the data is generated as $y \propto \sum_{j=1}^{r}\lambda_j \sigma\left(\langle \boldsymbol{\theta_j}, \boldsymbol{x}\rangle\right), \boldsymbol{x} \sim N(0,\boldsymbol{I}_d)$, $\sigma$ is the 2nd Hermite polynomial, and $\lbrace\boldsymbol{\theta}_j \rbrace_{j=1}^{r} \subset \mathbb{R}^d$ are orthonormal signal directions. We consider the extensive-width regime $r \asymp d^\beta$ for $\beta \in [0, 1)$, and assume a power-law decay on the (non-negative) second-layer coefficients $\lambda_j\asymp j^{-\alpha}$ for $\alpha \geq 0$. We present a sharp analysis of the SGD dynamics in the feature learning regime, for both the population limit and the finite-sample (online) discretization, and derive scaling laws for the prediction risk that highlight the power-law dependencies on the optimization time, sample size, and model width. Our analysis combines a precise characterization of the associated matrix Riccati differential equation with novel matrix monotonicity arguments to establish convergence guarantees for the infinite-dimensional effective dynamics.
Authors: Kabir Khan, Priya Sharma, Arjun Mehta, Neha Gupta, Ravi Narayanan
Abstract: Large Language Models (LLMs) suffer from a critical limitation: their knowledge is static and quickly becomes outdated. Retraining these massive models is computationally prohibitive, while existing knowledge editing techniques can be slow and may introduce unforeseen side effects. To address this, we propose DySK-Attn, a novel framework that enables LLMs to efficiently integrate real-time knowledge from a dynamic external source. Our approach synergizes an LLM with a dynamic Knowledge Graph (KG) that can be updated instantaneously. The core of our framework is a sparse knowledge attention mechanism, which allows the LLM to perform a coarse-to-fine grained search, efficiently identifying and focusing on a small, highly relevant subset of facts from the vast KG. This mechanism avoids the high computational cost of dense attention over the entire knowledge base and mitigates noise from irrelevant information. We demonstrate through extensive experiments on time-sensitive question-answering tasks that DySK-Attn significantly outperforms strong baselines, including standard Retrieval-Augmented Generation (RAG) and model editing techniques, in both factual accuracy for updated knowledge and computational efficiency. Our framework offers a scalable and effective solution for building LLMs that can stay current with the ever-changing world.
Authors: Zhenyi Zhao, Muthu Rama Krishnan Mookiah, Emanuele Trucco
Abstract: Purpose: This study aims to introduce the first adaptation of RETFound for joint optic disc (OD) and optic cup (OC) segmentation. RETFound is a well-known foundation model developed for fundus camera and optical coherence tomography images, which has shown promising performance in disease diagnosis. Methods: We propose FunduSegmenter, a model integrating a series of novel modules with RETFound, including a Pre-adapter, a Decoder, a Post-adapter, skip connections with Convolutional Block Attention Module and a Vision Transformer block adapter. The model is evaluated on a private dataset, GoDARTS, and four public datasets, IDRiD, Drishti-GS, RIM-ONE-r3, and REFUGE, through internal verification, external verification and domain generalization experiments. Results: An average Dice similarity coefficient of 90.51% was achieved in internal verification, which substantially outperformed the baselines (nnU-Net: 82.91%; DUNet: 89.17%; TransUNet: 87.91%). In all external verification experiments, the average results were about 3% higher than those of the best baseline, and were also competitive in domain generalization. Conclusions: This study explored the potential of the latent general representations learned by RETFound for OD and OC segmentation in fundus camera images. Our FunduSegmenter outperformed nearly all state-of-the-art baseline methods. The proposed modules are general and can be extended to fine-tuning other foundation models. Translational Relevance: The model shows strong stability and generalization on both in-distribution and out-of-distribution data, providing stable OD and OC segmentation. This is an essential step for many automated tasks, from setting the accurate retinal coordinate to biomarker discovery. The code and all trained weights are available at: [link to be added after the paper is accepted]
