Authors: Saif Ur Rehman, Wajid Yousuf
Abstract: The accurate solution of nonlinear hyperbolic partial differential equations (PDEs) remains a central challenge in computational science due to the presence of steep gradients, discontinuities, and multiscale structures that make conventional discretization-based solvers computationally demanding. Physics-Informed Neural Networks (PINNs) embed the governing equations into the learning process, enabling mesh-free solution of PDEs, yet they often struggle to capture steep gradients, discontinuities, and complex nonlinear wave interactions. To address these limitations, this study employs the Extended Physics-Informed Neural Network (XPINN) framework to solve the nonlinear Buckley-Leverett equation with a nonconvex flux function, which models immiscible two-phase flow in porous media. The computational domain is dynamically decomposed in space and time into evolving pre-shock and post-shock regions, allowing localized subnetworks to efficiently learn distinct flow behaviors. Coupling between subnetworks is achieved through the Rankine-Hugoniot jump condition, which enforces physically consistent flux continuity across the moving shock interface. Numerical experiments demonstrate that the proposed XPINN approach accurately captures discontinuous saturation fronts and compound wave interactions without requiring artificial diffusion or entropy corrections. Compared to standard PINNs, the XPINN framework achieves superior stability, faster convergence, and enhanced resolution of nonlinear wave dynamics using smaller, domain-specific models with fewer trainable parameters, establishing it as an effective and scalable tool for solving challenging hyperbolic PDEs in multiphase flow problems. The code of this work is available on github.com/saifkhanengr/XPINN-for-Buckley-Leverett.
Authors: Silin Zhou, Yao Chen, Shuo Shang, Lisi Chen, Bingsheng He, Ryosuke Shibasaki
Abstract: Trajectory representation learning (TRL) maps trajectories to vector embeddings and facilitates tasks such as trajectory classification and similarity search. State-of-the-art (SOTA) TRL methods transform raw GPS trajectories to grid or road trajectories to capture high-level travel semantics, i.e., regions and roads. However, they lose fine-grained spatial-temporal details as multiple GPS points are grouped into a single grid cell or road segment. To tackle this problem, we propose the BLUrred Encoding method, dubbed BLUE, which gradually reduces the precision of GPS coordinates to create hierarchical patches with multiple levels. The low-level patches are small and preserve fine-grained spatial-temporal details, while the high-level patches are large and capture overall travel patterns. To complement different patch levels with each other, our BLUE is an encoder-decoder model with a pyramid structure. At each patch level, a Transformer is used to learn the trajectory embedding at the current level, while pooling prepares inputs for the higher level in the encoder, and up-resolution provides guidance for the lower level in the decoder. BLUE is trained using the trajectory reconstruction task with the MSE loss. We compare BLUE with 8 SOTA TRL methods for 3 downstream tasks, the results show that BLUE consistently achieves higher accuracy than all baselines, outperforming the best-performing baselines by an average of 30.90%. Our code is available at https://github.com/slzhou-xy/BLUE.
Authors: Ci Lin, Tet Yeap, Iluju Kiringa, Biwei Zhang
Abstract: Deep neural networks are known to be vulnerable to adversarial perturbations, which are small and carefully crafted inputs that lead to incorrect predictions. In this paper, we propose DeepDefense, a novel defense framework that applies Gradient-Feature Alignment (GFA) regularization across multiple layers to suppress adversarial vulnerability. By aligning input gradients with internal feature representations, DeepDefense promotes a smoother loss landscape in tangential directions, thereby reducing the model's sensitivity to adversarial noise. We provide theoretical insights into how adversarial perturbation can be decomposed into radial and tangential components and demonstrate that alignment suppresses loss variation in tangential directions, where most attacks are effective. Empirically, our method achieves significant improvements in robustness across both gradient-based and optimization-based attacks. For example, on CIFAR-10, CNN models trained with DeepDefense outperform standard adversarial training by up to 15.2% under APGD attacks and 24.7% under FGSM attacks. Against optimization-based attacks such as DeepFool and EADEN, DeepDefense requires 20 to 30 times higher perturbation magnitudes to cause misclassification, indicating stronger decision boundaries and a flatter loss landscape. Our approach is architecture-agnostic, simple to implement, and highly effective, offering a promising direction for improving the adversarial robustness of deep learning models.
Authors: E. Zhixuan Zeng, Yuhao Chen, Alexander Wong
Abstract: Image generation models frequently encode social biases, including stereotypes tied to gender, race, and profession. Existing methods for analyzing these biases in diffusion models either focus narrowly on predefined categories or depend on manual interpretation of latent directions. These constraints limit scalability and hinder the discovery of subtle or unanticipated patterns. We introduce SCALEX, a framework for scalable and automated exploration of diffusion model latent spaces. SCALEX extracts semantically meaningful directions from H-space using only natural language prompts, enabling zero-shot interpretation without retraining or labelling. This allows systematic comparison across arbitrary concepts and large-scale discovery of internal model associations. We show that SCALEX detects gender bias in profession prompts, ranks semantic alignment across identity descriptors, and reveals clustered conceptual structure without supervision. By linking prompts to latent directions directly, SCALEX makes bias analysis in diffusion models more scalable, interpretable, and extensible than prior approaches.
Authors: Abdullah Al Shiam, Md. Khademul Islam Molla, Abu Saleh Musa Miah, Md. Abdus Samad Kamal
Abstract: A Brain Computer Interface (BCI) connects the human brain to the outside world, providing a direct communication channel. Electroencephalography (EEG) signals are commonly used in BCIs to reflect cognitive patterns related to motor function activities. However, due to the multichannel nature of EEG signals, explicit information processing is crucial to lessen computational complexity in BCI systems. This study proposes an innovative method based on brain region-specific channel selection and multi-domain feature fusion to improve classification accuracy. The novelty of the proposed approach lies in region-based channel selection, where EEG channels are grouped according to their functional relevance to distinct brain regions. By selecting channels based on specific regions involved in motor imagery (MI) tasks, this technique eliminates irrelevant channels, reducing data dimensionality and improving computational efficiency. This also ensures that the extracted features are more reflective of the brain actual activity related to motor tasks. Three distinct feature extraction methods Common Spatial Pattern (CSP), Fuzzy C-means clustering, and Tangent Space Mapping (TSM), are applied to each group of channels based on their brain region. Each method targets different characteristics of the EEG signal: CSP focuses on spatial patterns, Fuzzy C means identifies clusters within the data, and TSM captures non-linear patterns in the signal. The combined feature vector is used to classify motor imagery tasks (left hand, right hand, and right foot) using Support Vector Machine (SVM). The proposed method was validated on publicly available benchmark EEG datasets (IVA and I) from the BCI competition III and IV. The results show that the approach outperforms existing methods, achieving classification accuracies of 90.77% and 84.50% for datasets IVA and I, respectively.
Authors: Feilong Wang, Fuqiang Liu
Abstract: The integration of large language models (LLMs) into automated driving systems has opened new possibilities for reasoning and decision-making by transforming complex driving contexts into language-understandable representations. Recent studies demonstrate that fine-tuned LLMs can accurately predict vehicle trajectories and lane-change intentions by gathering and transforming data from surrounding vehicles. However, the robustness of such LLM-based prediction models for safety-critical driving systems remains unexplored, despite the increasing concerns about the trustworthiness of LLMs. This study addresses this gap by conducting a systematic vulnerability analysis of LLM-enabled vehicle trajectory prediction. We propose a one-feature differential evolution attack that perturbs a single kinematic feature of surrounding vehicles within the LLM's input prompts under a black-box setting. Experiments on the highD dataset reveal that even minor, physically plausible perturbations can significantly disrupt model outputs, underscoring the susceptibility of LLM-based predictors to adversarial manipulation. Further analyses reveal a trade-off between accuracy and robustness, examine the failure mechanism, and explore potential mitigation solutions. The findings provide the very first insights into adversarial vulnerabilities of LLM-driven automated vehicle models in the context of vehicular interactions and highlight the need for robustness-oriented design in future LLM-based intelligent transportation systems.