Authors: Alireza Sedighi Moghaddam, Mohammad Reza Mohammadi
Abstract: Labeled data is a fundamental component in training supervised deep learning models for computer vision tasks. However, the labeling process, especially for ordinal image classification where class boundaries are often ambiguous, is prone to error and noise. Such label noise can significantly degrade the performance and reliability of machine learning models. This paper addresses the problem of detecting and correcting label noise in ordinal image classification tasks. To this end, a novel data-centric method called ORDinal Adaptive Correction (ORDAC) is proposed for adaptive correction of noisy labels. The proposed approach leverages the capabilities of Label Distribution Learning (LDL) to model the inherent ambiguity and uncertainty present in ordinal labels. During training, ORDAC dynamically adjusts the mean and standard deviation of the label distribution for each sample. Rather than discarding potentially noisy samples, this approach aims to correct them and make optimal use of the entire training dataset. The effectiveness of the proposed method is evaluated on benchmark datasets for age estimation (Adience) and disease severity detection (Diabetic Retinopathy) under various asymmetric Gaussian noise scenarios. Results show that ORDAC and its extended versions (ORDAC_C and ORDAC_R) lead to significant improvements in model performance. For instance, on the Adience dataset with 40% noise, ORDAC_R reduced the mean absolute error from 0.86 to 0.62 and increased the recall metric from 0.37 to 0.49. The method also demonstrated its effectiveness in correcting intrinsic noise present in the original datasets. This research indicates that adaptive label correction using label distributions is an effective strategy to enhance the robustness and accuracy of ordinal classification models in the presence of noisy data.
Authors: Angxiu Ni
Abstract: We derive the divergence-kernel formula for the linear response of random dynamical systems. Specifically, the pathwise expression is for the parameter-derivative of the marginal or stationary density, not an averaged observable. Our formula works for multiplicative and parameterized noise over any period of time; it does not require hyperbolicity. Then we derive a Monte-Carlo algorithm for linear responses. We develop a new framework of generative models, DK-SDE, where the model is a parameterized SDE, that (1) directly uses the KL divergence between the empirical data distribution and the marginal density of the SDE as the training objective, and (2) accommodates parametrizations in both drift and diffusion over a long time span, allowing prior structural knowledge to be incorporated explicitly. The optimization is done by gradient-descent enabled by the divergence-kernel method, which involves only forward processes and therefore substantially reduces memory cost. We demonstrate the new model on a 20-dimensional Lorenz system.
Authors: Weibin Li, Wendu Li, Quanying Liu
Abstract: Higher-order brain connectivity (HOBC), which captures interactions among three or more brain regions, provides richer organizational information than traditional pairwise functional connectivity (FC). Recent studies have begun to infer latent HOBC from noninvasive imaging data, but they mainly focus on static analyses, limiting their applicability in dynamic prediction tasks. To address this gap, we propose DCHO, a unified approach for modeling and forecasting the temporal evolution of HOBC based on a Decomposition-Composition framework, which is applicable to both non-predictive tasks (state classification) and predictive tasks (brain dynamics forecasting). DCHO adopts a decomposition-composition strategy that reformulates the prediction task into two manageable subproblems: HOBC inference and latent trajectory prediction. In the inference stage, we propose a dual-view encoder to extract multiscale topological features and a latent combinatorial learner to capture high-level HOBC information. In the forecasting stage, we introduce a latent-space prediction loss to enhance the modeling of temporal trajectories. Extensive experiments on multiple neuroimaging datasets demonstrate that DCHO achieves superior performance in both non-predictive tasks (state classification) and predictive tasks (brain dynamics forecasting), significantly outperforming existing methods.