Authors: Zhe Yang, Wenrui Li, Hongtao Chen, Penghong Wang, Ruiqin Xiong, Xiaopeng Fan
Abstract: Multimodal learning aims to improve performance by leveraging data from multiple sources. During joint multimodal training, due to modality bias, the advantaged modality often dominates backpropagation, leading to imbalanced optimization. Existing methods still face two problems: First, the long-term dominance of the dominant modality weakens representation-output coupling in the late stages of training, resulting in the accumulation of redundant information. Second, previous methods often directly and uniformly adjust the gradients of the advantaged modality, ignoring the semantics and directionality between modalities. To address these limitations, we propose Adaptive Redundancy Regulation for Balanced Multimodal Information Refinement (RedReg), which is inspired by information bottleneck principle. Specifically, we construct a redundancy phase monitor that uses a joint criterion of effective gain growth rate and redundancy to trigger intervention only when redundancy is high. Furthermore, we design a co-information gating mechanism to estimate the contribution of the current dominant modality based on cross-modal semantics. When the task primarily relies on a single modality, the suppression term is automatically disabled to preserve modality-specific information. Finally, we project the gradient of the dominant modality onto the orthogonal complement of the joint multimodal gradient subspace and suppress the gradient according to redundancy. Experiments show that our method demonstrates superiority among current major methods in most scenarios. Ablation experiments verify the effectiveness of our method. The code is available at https://github.com/xia-zhe/RedReg.git
Authors: Niklas Erdmann, Lars Bentsen, Roy Stenbro, Heine Nygard Riise, Narada Dilp Warakagoda, Paal E. Engelstad
Abstract: Probabilistic forecasting is not only a way to add more information to a prediction of the future, but it also builds on weaknesses in point prediction. Sudden changes in a time series can still be captured by a cumulative distribution function (CDF), while a point prediction is likely to miss it entirely. The modeling of CDFs within forecasts has historically been limited to parametric approaches, but due to recent advances, this no longer has to be the case. We aim to advance the fields of probabilistic forecasting and monotonic networks by connecting them and propose an approach that permits the forecasting of implicit, complete, and nonparametric CDFs. For this purpose, we propose an adaptation to deep lattice networks (DLN) for monotonically constrained simultaneous/implicit quantile regression in time series forecasting. Quantile regression usually produces quantile crossovers, which need to be prevented to achieve a legitimate CDF. By leveraging long short term memory units (LSTM) as the embedding layer, and spreading quantile inputs to all sub-lattices of a DLN with an extended output size, we can produce a multi-horizon forecast of an implicit CDF due to the monotonic constraintability of DLNs that prevent quantile crossovers. We compare and evaluate our approach's performance to relevant state of the art within the context of a highly relevant application of time series forecasting: Day-ahead, hourly forecasts of solar irradiance observations. Our experiments show that the adaptation of a DLN performs just as well or even better than an unconstrained approach. Further comparison of the adapted DLN against a scalable monotonic neural network shows that our approach performs better. With this adaptation of DLNs, we intend to create more interest and crossover investigations in techniques of monotonic neural networks and probabilistic forecasting.
Authors: Xiuding Cai, Xueyao Wang, Sen Wang, Yaoyao Zhu, Jiao Chen, Yu Yao
Abstract: Intraoperative monitoring and prediction of vital signs are critical for ensuring patient safety and improving surgical outcomes. Despite recent advances in deep learning models for medical time-series forecasting, several challenges persist, including the lack of standardized benchmarks, incomplete data, and limited cross-center validation. To address these challenges, we introduce VitalBench, a novel benchmark specifically designed for intraoperative vital sign prediction. VitalBench includes data from over 4,000 surgeries across two independent medical centers, offering three evaluation tracks: complete data, incomplete data, and cross-center generalization. This framework reflects the real-world complexities of clinical practice, minimizing reliance on extensive preprocessing and incorporating masked loss techniques for robust and unbiased model evaluation. By providing a standardized and unified platform for model development and comparison, VitalBench enables researchers to focus on architectural innovation while ensuring consistency in data handling. This work lays the foundation for advancing predictive models for intraoperative vital sign forecasting, ensuring that these models are not only accurate but also robust and adaptable across diverse clinical environments. Our code and data are available at https://github.com/XiudingCai/VitalBench.
Authors: Jun-Hyoung Park, Ho-Jun Song, Seong-Whan Lee
Abstract: Deep learning-based molecular generation models have shown great potential in efficiently exploring vast chemical spaces by generating potential drug candidates with desired properties. However, these models often produce chemically invalid molecules, which limits the usable scope of the learned chemical space and poses significant challenges for practical applications. To address this issue, we propose ChemFixer, a framework designed to correct invalid molecules into valid ones. ChemFixer is built on a transformer architecture, pre-trained using masking techniques, and fine-tuned on a large-scale dataset of valid/invalid molecular pairs that we constructed. Through comprehensive evaluations across diverse generative models, ChemFixer improved molecular validity while effectively preserving the chemical and biological distributional properties of the original outputs. This indicates that ChemFixer can recover molecules that could not be previously generated, thereby expanding the diversity of potential drug candidates. Furthermore, ChemFixer was effectively applied to a drug-target interaction (DTI) prediction task using limited data, improving the validity of generated ligands and discovering promising ligand-protein pairs. These results suggest that ChemFixer is not only effective in data-limited scenarios, but also extensible to a wide range of downstream tasks. Taken together, ChemFixer shows promise as a practical tool for various stages of deep learning-based drug discovery, enhancing molecular validity and expanding accessible chemical space.
Authors: Han Wang, Deyi Ji, Junyu Lu, Lanyun Zhu, Hailong Zhang, Haiyang Wu, Liqun Liu, Peng Shu, Roy Ka-Wei Lee
Abstract: Accurate detection of offensive content on social media demands high-quality labeled data; however, such data is often scarce due to the low prevalence of offensive instances and the high cost of manual annotation. To address this low-resource challenge, we propose a self-training framework that leverages abundant unlabeled data through collaborative pseudo-labeling. Starting with a lightweight classifier trained on limited labeled data, our method iteratively assigns pseudo-labels to unlabeled instances with the support of Multi-Agent Vision-Language Models (MA-VLMs). Un-labeled data on which the classifier and MA-VLMs agree are designated as the Agreed-Unknown set, while conflicting samples form the Disagreed-Unknown set. To enhance label reliability, MA-VLMs simulate dual perspectives, moderator and user, capturing both regulatory and subjective viewpoints. The classifier is optimized using a novel Positive-Negative-Unlabeled (PNU) loss, which jointly exploits labeled, Agreed-Unknown, and Disagreed-Unknown data while mitigating pseudo-label noise. Experiments on benchmark datasets demonstrate that our framework substantially outperforms baselines under limited supervision and approaches the performance of large-scale models
Authors: Xiao Fan, Jingyan Jiang, Zhaoru Chen, Fanding Huang, Xiao Chen, Qinting Jiang, Bowen Zhang, Xing Tang, Zhi Wang
Abstract: Test-Time adaptation (TTA) has proven effective in mitigating performance drops under single-domain distribution shifts by updating model parameters during inference. However, real-world deployments often involve mixed distribution shifts, where test samples are affected by diverse and potentially conflicting domain factors, posing significant challenges even for SOTA TTA methods. A key limitation in existing approaches is their reliance on a unified adaptation path, which fails to account for the fact that optimal gradient directions can vary significantly across different domains. Moreover, current benchmarks focus only on synthetic or homogeneous shifts, failing to capture the complexity of real-world heterogeneous mixed distribution shifts. To address this, we propose MoETTA, a novel entropy-based TTA framework that integrates the Mixture-of-Experts (MoE) architecture. Rather than enforcing a single parameter update rule for all test samples, MoETTA introduces a set of structurally decoupled experts, enabling adaptation along diverse gradient directions. This design allows the model to better accommodate heterogeneous shifts through flexible and disentangled parameter updates. To simulate realistic deployment conditions, we introduce two new benchmarks: potpourri and potpourri+. While classical settings focus solely on synthetic corruptions, potpourri encompasses a broader range of domain shifts--including natural, artistic, and adversarial distortions--capturing more realistic deployment challenges. Additionally, potpourri+ further includes source-domain samples to evaluate robustness against catastrophic forgetting. Extensive experiments across three mixed distribution shifts settings show that MoETTA consistently outperforms strong baselines, establishing SOTA performance and highlighting the benefit of modeling multiple adaptation directions via expert-level diversity.
Authors: Jiaxin Qi, Yan Cui, Jianqiang Huang, Gaogang Xie
Abstract: Classes, as fundamental elements of Computer Vision, have been extensively studied within incremental learning frameworks. In contrast, tokens, which play essential roles in many research fields, exhibit similar characteristics of growth, yet investigations into their incremental learning remain significantly scarce. This research gap primarily stems from the holistic nature of tokens in language, which imposes significant challenges on the design of incremental learning frameworks for them. To overcome this obstacle, in this work, we turn to a type of token, gene, for a large-scale biological dataset--single-cell transcriptomics--to formulate a pipeline for gene incremental learning and establish corresponding evaluations. We found that the forgetting problem also exists in gene incremental learning, thus we adapted existing class incremental learning methods to mitigate the forgetting of genes. Through extensive experiments, we demonstrated the soundness of our framework design and evaluations, as well as the effectiveness of our method adaptations. Finally, we provide a complete benchmark for gene incremental learning in single-cell transcriptomics.
Authors: Arun Thangamani, Md Asghar Ahmad Shahid, Adam Siemieniuk, Rolf Morel, Renato Golin, Alexander Heinecke
Abstract: The rapidly evolving landscape of AI and machine learning workloads has widened the gap between high-level domain operations and efficient hardware utilization. Achieving near-peak performance still demands deep hardware expertise-experts either handcraft target-specific kernels (e.g., DeepSeek) or rely on specialized libraries (e.g., CUTLASS)-both of which add complexity and limit scalability for most ML practitioners. This paper introduces a compilation scheme that automatically generates scalable, high-performance microkernels by leveraging the MLIR dialects to bridge domain-level operations and processor capabilities. Our approach removes dependence on low-level libraries by enabling the compiler to auto-generate near-optimal code directly. At its core is a mechanism for composing nanokernels from low-level IR constructs with near-optimal register utilization, forming efficient microkernels tailored to each target. We implement this technique in an MLIR-based compiler supporting both vector and tile based CPU instructions. Experiments show that the generated nanokernels are of production-quality, and competitive with state-of-the-art microkernel libraries.
Authors: Shengjie Sun, Jiafei Lyu, Runze Liu, Mengbei Yan, Bo Liu, Deheng Ye, Xiu Li
Abstract: Offline imitation learning (offline IL) enables training effective policies without requiring explicit reward annotations. Recent approaches attempt to estimate rewards for unlabeled datasets using a small set of expert demonstrations. However, these methods often assume that the similarity between a trajectory and an expert demonstration is positively correlated with the reward, which oversimplifies the underlying reward structure. We propose PROF, a novel framework that leverages large language models (LLMs) to generate and improve executable reward function codes from natural language descriptions and a single expert trajectory. We propose Reward Preference Ranking (RPR), a novel reward function quality assessment and ranking strategy without requiring environment interactions or RL training. RPR calculates the dominance scores of the reward functions, where higher scores indicate better alignment with expert preferences. By alternating between RPR and text-based gradient optimization, PROF fully automates the selection and refinement of optimal reward functions for downstream policy learning. Empirical results on D4RL demonstrate that PROF surpasses or matches recent strong baselines across numerous datasets and domains, highlighting the effectiveness of our approach.
Authors: Kaizheng Wang, Fabio Cuzzolin, David Moens, Hans Hallez
Abstract: Deep ensembles (DE) have emerged as a powerful approach for quantifying predictive uncertainty and distinguishing its aleatoric and epistemic components, thereby enhancing model robustness and reliability. However, their high computational and memory costs during inference pose significant challenges for wide practical deployment. To overcome this issue, we propose credal ensemble distillation (CED), a novel framework that compresses a DE into a single model, CREDIT, for classification tasks. Instead of a single softmax probability distribution, CREDIT predicts class-wise probability intervals that define a credal set, a convex set of probability distributions, for uncertainty quantification. Empirical results on out-of-distribution detection benchmarks demonstrate that CED achieves superior or comparable uncertainty estimation compared to several existing baselines, while substantially reducing inference overhead compared to DE.
Authors: Sibgat Ul Islam, Jawad Ibn Ahad, Fuad Rahman, Mohammad Ruhul Amin, Nabeel Mohammed, Shafin Rahman
Abstract: Knowledge Distillation (KD) trains a smaller student model using a large, pre-trained teacher model, with temperature as a key hyperparameter controlling the softness of output probabilities. Traditional methods use a fixed temperature throughout training, which is suboptimal. Moreover, architectural differences between teacher and student often result in mismatched logit magnitudes. We demonstrate that students benefit from softer probabilities early in training but require sharper probabilities in later stages. We introduce Dynamic Temperature Scheduler (DTS), which adjusts temperature dynamically based on the cross-entropy loss gap between teacher and student. To our knowledge, this is the first temperature scheduling method that adapts based on the divergence between teacher and student distributions. Our method integrates seamlessly with existing KD frameworks. We validate DTS across multiple KD strategies on vision (CIFAR-100, Tiny-ImageNet) and NLP tasks (GLUE, Dolly, SelfIns, UnNI, S-NI), consistently outperforming static-temperature baselines. Code is available at https://github.com/Sibgat-Ul/DTS.
Authors: Joey Velez-Ginorio, Nada Amin, Konrad Kording, Steve Zdancewic
Abstract: We don't program neural networks directly. Instead, we rely on an indirect style where learning algorithms, like gradient descent, determine a neural network's function by learning from data. This indirect style is often a virtue; it empowers us to solve problems that were previously impossible. But it lacks discrete structure. We can't compile most algorithms into a neural network -- even if these algorithms could help the network learn. This limitation occurs because discrete algorithms are not obviously differentiable, making them incompatible with the gradient-based learning algorithms that determine a neural network's function. To address this, we introduce $\textsf{Cajal}$: a typed, higher-order and linear programming language intended to be a minimal vehicle for exploring a direct style of programming neural networks. We prove $\textsf{Cajal}$ programs compile to linear neurons, allowing discrete algorithms to be expressed in a differentiable form compatible with gradient-based learning. With our implementation of $\textsf{Cajal}$, we conduct several experiments where we link these linear neurons against other neural networks to determine part of their function prior to learning. Linking with these neurons allows networks to learn faster, with greater data-efficiency, and in a way that's easier to debug. A key lesson is that linear programming languages provide a path towards directly programming neural networks, enabling a rich interplay between learning and the discrete structures of ordinary programming.
Authors: Nihal Mehta
Abstract: This paper presents a mathematical interpretation of self-attention by connecting it to distributional semantics principles. We show that self-attention emerges from projecting corpus-level co-occurrence statistics into sequence context. Starting from the co-occurrence matrix underlying GloVe embeddings, we demonstrate how the projection naturally captures contextual influence, with the query-key-value mechanism arising as the natural asymmetric extension for modeling directional relationships. Positional encodings and multi-head attention then follow as structured refinements of this same projection principle. Our analysis demonstrates that the Transformer architecture's particular algebraic form follows from these projection principles rather than being an arbitrary design choice.
Authors: Huijie Guo, Jingyao Wang, Peizheng Guo, Xingchen Shen, Changwen Zheng, Wenwen Qiang
Abstract: In this paper, we explore the transferability of SSL by addressing two central questions: (i) what is the representation transferability of SSL, and (ii) how can we effectively model this transferability? Transferability is defined as the ability of a representation learned from one task to support the objective of another. Inspired by the meta-learning paradigm, we construct multiple SSL tasks within each training batch to support explicitly modeling transferability. Based on empirical evidence and causal analysis, we find that although introducing task-level information improves transferability, it is still hindered by task conflict. To address this issue, we propose a Task Conflict Calibration (TC$^2$) method to alleviate the impact of task conflict. Specifically, it first splits batches to create multiple SSL tasks, infusing task-level information. Next, it uses a factor extraction network to produce causal generative factors for all tasks and a weight extraction network to assign dedicated weights to each sample, employing data reconstruction, orthogonality, and sparsity to ensure effectiveness. Finally, TC$^2$ calibrates sample representations during SSL training and integrates into the pipeline via a two-stage bi-level optimization framework to boost the transferability of learned representations. Experimental results on multiple downstream tasks demonstrate that our method consistently improves the transferability of SSL models.
Authors: Samuel Nathanson, Rebecca Williams, Cynthia Matuszek
Abstract: Large language models (LLMs) increasingly operate in multi-agent and safety-critical settings, raising open questions about how their vulnerabilities scale when models interact adversarially. This study examines whether larger models can systematically jailbreak smaller ones - eliciting harmful or restricted behavior despite alignment safeguards. Using standardized adversarial tasks from JailbreakBench, we simulate over 6,000 multi-turn attacker-target exchanges across major LLM families and scales (0.6B-120B parameters), measuring both harm score and refusal behavior as indicators of adversarial potency and alignment integrity. Each interaction is evaluated through aggregated harm and refusal scores assigned by three independent LLM judges, providing a consistent, model-based measure of adversarial outcomes. Aggregating results across prompts, we find a strong and statistically significant correlation between mean harm and the logarithm of the attacker-to-target size ratio (Pearson r = 0.51, p < 0.001; Spearman rho = 0.52, p < 0.001), indicating that relative model size correlates with the likelihood and severity of harmful completions. Mean harm score variance is higher across attackers (0.18) than across targets (0.10), suggesting that attacker-side behavioral diversity contributes more to adversarial outcomes than target susceptibility. Attacker refusal frequency is strongly and negatively correlated with harm (rho = -0.93, p < 0.001), showing that attacker-side alignment mitigates harmful responses. These findings reveal that size asymmetry influences robustness and provide exploratory evidence for adversarial scaling patterns, motivating more controlled investigations into inter-model alignment and safety.