Authors: Antonio Scardace, Lemuel Puglisi, Francesco Guarnera, Sebastiano Battiato, Daniele Rav\`i
Abstract: Deep generative models have emerged as a transformative tool in medical imaging, offering substantial potential for synthetic data generation. However, recent empirical studies highlight a critical vulnerability: these models can memorize sensitive training data, posing significant risks of unauthorized patient information disclosure. Detecting memorization in generative models remains particularly challenging, necessitating scalable methods capable of identifying training data leakage across large sets of generated samples. In this work, we propose DeepSSIM, a novel self-supervised metric for quantifying memorization in generative models. DeepSSIM is trained to: i) project images into a learned embedding space and ii) force the cosine similarity between embeddings to match the ground-truth SSIM (Structural Similarity Index) scores computed in the image space. To capture domain-specific anatomical features, training incorporates structure-preserving augmentations, allowing DeepSSIM to estimate similarity reliably without requiring precise spatial alignment. We evaluate DeepSSIM in a case study involving synthetic brain MRI data generated by a Latent Diffusion Model (LDM) trained under memorization-prone conditions, using 2,195 MRI scans from two publicly available datasets (IXI and CoRR). Compared to state-of-the-art memorization metrics, DeepSSIM achieves superior performance, improving F1 scores by an average of +52.03% over the best existing method. Code and data of our approach are publicly available at the following link: https://github.com/brAIn-science/DeepSSIM.
Authors: Thanh-Long V. Le, Myeongho Jeon, Kim Vu, Viet Lai, Eunho Yang
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) is a powerful framework for improving the reasoning abilities of Large Language Models (LLMs). However, current methods such as GRPO rely only on problems where the model responses to the same input differ in correctness, while ignoring those where all responses receive the same reward -- so-called zero-variance prompts. In this work, we argue that such prompts are not useless but can, in fact, provide meaningful feedback for policy optimization. To this end, we introduce RL with Zero-Variance Prompts (RL-ZVP), a novel algorithm that extract learning signals from zero-variance prompts. RL-ZVP directly rewards correctness and penalizes errors even without contrasting responses, modulating feedback with token-level characteristics to preserve informative, nuanced signals. Across six math reasoning benchmarks, RL-ZVP achieves significant improvements of up to 8.61 points in accuracy and 7.77 points in pass rate over GRPO, while consistently outperforming other baselines that filter out zero-variance prompts. These results highlight the untapped potential of learning from zero-variance prompts in RLVR.
Authors: Yuzhen Long, Songze Li
Abstract: Autonomous driving systems increasingly rely on multi-agent architectures powered by large language models (LLMs), where specialized agents collaborate to perceive, reason, and plan. A key component of these systems is the shared function library, a collection of software tools that agents use to process sensor data and navigate complex driving environments. Despite its critical role in agent decision-making, the function library remains an under-explored vulnerability. In this paper, we introduce FuncPoison, a novel poisoning-based attack targeting the function library to manipulate the behavior of LLM-driven multi-agent autonomous systems. FuncPoison exploits two key weaknesses in how agents access the function library: (1) agents rely on text-based instructions to select tools; and (2) these tools are activated using standardized command formats that attackers can replicate. By injecting malicious tools with deceptive instructions, FuncPoison manipulates one agent s decisions--such as misinterpreting road conditions--triggering cascading errors that mislead other agents in the system. We experimentally evaluate FuncPoison on two representative multi-agent autonomous driving systems, demonstrating its ability to significantly degrade trajectory accuracy, flexibly target specific agents to induce coordinated misbehavior, and evade diverse defense mechanisms. Our results reveal that the function library, often considered a simple toolset, can serve as a critical attack surface in LLM-based autonomous driving systems, raising elevated concerns on their reliability.
Authors: Massinissa Merouani, Afif Boudaoud, Riyadh Baghdadi
Abstract: The advancement of machine learning for compiler optimization, particularly within the polyhedral model, is constrained by the scarcity of large-scale, public performance datasets. This data bottleneck forces researchers to undertake costly data generation campaigns, slowing down innovation and hindering reproducible research learned code optimization. To address this gap, we introduce LOOPerSet, a new public dataset containing 28 million labeled data points derived from 220,000 unique, synthetically generated polyhedral programs. Each data point maps a program and a complex sequence of semantics-preserving transformations (such as fusion, skewing, tiling, and parallelism)to a ground truth performance measurement (execution time). The scale and diversity of LOOPerSet make it a valuable resource for training and evaluating learned cost models, benchmarking new model architectures, and exploring the frontiers of automated polyhedral scheduling. The dataset is released under a permissive license to foster reproducible research and lower the barrier to entry for data-driven compiler optimization.
Authors: Ashley Chen
Abstract: Autism spectrum disorder (ASD) has been associated with structural alterations across cortical and subcortical regions. Quantitative neuroimaging enables large-scale analysis of these neuroanatomical patterns. This project used structural MRI (T1-weighted) data from the publicly available ABIDE I dataset (n = 1,112) to classify ASD and control participants using a hybrid model. A 3D convolutional neural network (CNN) was trained to learn neuroanatomical feature representations, which were then passed to a support vector machine (SVM) for final classification. Gradient-weighted class activation mapping (Grad-CAM) was applied to the CNN to visualize the brain regions that contributed most to the model predictions. The Grad-CAM difference maps showed strongest relevance along cortical boundary regions, with additional emphasis in midline frontal-temporal-parietal areas, which is broadly consistent with prior ASD neuroimaging findings.
Authors: Jishu Adhikary, Raju Maiti
Abstract: Rainfall forecasting plays a critical role in climate adaptation, agriculture, and water resource management. This study develops long-term forecasts of monthly rainfall across 19 districts of West Bengal using a century-scale dataset spanning 1900-2019. Daily rainfall records are aggregated into monthly series, resulting in 120 years of observations for each district. The forecasting task involves predicting the next 108 months (9 years, 2011-2019) while accounting for temporal dependencies and spatial interactions among districts. To address the nonlinear and complex structure of rainfall dynamics, we propose a hierarchical modeling framework that combines regression-based forecasting of yearly features with multi-layer perceptrons (MLPs) for monthly prediction. Yearly features, such as annual totals, quarterly proportions, variability measures, skewness, and extremes, are first forecasted using regression models that incorporate both own lags and neighboring-district lags. These forecasts are then integrated as auxiliary inputs into an MLP model, which captures nonlinear temporal patterns and spatial dependencies in the monthly series. The results demonstrate that the hierarchical regression-MLP architecture provides robust long-term spatio-temporal forecasts, offering valuable insights for agriculture, irrigation planning, and water conservation strategies.
Authors: Quan Nguyen-Tri, Mukul Ranjan, Zhiqiang Shen
Abstract: This work studies how to adaptively recompute key-value (KV) caches for diffusion large language models (DLMs) to maximize prediction accuracy while minimizing decoding latency. Prior methods' decoders recompute QKV for all tokens at every denoising step and layer, despite KV states changing little across most steps, especially in shallow layers, leading to substantial redundancy. We make three observations: (1) distant ${\bf MASK}$ tokens primarily act as a length-bias and can be cached block-wise beyond the active prediction window; (2) KV dynamics increase with depth, suggesting that selective refresh starting from deeper layers is sufficient; and (3) the most-attended token exhibits the smallest KV drift, providing a conservative lower bound on cache change for other tokens. Building on these, we propose ${\bf Elastic-Cache}$, a training-free, architecture-agnostic strategy that jointly decides ${when}$ to refresh (via an attention-aware drift test on the most-attended token) and ${where}$ to refresh (via a depth-aware schedule that recomputes from a chosen layer onward while reusing shallow-layer caches and off-window MASK caches). Unlike fixed-period schemes, Elastic-Cache performs adaptive, layer-aware cache updates for diffusion LLMs, reducing redundant computation and accelerating decoding with negligible loss in generation quality. Experiments on LLaDA-Instruct, LLaDA-1.5, and LLaDA-V across mathematical reasoning and code generation tasks demonstrate consistent speedups: $8.7\times$ on GSM8K (256 tokens), and $45.1\times$ on longer sequences, while consistently maintaining higher accuracy than the baseline. Our method achieves significantly higher throughput ($6.8\times$ on GSM8K) than existing confidence-based approaches while preserving generation quality, enabling practical deployment of diffusion LLMs.