Authors: Emanuel Covaci, Fabian Galis, Radu Balan, Daniela Zaharie, Darian Onchis
Abstract: Understanding the decision of large deep learning models is a critical challenge for building transparent and trustworthy systems. Although the current post hoc explanation methods offer valuable insights into feature importance, they are inherently disconnected from the model training process, limiting their faithfulness and utility. In this work, we introduce a novel differentiable approach to global explainability by design, integrating feature importance estimation directly into model training. Central to our method is the ScoresActivation function, a feature-ranking mechanism embedded within the learning pipeline. This integration enables models to prioritize features according to their contribution to predictive performance in a differentiable and end-to-end trainable manner. Evaluations across benchmark datasets show that our approach yields globally faithful, stable feature rankings aligned with SHAP values and ground-truth feature importance, while maintaining high predictive performance. Moreover, feature scoring is 150 times faster than the classical SHAP method, requiring only 2 seconds during training compared to SHAP's 300 seconds for feature ranking in the same configuration. Our method also improves classification accuracy by 11.24% with 10 features (5 relevant) and 29.33% with 16 features (5 relevant, 11 irrelevant), demonstrating robustness to irrelevant inputs. This work bridges the gap between model accuracy and interpretability, offering a scalable framework for inherently explainable machine learning.
Authors: Zelei Shao, Vikranth Srivatsa, Sanjana Srivastava, Qingyang Wu, Alpay Ariyak, Xiaoxia Wu, Ameen Patel, Jue Wang, Percy Liang, Tri Dao, Ce Zhang, Yiying Zhang, Ben Athiwaratkun, Chenfeng Xu, Junxiong Wang
Abstract: Reinforcement learning(RL) post-training has become essential for aligning large language models (LLMs), yet its efficiency is increasingly constrained by the rollout phase, where long trajectories are generated token by token. We identify a major bottleneck:the long-tail distribution of rollout lengths, where a small fraction of long generations dominates wall clock time and a complementary opportunity; the availability of historical rollouts that reveal stable prompt level patterns across training epochs. Motivated by these observations, we propose DAS, a Distribution Aware Speculative decoding framework that accelerates RL rollouts without altering model outputs. DAS integrates two key ideas: an adaptive, nonparametric drafter built from recent rollouts using an incrementally maintained suffix tree, and a length aware speculation policy that allocates more aggressive draft budgets to long trajectories that dominate makespan. This design exploits rollout history to sustain acceptance while balancing base and token level costs during decoding. Experiments on math and code reasoning tasks show that DAS reduces rollout time up to 50% while preserving identical training curves, demonstrating that distribution-aware speculative decoding can significantly accelerate RL post training without compromising learning quality.
Authors: Saleh Momeni, Changnan Xiao, Bing Liu
Abstract: This paper studies the problem of class-incremental learning (CIL), a core setting within continual learning where a model learns a sequence of tasks, each containing a distinct set of classes. Traditional CIL methods, which do not leverage pre-trained models (PTMs), suffer from catastrophic forgetting (CF) due to the need to incrementally learn both feature representations and the classifier. The integration of PTMs into CIL has recently led to efficient approaches that treat the PTM as a fixed feature extractor combined with analytic classifiers, achieving state-of-the-art performance. However, they still face a major limitation: the inability to continually adapt feature representations to best suit the CIL tasks, leading to suboptimal performance. To address this, we propose AnaCP (Analytic Contrastive Projection), a novel method that preserves the efficiency of analytic classifiers while enabling incremental feature adaptation without gradient-based training, thereby eliminating the CF caused by gradient updates. Our experiments show that AnaCP not only outperforms existing baselines but also achieves the accuracy level of joint training, which is regarded as the upper bound of CIL.
Authors: Nicolas M. Cuadrado A., Mohannad Takrouri, Ji\v{r}\'i N\v{e}me\v{c}ek, Martin Tak\'a\v{c}, Jakub Mare\v{c}ek
Abstract: Investment planning in power utilities, such as generation and transmission expansion, requires decade-long forecasts under profound uncertainty. Forecasting of energy mix and energy use decades ahead is nontrivial. Classical approaches focus on generating a finite number of scenarios (modeled as a mixture of Diracs in statistical theory terms), which limits insight into scenario-specific volatility and hinders robust decision-making. We propose an alternative using tractable probabilistic models (TPMs), particularly sum-product networks (SPNs). These models enable exact, scalable inference of key quantities such as scenario likelihoods, marginals, and conditional probabilities, supporting robust scenario expansion and risk assessment. This framework enables direct embedding of chance-constrained optimization into investment planning, enforcing safety or reliability with prescribed confidence levels. TPMs allow both scenario analysis and volatility quantification by compactly representing high-dimensional uncertainties. We demonstrate the approach's effectiveness through a representative power system planning case study, illustrating computational and reliability advantages over traditional scenario-based models.
Authors: Kai Chen, Chen Gong, Tianhao Wang
Abstract: In differentially private (DP) tabular data synthesis, the consensus is that statistical models are better than neural network (NN)-based methods. However, we argue that this conclusion is incomplete and overlooks the challenge of densely correlated datasets, where intricate dependencies can overwhelm statistical models. In such complex scenarios, neural networks are more suitable due to their capacity to fit complex distributions by learning directly from samples. Despite this potential, existing NN-based algorithms still suffer from significant limitations. We therefore propose MargNet, incorporating successful algorithmic designs of statistical models into neural networks. MargNet applies an adaptive marginal selection strategy and trains the neural networks to generate data that conforms to the selected marginals. On sparsely correlated datasets, our approach achieves utility close to the best statistical method while offering an average 7$\times$ speedup over it. More importantly, on densely correlated datasets, MargNet establishes a new state-of-the-art, reducing fidelity error by up to 26\% compared to the previous best. We release our code on GitHub.\footnote{https://github.com/KaiChen9909/margnet}
Authors: Federico Battini
Abstract: Accurate renewable energy forecasting is essential to reduce dependence on fossil fuels and enabling grid decarbonization. However, current approaches fail to effectively integrate the rich spatial context of weather patterns with their temporal evolution. This work introduces a novel approach that treats weather maps as tokens in transformer sequences to predict renewable energy. Hourly weather maps are encoded as spatial tokens using a lightweight convolutional neural network, and then processed by a transformer to capture temporal dynamics across a 45-hour forecast horizon. Despite disadvantages in input initialization, evaluation against ENTSO-E operational forecasts shows a reduction in RMSE of about 60\% and 20\% for wind and solar respectively. A live dashboard showing daily forecasts is available at: https://www.sardiniaforecast.ifabfoundation.it.
Authors: Cristina L\'opez Amado, Tassilo Schwarz, Yu Tian, Renaud Lambiotte
Abstract: Graph Neural Networks (GNNs) have achieved remarkable success across diverse applications, yet they remain limited by oversmoothing and poor performance on heterophilic graphs. To address these challenges, we introduce a novel framework that equips graphs with a complex-weighted structure, assigning each edge a complex number to drive a diffusion process that extends random walks into the complex domain. We prove that this diffusion is highly expressive: with appropriately chosen complex weights, any node-classification task can be solved in the steady state of a complex random walk. Building on this insight, we propose the Complex-Weighted Convolutional Network (CWCN), which learns suitable complex-weighted structures directly from data while enriching diffusion with learnable matrices and nonlinear activations. CWCN is simple to implement, requires no additional hyperparameters beyond those of standard GNNs, and achieves competitive performance on benchmark datasets. Our results demonstrate that complex-weighted diffusion provides a principled and general mechanism for enhancing GNN expressiveness, opening new avenues for models that are both theoretically grounded and practically effective.
Authors: Micha{\l} Iwaniuk, Mateusz Jarosz, Bart{\l}omiej Borycki, Bartosz Jezierski, Jan Cwalina, Stanis{\l}aw Ka\'zmierczak, Jacek Ma\'ndziuk
Abstract: Random Forests (RFs) typically train each tree on a bootstrap sample of the same size as the training set, i.e., bootstrap rate (BR) equals 1.0. We systematically examine how varying BR from 0.2 to 5.0 affects RF performance across 39 heterogeneous regression datasets and 16 RF configurations, evaluating with repeated two-fold cross-validation and mean squared error. Our results demonstrate that tuning the BR can yield significant improvements over the default: the best setup relied on BR \leq 1.0 for 24 datasets, BR > 1.0 for 15, and BR = 1.0 was optimal in 4 cases only. We establish a link between dataset characteristics and the preferred BR: datasets with strong global feature-target relationships favor higher BRs, while those with higher local target variance benefit from lower BRs. To further investigate this relationship, we conducted experiments on synthetic datasets with controlled noise levels. These experiments reproduce the observed bias-variance trade-off: in low-noise scenarios, higher BRs effectively reduce model bias, whereas in high-noise settings, lower BRs help reduce model variance. Overall, BR is an influential hyperparameter that should be tuned to optimize RF regression models.
Authors: Mingtao Xia, Qijing Shen
Abstract: In this paper, we analyze the scalability of a recent Wasserstein-distance approach for training stochastic neural networks (SNNs) to reconstruct multidimensional random field models. We prove a generalization error bound for reconstructing multidimensional random field models on training stochastic neural networks with a limited number of training data. Our results indicate that when noise is heterogeneous across dimensions, the convergence rate of the generalization error may not depend explicitly on the model's dimensionality, partially alleviating the "curse of dimensionality" for learning multidimensional random field models from a finite number of data points. Additionally, we improve the previous Wasserstein-distance SNN training approach and showcase the robustness of the SNN. Through numerical experiments on different multidimensional uncertainty quantification tasks, we show that our Wasserstein-distance approach can successfully train stochastic neural networks to learn multidimensional uncertainty models.
Authors: Ashwin Saraswatula, David Klindt
Abstract: Sparse autoencoders (SAEs) have emerged as a promising approach for learning interpretable features from neural network activations. However, the optimization landscape for SAE training can be challenging due to correlations in the input data. We demonstrate that applying PCA Whitening to input activations -- a standard preprocessing technique in classical sparse coding -- improves SAE performance across multiple metrics. Through theoretical analysis and simulation, we show that whitening transforms the optimization landscape, making it more convex and easier to navigate. We evaluate both ReLU and Top-K SAEs across diverse model architectures, widths, and sparsity regimes. Empirical evaluation on SAEBench, a comprehensive benchmark for sparse autoencoders, reveals that whitening consistently improves interpretability metrics, including sparse probing accuracy and feature disentanglement, despite minor drops in reconstruction quality. Our results challenge the assumption that interpretability aligns with an optimal sparsity--fidelity trade-off and suggest that whitening should be considered as a default preprocessing step for SAE training, particularly when interpretability is prioritized over perfect reconstruction.
Authors: Hilaf Hasson, Ruocheng Guo
Abstract: We present a practical framework for detecting errors in LLM-generated SQL by estimating uncertainty at the level of individual nodes in the query's abstract syntax tree (AST). Our approach proceeds in two stages. First, we introduce a semantically aware labeling algorithm that, given a generated SQL and a gold reference, assigns node-level correctness without over-penalizing structural containers or alias variation. Second, we represent each node with a rich set of schema-aware and lexical features - capturing identifier validity, alias resolution, type compatibility, ambiguity in scope, and typo signals - and train a supervised classifier to predict per-node error probabilities. We interpret these probabilities as calibrated uncertainty, enabling fine-grained diagnostics that pinpoint exactly where a query is likely to be wrong. Across multiple databases and datasets, our method substantially outperforms token log-probabilities: average AUC improves by +27.44% while maintaining robustness under cross-database evaluation. Beyond serving as an accuracy signal, node-level uncertainty supports targeted repair, human-in-the-loop review, and downstream selective execution. Together, these results establish node-centric, semantically grounded uncertainty estimation as a strong and interpretable alternative to aggregate sequence level confidence measures.
Authors: Michael Menart, Aleksandar Nikolov
Abstract: We study the running time, in terms of first order oracle queries, of differentially private empirical/population risk minimization of Lipschitz convex losses. We first consider the setting where the loss is non-smooth and the optimizer interacts with a private proxy oracle, which sends only private messages about a minibatch of gradients. In this setting, we show that expected running time $\Omega(\min\{\frac{\sqrt{d}}{\alpha^2}, \frac{d}{\log(1/\alpha)}\})$ is necessary to achieve $\alpha$ excess risk on problems of dimension $d$ when $d \geq 1/\alpha^2$. Upper bounds via DP-SGD show these results are tight when $d>\tilde{\Omega}(1/\alpha^4)$. We further show our lower bound can be strengthened to $\Omega(\min\{\frac{d}{\bar{m}\alpha^2}, \frac{d}{\log(1/\alpha)} \})$ for algorithms which use minibatches of size at most $\bar{m} < \sqrt{d}$. We next consider smooth losses, where we relax the private oracle assumption and give lower bounds under only the condition that the optimizer is private. Here, we lower bound the expected number of first order oracle calls by $\tilde{\Omega}\big(\frac{\sqrt{d}}{\alpha} + \min\{\frac{1}{\alpha^2}, n\}\big)$, where $n$ is the size of the dataset. Modifications to existing algorithms show this bound is nearly tight. Compared to non-private lower bounds, our results show that differentially private optimizers pay a dimension dependent runtime penalty. Finally, as a natural extension of our proof technique, we show lower bounds in the non-smooth setting for optimizers interacting with information limited oracles. Specifically, if the proxy oracle transmits at most $\Gamma$-bits of information about the gradients in the minibatch, then $\Omega\big(\min\{\frac{d}{\alpha^2\Gamma}, \frac{d}{\log(1/\alpha)}\}\big)$ oracle calls are needed. This result shows fundamental limitations of gradient quantization techniques in optimization.
Authors: William Zhao, Guy Van den Broeck, Benjie Wang
Abstract: Bayesian networks (BNs) are a widely used class of probabilistic graphical models employed in numerous application domains. However, inferring the network's graphical structure from data remains challenging. Bayesian structure learners approach this problem by inferring a posterior distribution over the possible directed acyclic graphs underlying the BN. The inference process often requires marginalizing over probability distributions, which is typically done using dynamic programming methods that restrict the set of possible parents for each node. Instead, we present a novel method that utilizes tractable probabilistic circuits to circumvent this restriction. This method utilizes a new learning routine that trains these circuits on both the original distribution and marginal queries. The architecture of probabilistic circuits then inherently allows for fast and exact marginalization on the learned distribution. We then show empirically that utilizing our method to answer marginals allows Bayesian structure learners to improve their performance compared to current methods.
Authors: Quoc Viet Vo, Tashreque M. Haq, Paul Montague, Tamas Abraham, Ehsan Abbasnejad, Damith C. Ranasinghe
Abstract: Certified defenses promise provable robustness guarantees. We study the malicious exploitation of probabilistic certification frameworks to better understand the limits of guarantee provisions. Now, the objective is to not only mislead a classifier, but also manipulate the certification process to generate a robustness guarantee for an adversarial input certificate spoofing. A recent study in ICLR demonstrated that crafting large perturbations can shift inputs far into regions capable of generating a certificate for an incorrect class. Our study investigates if perturbations needed to cause a misclassification and yet coax a certified model into issuing a deceptive, large robustness radius for a target class can still be made small and imperceptible. We explore the idea of region-focused adversarial examples to craft imperceptible perturbations, spoof certificates and achieve certification radii larger than the source class ghost certificates. Extensive evaluations with the ImageNet demonstrate the ability to effectively bypass state-of-the-art certified defenses such as Densepure. Our work underscores the need to better understand the limits of robustness certification methods.