Authors: Massinissa Merouani, Islem Kara Bernou, Riyadh Baghdadi
Abstract: Automatic code optimization remains a difficult challenge, particularly for complex loop nests on modern hardware. This paper investigates a novel approach to code optimization where Large Language Models (LLMs) guide the process through a closed-loop interaction with a compiler. We present ComPilot, an experimental framework that leverages off-the-shelf LLMs, without any task-specific fine-tuning, as interactive optimization agents. ComPilot establishes a feedback loop where an LLM proposes transformations for a given loop nest to a compiler. The compiler attempts the transformations, reporting back legality status and measured speedup or slowdown. The LLM utilizes this concrete feedback to iteratively refine its optimization strategy. Our extensive evaluation across the PolyBench benchmark suite demonstrates the effectiveness of this zero-shot approach. ComPilot achieves geometric mean speedups of 2.66x (single run) and 3.54x (best-of-5 runs) over the original code. Furthermore, ComPilot demonstrates competitive performance against the state-of-the-art Pluto polyhedral optimizer, outperforming it in many cases. This experimental study demonstrates that general-purpose LLMs can effectively guide the code optimization process when grounded by compiler feedback, opening promising research directions for agentic AI in code optimization.
Authors: Yoonjae Seo, Ermal Elbasani, Jaehong Lee
Abstract: Most neural network quantization methods apply uniform bit precision across spatial regions, disregarding the heterogeneous complexity inherent in visual data. This paper introduces MCAQ-YOLO, a practical framework for tile-wise spatial mixed-precision quantization in real-time object detectors. Morphological complexity--quantified through five complementary metrics (fractal dimension, texture entropy, gradient variance, edge density, and contour complexity)--is proposed as a signal-centric predictor of spatial quantization sensitivity. A calibration-time analysis design enables spatial bit allocation with only 0.3ms inference overhead, achieving 151 FPS throughput. Additionally, a curriculum-based training scheme that progressively increases quantization difficulty is introduced to stabilize optimization and accelerate convergence. On a construction safety equipment dataset exhibiting high morphological variability, MCAQ-YOLO achieves 85.6% mAP@0.5 with an average bit-width of 4.2 bits and a 7.6x compression ratio, outperforming uniform 4-bit quantization by 3.5 percentage points. Cross-dataset evaluation on COCO 2017 (+2.9%) and Pascal VOC 2012 (+2.3%) demonstrates consistent improvements, with performance gains correlating with within-image complexity variation.
Authors: Yuanchao Wang, Tian Qin, Eduardo Valle, Bruno Abrahao
Abstract: Out-of-distribution (OOD) detection is critical for deploying image classifiers in safety-sensitive environments, yet existing detectors often struggle when OOD samples are semantically similar to the in-distribution (ID) classes. We present BootOOD, a fully self-supervised OOD detection framework that bootstraps exclusively from ID data and is explicitly designed to handle semantically challenging OOD samples. BootOOD synthesizes pseudo-OOD features through simple transformations of ID representations and leverages Neural Collapse (NC), where ID features cluster tightly around class means with consistent feature norms. Unlike prior approaches that aim to constrain OOD features into subspaces orthogonal to the collapsed ID means, BootOOD introduces a lightweight auxiliary head that performs radius-based classification on feature norms. This design decouples OOD detection from the primary classifier and imposes a relaxed requirement: OOD samples are learned to have smaller feature norms than ID features, which is easier to satisfy when ID and OOD are semantically close. Experiments on CIFAR-10, CIFAR-100, and ImageNet-200 show that BootOOD outperforms prior post-hoc methods, surpasses training-based methods without outlier exposure, and is competitive with state-of-the-art outlier-exposure approaches while maintaining or improving ID accuracy.