Authors: Erum Mushtaq, Anil Ramakrishna, Satyapriya Krishna, Sattvik Sahai, Prasoon Goyal, Kai-Wei Chang, Tao Zhang, Rahul Gupta
Abstract: Recent work has shown that fine-tuning on insecure code data can trigger an emergent misalignment (EMA) phenomenon, where models generate malicious responses even to prompts unrelated to the original insecure code-writing task. Such cross-domain generalization of harmful behavior underscores the need for a deeper understanding of the algorithms, tasks, and datasets that induce emergent misalignment. In this work, we extend this study by demonstrating that emergent misalignment can also arise from narrow refusal unlearning in specific domains. We perform refusal unlearning on Cybersecurity and Safety concept, and evaluate EMA by monitoring refusal scores across seven responsible AI (RAI) domains, Cybersecurity, Safety, Toxicity, Bias, Sensitive Content, Medical/Legal, and Privacy. Our work shows that narrow domain unlearning can yield compliance responses for the targeted concept, however, it may also propagate EMA to unrelated domains. Among the two intervened concepts, Cybersecurity and Safety, we find that the safety concept can have larger EMA impact, i.e, causing lower refusal scores, across other unrelated domains such as bias. We observe this effect consistently across two model families, Mistral-7b-0.3v, and Qwen-7b-2.5. Further, we show that refusal unlearning augmented with cross-entropy loss function on a small set of retain data from the affected domains can largely, if not fully, restore alignment across the impacted domains while having lower refusal rate on the concept we perform unlearning on. To investigate the underlying causes of EMA, we analyze concept entanglements at the representation level via concept vectors. Our analysis reveals that concepts with higher representation similarity in earlier layers are more susceptible to EMA after intervention when the refusal stream is altered through targeted refusal unlearning.
Authors: Semen Leontev
Abstract: Small and medium-sized enterprises (SMEs) represent 99.9% of U.S. businesses yet remain systematically excluded from AI due to a mismatch between their operational scale and modern machine learning's data requirements. This paper introduces SmallML, a Bayesian transfer learning framework achieving enterprise-level prediction accuracy with datasets as small as 50-200 observations. We develop a three-layer architecture integrating transfer learning, hierarchical Bayesian modeling, and conformal prediction. Layer 1 extracts informative priors from 22,673 public records using a SHAP-based procedure transferring knowledge from gradient boosting to logistic regression. Layer 2 implements hierarchical pooling across J=5-50 SMEs with adaptive shrinkage, balancing population patterns with entity-specific characteristics. Layer 3 provides conformal sets with finite-sample coverage guarantees P(y in C(x)) >= 1-alpha for distribution-free uncertainty quantification. Validation on customer churn data demonstrates 96.7% +/- 4.2% AUC with 100 observations per business -- a +24.2 point improvement over independent logistic regression (72.5% +/- 8.1%), with p < 0.000001. Conformal prediction achieves 92% empirical coverage at 90% target. Training completes in 33 minutes on standard CPU hardware. By enabling enterprise-grade predictions for 33 million U.S. SMEs previously excluded from machine learning, SmallML addresses a critical gap in AI democratization. Keywords: Bayesian transfer learning, hierarchical models, conformal prediction, small-data analytics, SME machine learning
Authors: Marios Papamichals, Regina Ruane
Abstract: Generative models on curved spaces rely on charts to map Euclidean spaces to manifolds. Exponential maps preserve geodesics but have stiff, radius-dependent Jacobians, while volume-preserving charts maintain densities but distort geodesic distances. Both approaches entangle curvature with model parameters, inflating gradient variance. In high-dimensional latent normalizing flows, the wrapped exponential prior can stretch radii far beyond the curvature scale, leading to poor test likelihoods and stiff solvers. We introduce Radial Compensation (RC), an information-geometric method that selects the base density in the tangent space so that the likelihood depends only on geodesic distance from a pole, decoupling parameter semantics from curvature. RC lets radial parameters retain their usual meaning in geodesic units, while the chart can be tuned as a numerical preconditioner. We extend RC to manifolds with known geodesic polar volume and show that RC is the only construction for geodesic-radial likelihoods with curvature-invariant Fisher information. We derive the Balanced-Exponential (bExp) chart family, balancing volume distortion and geodesic error. Under RC, all bExp settings preserve the same manifold density and Fisher information, with smaller dial values reducing gradient variance and flow cost. Empirically, RC yields stable generative models across densities, VAEs, flows on images and graphs, and protein models. RC improves likelihoods, restores clean geodesic radii, and prevents radius blow-ups in high-dimensional flows, making RC-bExp a robust default for likelihood-trained generative models on manifolds.
Authors: Xianghe Liu, Jiajia Liu, Chuxian Xu, Minghan Wang, Hongbo Peng, Tao Sun, Jiaqi Xu
Abstract: In precision sports such as archery, athletes' performance depends on both biomechanical stability and psychological resilience. Traditional motion analysis systems are often expensive and intrusive, limiting their use in natural training environments. To address this limitation, we propose a machine learning-based multimodal framework that integrates wearable sensor data for simultaneous action recognition and stress estimation. Using a self-developed wrist-worn device equipped with an accelerometer and photoplethysmography (PPG) sensor, we collected synchronized motion and physiological data during real archery sessions. For motion recognition, we introduce a novel feature--Smoothed Differential Acceleration (SmoothDiff)--and employ a Long Short-Term Memory (LSTM) model to identify motion phases, achieving 96.8% accuracy and 95.9% F1-score. For stress estimation, we extract heart rate variability (HRV) features from PPG signals and apply a Multi-Layer Perceptron (MLP) classifier, achieving 80% accuracy in distinguishing high- and low-stress levels. The proposed framework demonstrates that integrating motion and physiological sensing can provide meaningful insights into athletes' technical and mental states. This approach offers a foundation for developing intelligent, real-time feedback systems for training optimization in archery and other precision sports.
Authors: Kelin Ren, Chan-Yang Ju, Dong-Ho Lee
Abstract: Medication recommendation systems play a crucial role in assisting clinicians with personalized treatment decisions. While existing approaches have made significant progress in learning medication representations, they suffer from two fundamental limitations: (i) treating medical entities as independent features without modeling their synergistic effects on medication selection; (ii) employing static causal relationships that fail to adapt to patient-specific contexts and health states. To address these challenges, we propose CafeMed, a framework that integrates dynamic causal reasoning with cross-modal attention for safe and accurate medication recommendation. CafeMed introduces two key components: the Causal Weight Generator (CWG) that transforms static causal effects into dynamic modulation weights based on individual patient states, and the Channel Harmonized Attention Refinement Module (CHARM) that captures complex interdependencies between diagnoses and procedures. This design enables CafeMed to model how different medical conditions jointly influence treatment decisions while maintaining medication safety constraints. Extensive experiments on MIMIC-III and MIMIC-IV datasets demonstrate that CafeMed significantly outperforms state-of-the-art baselines, achieving superior accuracy in medication prediction while maintaining the lower drug--drug interaction rates. Our results indicate that incorporating dynamic causal relationships and cross-modal synergies leads to more clinically-aligned and personalized medication recommendations. Our code is released publicly at https://github.com/rkl71/CafeMed.
Authors: Nakkyu Yang, Yechan Lee, SooJean Han
Abstract: Classifier-Free Guidance (CFG) has become a mainstream approach for simultaneously improving prompt fidelity and generation quality in conditional generative models. During training, CFG stochastically alternates between conditional and null prompts to enable both conditional and unconditional generation. However, during sampling, CFG outputs both null and conditional prompts simultaneously, leading to inconsistent noise estimates between the training and sampling processes. To reduce this error, we propose CFG-EC, a versatile correction scheme augmentable to any CFG-based method by refining the unconditional noise predictions. CFG-EC actively realigns the unconditional noise error component to be orthogonal to the conditional error component. This corrective maneuver prevents interference between the two guidance components, thereby constraining the sampling error's upper bound and establishing more reliable guidance trajectories for high-fidelity image generation. Our numerical experiments show that CFG-EC handles the unconditional component more effectively than CFG and CFG++, delivering a marked performance increase in the low guidance sampling regime and consistently higher prompt alignment across the board.
Authors: Qiqi Xiao, Ziqi Ye, Yinghui He, Jianwei Liu, Guanding Yu
Abstract: To promote the practicality of deep learning-based localization, existing studies aim to address the issue of scenario dependence through meta-learning. However, these studies primarily focus on variations in environmental layouts while overlooking the impact of changes in device configurations, such as bandwidth, the number of access points (APs), and the number of antennas used. Unlike environmental changes, variations in device configurations affect the dimensionality of channel state information (CSI), thereby compromising neural network usability. To address this issue, we propose Meta-SimGNN, a novel WiFi localization system that integrates graph neural networks with meta-learning to improve localization generalization and robustness. First, we introduce a fine-grained CSI graph construction scheme, where each AP is treated as a graph node, allowing for adaptability to changes in the number of APs. To structure the features of each node, we propose an amplitude-phase fusion method and a feature extraction method. The former utilizes both amplitude and phase to construct CSI images, enhancing data reliability, while the latter extracts dimension-consistent features to address variations in bandwidth and the number of antennas. Second, a similarity-guided meta-learning strategy is developed to enhance adaptability in diverse scenarios. The initial model parameters for the fine-tuning stage are determined by comparing the similarity between the new scenario and historical scenarios, facilitating rapid adaptation of the model to the new localization scenario. Extensive experimental results over commodity WiFi devices in different scenarios show that Meta-SimGNN outperforms the baseline methods in terms of localization generalization and accuracy.