Authors: Samuele Dell'Erba, Andrew D. Bagdanov
Abstract: Diffusion models have established the state-of-the-art in text-to-image generation, but their performance often relies on a diffusion prior network to translate text embeddings into the visual manifold for easier decoding. These priors are computationally expensive and require extensive training on massive datasets. In this work, we challenge the necessity of a trained prior at all by employing Optimization-based Visual Inversion (OVI), a training-free and zero-shot alternative, to replace the need for a prior. OVI initializes a latent visual representation from random pseudo-tokens and iteratively optimizes it to maximize the cosine similarity with the input textual prompt embedding. We further propose two novel constraints, a Mahalanobis-based and a Nearest-Neighbor loss, to regularize the OVI optimization process toward the distribution of realistic images. Our experiments, conducted on Kandinsky 2.2, show that OVI can serve as an alternative to traditional priors. More importantly, our analysis reveals a critical flaw in current evaluation benchmarks like T2I-CompBench++, where simply using the text embedding as a prior achieves surprisingly high scores, despite lower perceptual quality. Our constrained OVI methods improve visual fidelity over this baseline, with the Nearest-Neighbor approach proving particularly effective. It achieves quantitative scores comparable to or higher than the state-of-the-art data-efficient prior, underscoring the potential of optimization-based strategies as viable, training-free alternatives to traditional priors. The code will be publicly available upon acceptance.
Authors: Yesid Fonseca, Manuel S. R\'ios, Nicanor Quijano, Luis F. Giraldo
Abstract: Patch foraging involves the deliberate and planned process of determining the optimal time to depart from a resource-rich region and investigate potentially more beneficial alternatives. The Marginal Value Theorem (MVT) is frequently used to characterize this process, offering an optimality model for such foraging behaviors. Although this model has been widely used to make predictions in behavioral ecology, discovering the computational mechanisms that facilitate the emergence of optimal patch-foraging decisions in biological foragers remains under investigation. Here, we show that artificial foragers equipped with learned world models naturally converge to MVT-aligned strategies. Using a model-based reinforcement learning agent that acquires a parsimonious predictive representation of its environment, we demonstrate that anticipatory capabilities, rather than reward maximization alone, drive efficient patch-leaving behavior. Compared with standard model-free RL agents, these model-based agents exhibit decision patterns similar to many of their biological counterparts, suggesting that predictive world models can serve as a foundation for more explainable and biologically grounded decision-making in AI systems. Overall, our findings highlight the value of ecological optimality principles for advancing interpretable and adaptive AI.
Authors: Mahsa Lavaei, Zahra Abadi, Salar Beigzad, Alireza Maleki
Abstract: Medical image classification is a critical task in healthcare, enabling accurate and timely diagnosis. However, deploying deep learning models on resource-constrained edge devices presents significant challenges due to computational and memory limitations. This research investigates a resource-efficient approach to medical image classification by employing model quantization techniques. Quantization reduces the precision of model parameters and activations, significantly lowering computational overhead and memory requirements without sacrificing classification accuracy. The study focuses on the optimization of quantization-aware training (QAT) and post-training quantization (PTQ) methods tailored for edge devices, analyzing their impact on model performance across medical imaging datasets. Experimental results demonstrate that quantized models achieve substantial reductions in model size and inference latency, enabling real-time processing on edge hardware while maintaining clinically acceptable diagnostic accuracy. This work provides a practical pathway for deploying AI-driven medical diagnostics in remote and resource-limited settings, enhancing the accessibility and scalability of healthcare technologies.
Authors: Shengdu Chai, Chen Lin, Xinyang Dong, Yuqiang Li, Wanli Ouyang, Lei Wang, X. C. Xie
Abstract: The crystal structure of high-pressure solid hydrogen remains a fundamental open problem. Although the research frontier has mostly shifted toward ultra-high pressure phases above 400 GPa, we show that even the broken symmetry phase observed around 130~GPa requires revisiting due to its intricate coupling of electronic and nuclear degrees of freedom. Here, we develop a first principle quantum Monte Carlo framework based on a deep neural network wave function that treats both electrons and nuclei quantum mechanically within the constant pressure ensemble. Our calculations reveal an unreported ground-state structure candidate for the broken symmetry phase with $Cmcm$ space group symmetry, and we test its stability up to 96 atoms. The predicted structure quantitatively matches the experimental equation of state and X-ray diffraction patterns. Furthermore, our group-theoretical analysis shows that the $Cmcm$ structure is compatible with existing Raman and infrared spectroscopic data. Crucially, static density functional theory calculation reveals the $Cmcm$ structure as a dynamically unstable saddle point on the Born-Oppenheimer potential energy surface, demonstrating that a full quantum many-body treatment of the problem is necessary. These results shed new light on the phase diagram of high-pressure hydrogen and call for further experimental verifications.
Authors: Christopher Regan, Ying Xie
Abstract: We introduce obfuscation testing, a novel methodology for validating whether large language models detect structural market patterns through causal reasoning rather than temporal association. Testing three dealer hedging constraint patterns (gamma positioning, stock pinning, 0DTE hedging) on 242 trading days (95.6% coverage) of S&P 500 options data, we find LLMs achieve 71.5% detection rate using unbiased prompts that provide only raw gamma exposure values without regime labels or temporal context. The WHO-WHOM-WHAT causal framework forces models to identify the economic actors (dealers), affected parties (directional traders), and structural mechanisms (forced hedging) underlying observed market dynamics. Critically, detection accuracy (91.2%) remains stable even as economic profitability varies quarterly, demonstrating that models identify structural constraints rather than profitable patterns. When prompted with regime labels, detection increases to 100%, but the 71.5% unbiased rate validates genuine pattern recognition. Our findings suggest LLMs possess emergent capabilities for detecting complex financial mechanisms through pure structural reasoning, with implications for systematic strategy development, risk management, and our understanding of how transformer architectures process financial market dynamics.
Authors: Xiaolong Wu, Qifeng Liao
Abstract: Solving high-dimensional Fokker-Planck (FP) equations is a challenge in computational physics and stochastic dynamics, due to the curse of dimensionality (CoD) and the bottleneck of evaluating second-order diffusion terms. Existing deep learning approaches, such as Physics-Informed Neural Networks, face computational challenges as dimensionality increases, driven by the $O(d^2)$ complexity of automatic differentiation for second-order derivatives. While recent probability flow approaches bypass this by learning score functions or matching velocity fields, they often involve serial operations or depend on sampling efficiency in complex distributions. To address these issues, we propose the Adaptive Probability Flow Residual Minimization (A-PFRM) method. We reformulate the second-order FP equation into an equivalent first-order deterministic Probability Flow ODE (PF-ODE) constraint, which avoids explicit Hessian computation. Unlike score matching or velocity matching, A-PFRM solves this problem by minimizing the residual of the continuity equation induced by the PF-ODE. We leverage Continuous Normalizing Flows combined with the Hutchinson Trace Estimator to reduce the training complexity to linear scale $O(d)$, achieving an effective $O(1)$ wall-clock time on GPUs. To address data sparsity in high dimensions, we apply a generative adaptive sampling strategy and theoretically prove that dynamically aligning collocation points with the evolving probability mass is a necessary condition to bound the approximation error. Experiments on diverse benchmarks -- ranging from anisotropic Ornstein-Uhlenbeck (OU) processes and high-dimensional Brownian motions with time-varying diffusion terms, to Geometric OU processes featuring non-Gaussian solutions -- demonstrate that A-PFRM effectively mitigates the CoD, maintaining high accuracy and constant temporal cost for problems up to 100 dimensions.