Authors: Iden Kalemaj, Luca Melis, Maxime Boucher, Ilya Mironov, Saeed Mahloujifar
Abstract: Differential privacy (DP) auditing is essential for evaluating privacy guarantees in machine learning systems. Existing auditing methods, however, pose a significant challenge for large-scale systems since they require modifying the training dataset -- for instance, by injecting out-of-distribution canaries or removing samples from training. Such interventions on the training data pipeline are resource-intensive and involve considerable engineering overhead. We introduce a novel observational auditing framework that leverages the inherent randomness of data distributions, enabling privacy evaluation without altering the original dataset. Our approach extends privacy auditing beyond traditional membership inference to protected attributes, with labels as a special case, addressing a key gap in existing techniques. We provide theoretical foundations for our method and perform experiments on Criteo and CIFAR-10 datasets that demonstrate its effectiveness in auditing label privacy guarantees. This work opens new avenues for practical privacy auditing in large-scale production environments.
Authors: Wenfeng Wang, Jiacheng Liu, Xiaofeng Hou, Xinfeng Xia, Peng Tang, Mingxuan Zhang, Chao Li, Minyi Guo
Abstract: The immense memory requirements of state-of-the-art Mixture-of-Experts (MoE) models present a significant challenge for inference, often exceeding the capacity of a single accelerator. While offloading experts to host memory is a common solution, it introduces a severe I/O bottleneck over the PCIe bus, as the data-dependent nature of expert selection places these synchronous transfers directly on the critical path of execution, crippling performance. This paper argues that the I/O bottleneck can be overcome by trading a small amount of cheap, on-device computation to hide the immense cost of data movement. We present MoE-SpeQ, a new inference system built on a novel co-design of speculative execution and expert offloading. MoE-SpeQ employs a small, on-device draft model to predict the sequence of required experts for future tokens. This foresight enables a runtime orchestrator to prefetch these experts from host memory, effectively overlapping the expensive I/O with useful computation and hiding the latency from the critical path. To maximize performance, an adaptive governor, guided by an Amortization Roofline Model, dynamically tunes the speculation strategy to the underlying hardware. Our evaluation on memory-constrained devices shows that for the Phi-MoE model, MoE-SpeQ achieves at most 2.34x speedup over the state-of-the-art offloading framework. Our work establishes a new, principled approach for managing data-dependent memory access in resource-limited environments, making MoE inference more accessible on commodity hardware.
Authors: Agamdeep Singh, Ashish Tiwari, Hosein Hasanbeig, Priyanshu Gupta
Abstract: Many machine learning tasks involve inherent subjectivity, where annotators naturally provide varied labels. Standard practice collapses these label distributions into single labels, aggregating diverse human judgments into point estimates. We argue that this approach is epistemically misaligned for ambiguous data--the annotation distribution itself should be regarded as the ground truth. Training on collapsed single labels forces models to express false confidence on fundamentally ambiguous cases, creating a misalignment between model certainty and the diversity of human perception. We demonstrate empirically that soft-label training, which treats annotation distributions as ground truth, preserves epistemic uncertainty. Across both vision and NLP tasks, soft-label training achieves 32% lower KL divergence from human annotations and 61% stronger correlation between model and annotation entropy, while matching the accuracy of hard-label training. Our work repositions annotation distributions from noisy signals to be aggregated away, to faithful representations of epistemic uncertainty that models should learn to reproduce.
Authors: Jessy Xinyi Han, Devavrat Shah
Abstract: Estimating causal effects on time-to-event outcomes from observational data is particularly challenging due to censoring, limited sample sizes, and non-random treatment assignment. The need for answering such "when-if" questions--how the timing of an event would change under a specified intervention--commonly arises in real-world settings with heterogeneous treatment adoption and confounding. To address these challenges, we propose Synthetic Survival Control (SSC) to estimate counterfactual hazard trajectories in a panel data setting where multiple units experience potentially different treatments over multiple periods. In such a setting, SSC estimates the counterfactual hazard trajectory for a unit of interest as a weighted combination of the observed trajectories from other units. To provide formal justification, we introduce a panel framework with a low-rank structure for causal survival analysis. Indeed, such a structure naturally arises under classical parametric survival models. Within this framework, for the causal estimand of interest, we establish identification and finite sample guarantees for SSC. We validate our approach using a multi-country clinical dataset of cancer treatment outcomes, where the staggered introduction of new therapies creates a quasi-experimental setting. Empirically, we find that access to novel treatments is associated with improved survival, as reflected by lower post-intervention hazard trajectories relative to their synthetic counterparts. Given the broad relevance of survival analysis across medicine, economics, and public policy, our framework offers a general and interpretable tool for counterfactual survival inference using observational data.
Authors: Promise Ekpo, Saesha Agarwal, Felix Grimm, Lekan Molu, Angelique Taylor
Abstract: Enforcing a fair workload allocation among multiple agents tasked to achieve an objective in learning enabled demand side healthcare worker settings is crucial for consistent and reliable performance at runtime. Existing multi-agent reinforcement learning (MARL) approaches steer fairness by shaping reward through post hoc orchestrations, leaving no certifiable self-enforceable fairness that is immutable by individual agents at runtime. Contextualized within a setting where each agent shares resources with others, we address this shortcoming with a learning enabled optimization scheme among self-interested decision makers whose individual actions affect those of other agents. This extends the problem to a generalized Nash equilibrium (GNE) game-theoretic framework where we steer group policy to a safe and locally efficient equilibrium, so that no agent can improve its utility function by unilaterally changing its decisions. Fair-GNE models MARL as a constrained generalized Nash equilibrium-seeking (GNE) game, prescribing an ideal equitable collective equilibrium within the problem's natural fabric. Our hypothesis is rigorously evaluated in our custom-designed high-fidelity resuscitation simulator. Across all our numerical experiments, Fair-GNE achieves significant improvement in workload balance over fixed-penalty baselines (0.89 vs.\ 0.33 JFI, $p < 0.01$) while maintaining 86\% task success, demonstrating statistically significant fairness gains through adaptive constraint enforcement. Our results communicate our formulations, evaluation metrics, and equilibrium-seeking innovations in large multi-agent learning-based healthcare systems with clarity and principled fairness enforcement.
Authors: Fatima Kazi, Alex Young, Yash Inani, Setareh Rafatirad
Abstract: Large Language Models (LLMs) inherit explicit and implicit biases from their training datasets. Identifying and mitigating biases in LLMs is crucial to ensure fair outputs, as they can perpetuate harmful stereotypes and misinformation. This study highlights the need to address biases in LLMs amid growing generative AI. We studied bias-specific benchmarks such as StereoSet and CrowSPairs to evaluate the existence of various biases in multiple generative models such as BERT and GPT 3.5. We proposed an automated Bias-Identification Framework to recognize various social biases in LLMs such as gender, race, profession, and religion. We adopted a two-pronged approach to detect explicit and implicit biases in text data. Results indicated fine-tuned models struggle with gender biases but excelled at identifying and avoiding racial biases. Our findings illustrated that despite having some success, LLMs often over-relied on keywords. To illuminate the capability of the analyzed LLMs in detecting implicit biases, we employed Bag-of-Words analysis and unveiled indications of implicit stereotyping within the vocabulary. To bolster the model performance, we applied an enhancement strategy involving fine-tuning models using prompting techniques and data augmentation of the bias benchmarks. The fine-tuned models exhibited promising adaptability during cross-dataset testing and significantly enhanced performance on implicit bias benchmarks, with performance gains of up to 20%.
Authors: Junpeng Zhao, Lin Li, Kaixi Hu, Kaize Shi, Jingling Yuan
Abstract: Signed graphs model complex relationships through positive and negative edges, with widespread real-world applications. Given the sensitive nature of such data, selective removal mechanisms have become essential for privacy protection. While graph unlearning enables the removal of specific data influences from Graph Neural Networks (GNNs), existing methods are designed for conventional GNNs and overlook the unique heterogeneous properties of signed graphs. When applied to Signed Graph Neural Networks (SGNNs), these methods lose critical sign information, degrading both model utility and unlearning effectiveness. To address these challenges, we propose Certified Signed Graph Unlearning (CSGU), which provides provable privacy guarantees while preserving the sociological principles underlying SGNNs. CSGU employs a three-stage method: (1) efficiently identifying minimal influenced neighborhoods via triangular structures, (2) applying sociological theories to quantify node importance for optimal privacy budget allocation, and (3) performing importance-weighted parameter updates to achieve certified modifications with minimal utility degradation. Extensive experiments demonstrate that CSGU outperforms existing methods, achieving superior performance in both utility preservation and unlearning effectiveness on SGNNs.