Authors: Sameera K. M., Serena Nicolazzo, Antonino Nocera, Vinod P., Rafidha Rehiman K. A
Abstract: Federated Learning (FL) has recently emerged as a revolutionary approach to collaborative training Machine Learning models. In particular, it enables decentralized model training while preserving data privacy, but its distributed nature makes it highly vulnerable to a severe attack known as Data Poisoning. In such scenarios, malicious clients inject manipulated data into the training process, thereby degrading global model performance or causing targeted misclassification. In this paper, we present a novel defense mechanism called GShield, designed to detect and mitigate malicious and low-quality updates, especially under non-independent and identically distributed (non-IID) data scenarios. GShield operates by learning the distribution of benign gradients through clustering and Gaussian modeling during an initial round, enabling it to establish a reliable baseline of trusted client behavior. With this benign profile, GShield selectively aggregates only those updates that align with the expected gradient patterns, effectively isolating adversarial clients and preserving the integrity of the global model. An extensive experimental campaign demonstrates that our proposed defense significantly improves model robustness compared to the state-of-the-art methods while maintaining a high accuracy of performance across both tabular and image datasets. Furthermore, GShield improves the accuracy of the targeted class by 43\% to 65\% after detecting malicious and low-quality clients.
Authors: Ilias Aarab
Abstract: A growing empirical literature suggests that equity-premium predictability is state dependent, with much of the forecasting power concentrated around recessionary periods (Henkel et al., 2011; Dangl and Halling, 2012; Devpura et al., 2018). I study U.S. stock return predictability across economic regimes and document strong evidence of time-varying expected returns across both expansionary and contractionary states. I contribute in two ways. First, I introduce a state-switching predictive regression in which the market state is defined in real time using the slope of the yield curve. Relative to the standard one-state predictive regression, the state-switching specification increases both in-sample and out-of-sample performance for the set of popular predictors considered by Welch and Goyal (2008), improving the out-of-sample performance of most predictors in economically meaningful ways. Second, I propose a new aggregate predictor, the Aligned Economic Index, constructed via partial least squares (PLS). Under the state-switching model, the Aligned Economic Index exhibits statistically and economically significant predictive power in sample and out of sample, and it outperforms widely used benchmark predictors and alternative predictor-combination methods.
Authors: Haipeng Luo, Huawen Feng, Qingfeng Sun, Can Xu, Kai Zheng, Yufei Wang, Tao Yang, Han Hu, Yansong Tang, Di Wang
Abstract: Large Reasoning Models (LRMs) like o3 and DeepSeek-R1 have achieved remarkable progress in natural language reasoning with long chain-of-thought. However, they remain computationally inefficient and struggle with accuracy when solving problems requiring complex mathematical operations. In this work, we present AgentMath, an agent framework that seamlessly integrates language models' reasoning capabilities with code interpreters' computational precision to efficiently tackle complex mathematical problems. Our approach introduces three key innovations: (1) An automated method that converts natural language chain-of-thought into structured tool-augmented trajectories, generating high-quality supervised fine-tuning (SFT) data to alleviate data scarcity; (2) A novel agentic reinforcement learning (RL) paradigm that dynamically interleaves natural language generation with real-time code execution. This enables models to autonomously learn optimal tool-use strategies through multi-round interactive feedback, while fostering emergent capabilities in code refinement and error correction; (3) An efficient training system incorporating innovative techniques, including request-level asynchronous rollout scheduling, agentic partial rollout, and prefix-aware weighted load balancing, achieving 4-5x speedup and making efficient RL training feasible on ultra-long sequences with scenarios with massive tool invocation. The evaluations show that AgentMath achieves state-of-the-art performance on challenging mathematical competition benchmarks including AIME24, AIME25, and HMMT25. Specifically, AgentMath-30B-A3B attains 90.6%, 86.4%, and 73.8% accuracy respectively, achieving advanced performance. The results validate the effectiveness of our approach and pave the way for building more sophisticated and scalable mathematical reasoning agents.