Authors: Zheyu Lin, Jirui Yang, Hengqi Guo, Yubing Bao, Yao Guan
Abstract: Evaluating the safety robustness of LLMs is critical for their deployment. However, mainstream Red Teaming methods rely on online generation and black-box output analysis. These approaches are not only costly but also suffer from feedback latency, making them unsuitable for agile diagnostics after training a new model. To address this, we propose N-GLARE (A Non-Generative, Latent Representation-Efficient LLM Safety Evaluator). N-GLARE operates entirely on the model's latent representations, bypassing the need for full text generation. It characterizes hidden layer dynamics by analyzing the APT (Angular-Probabilistic Trajectory) of latent representations and introducing the JSS (Jensen-Shannon Separability) metric. Experiments on over 40 models and 20 red teaming strategies demonstrate that the JSS metric exhibits high consistency with the safety rankings derived from Red Teaming. N-GLARE reproduces the discriminative trends of large-scale red-teaming tests at less than 1\% of the token cost and the runtime cost, providing an efficient output-free evaluation proxy for real-time diagnostics.
Authors: Xinlei Xiong, Wenbo Hu, Shuxun Zhou, Kaifeng Bi, Lingxi Xie, Ying Liu, Richang Hong, Qi Tian
Abstract: Weather forecasting is fundamentally challenged by the chaotic nature of the atmosphere, necessitating probabilistic approaches to quantify uncertainty. While traditional ensemble prediction (EPS) addresses this through computationally intensive simulations, recent advances in Bayesian Deep Learning (BDL) offer a promising but often disconnected alternative. We bridge these paradigms through a unified hybrid Bayesian Deep Learning framework for ensemble weather forecasting that explicitly decomposes predictive uncertainty into epistemic and aleatoric components, learned via variational inference and a physics-informed stochastic perturbation scheme modeling flow-dependent atmospheric dynamics, respectively. We further establish a unified theoretical framework that rigorously connects BDL and EPS, providing formal theorems that decompose total predictive uncertainty into epistemic and aleatoric components under the hybrid BDL framework. We validate our framework on the large-scale 40-year ERA5 reanalysis dataset (1979-2019) with 0.25{\deg} spatial resolution. Experimental results show that our method not only improves forecast accuracy and yields better-calibrated uncertainty quantification but also achieves superior computational efficiency compared to state-of-the-art probabilistic diffusion models. We commit to making our code open-source upon acceptance of this paper.
Authors: Yaniv Oren, Joery A. de Vries, Pascal R. van der Vaart, Matthijs T. J. Spaan, Wendelin B\"ohmer
Abstract: Model-based reinforcement learning (RL) methods that leverage search are responsible for many milestone breakthroughs in RL. Sequential Monte Carlo (SMC) recently emerged as an alternative to the Monte Carlo Tree Search (MCTS) algorithm which drove these breakthroughs. SMC is easier to parallelize and more suitable to GPU acceleration. However, it also suffers from large variance and path degeneracy which prevent it from scaling well with increased search depth, i.e., increased sequential compute. To address these problems, we introduce Twice Sequential Monte Carlo Tree Search (TSMCTS). Across discrete and continuous environments TSMCTS outperforms the SMC baseline as well as a popular modern version of MCTS. Through variance reduction and mitigation of path degeneracy, TSMCTS scales favorably with sequential compute while retaining the properties that make SMC natural to parallelize.
Authors: Jim Broadbent, Felix Cohen, Frederik Hvilsh{\o}j, Eric Landau, Eren Sasoglu
Abstract: We simplify space binding by focusing on two core components, a single encoder per modality and high-quality data; enabling training state-of-the-art models on a single GPU in a few hours as opposed to multiple days. We present EBind, an Easy, data-centric, and parameter-efficient method to Bind the embedding spaces of multiple contrastive models. We demonstrate that a simple 1.8B-parameter image-text-video-audio-3D model can outperform models 4 to 17x the size. The key to achieving this is a carefully curated dataset of three complementary data sources: i) 6.7M fully-automated multimodal quintuples sourced via SOTA retrieval models, ii) 1M diverse, semi-automated triples annotated by humans as negative, partial, or positive matches, and iii) 3.4M pre-existing captioned data items. We use 13 different evaluations to demonstrate the value of each data source. Due to limitations with existing benchmarks, we further introduce the first high-quality, consensus-annotated zero-shot classification benchmark between audio and PCs. In contrast to related work, we will open-source our code, model weights, and datasets.
Authors: Yosuke Nishimoto, Takashi Matsubara
Abstract: World models have been developed to support sample-efficient deep reinforcement learning agents. However, it remains challenging for world models to accurately replicate environments that are high-dimensional, non-stationary, and composed of multiple objects with rich interactions since most world models learn holistic representations of all environmental components. By contrast, humans perceive the environment by decomposing it into discrete objects, facilitating efficient decision-making. Motivated by this insight, we propose \emph{Slot Transformer Imagination with CAusality-aware reinforcement learning} (STICA), a unified framework in which object-centric Transformers serve as the world model and causality-aware policy and value networks. STICA represents each observation as a set of object-centric tokens, together with tokens for the agent action and the resulting reward, enabling the world model to predict token-level dynamics and interactions. The policy and value networks then estimate token-level cause--effect relations and use them in the attention layers, yielding causality-guided decision-making. Experiments on object-rich benchmarks demonstrate that STICA consistently outperforms state-of-the-art agents in both sample efficiency and final performance.
Authors: Pietro Sittoni, Francesco Tudisco
Abstract: Recent work in deep learning has opened new possibilities for solving classical algorithmic tasks using end-to-end learned models. In this work, we investigate the fundamental task of solving linear systems, particularly those that are ill-conditioned. Existing numerical methods for ill-conditioned systems often require careful parameter tuning, preconditioning, or domain-specific expertise to ensure accuracy and stability. In this work, we propose Algebraformer, a Transformer-based architecture that learns to solve linear systems end-to-end, even in the presence of severe ill-conditioning. Our model leverages a novel encoding scheme that enables efficient representation of matrix and vector inputs, with a memory complexity of $O(n^2)$, supporting scalable inference. We demonstrate its effectiveness on application-driven linear problems, including interpolation tasks from spectral methods for boundary value problems and acceleration of the Newton method. Algebraformer achieves competitive accuracy with significantly lower computational overhead at test time, demonstrating that general-purpose neural architectures can effectively reduce complexity in traditional scientific computing pipelines.
Authors: Rui Zhang, Chao Li, Kezhong Liu, Chen Wang, Bolong Zheng, Hongbo Jiang
Abstract: Vessel trajectory prediction is fundamental to intelligent maritime systems. Within this domain, short-term prediction of rapid behavioral changes in complex maritime environments has established multimodal trajectory prediction (MTP) as a promising research area. However, existing vessel MTP methods suffer from limited scenario applicability and insufficient explainability. To address these challenges, we propose a unified MTP framework incorporating explainable navigation intentions, which we classify into sustained and transient categories. Our method constructs sustained intention trees from historical trajectories and models dynamic transient intentions using a Conditional Variational Autoencoder (CVAE), while using a non-local attention mechanism to maintain global scenario consistency. Experiments on real Automatic Identification System (AIS) datasets demonstrates our method's broad applicability across diverse scenarios, achieving significant improvements in both ADE and FDE. Furthermore, our method improves explainability by explicitly revealing the navigational intentions underlying each predicted trajectory.
Authors: Frederik Hoppe, Lars Kleinemeier, Astrid Franz, Udo G\"obel
Abstract: Recent foundation models for tabular data achieve strong task-specific performance via in-context learning. Nevertheless, they focus on direct prediction by encapsulating both representation learning and task-specific inference inside a single, resource-intensive network. This work specifically focuses on representation learning, i.e., on transferable, task-agnostic embeddings. We systematically evaluate task-agnostic representations from tabular foundation models (TabPFN and TabICL) alongside with classical feature engineering (TableVectorizer) across a variety of application tasks as outlier detection (ADBench) and supervised learning (TabArena Lite). We find that simple TableVectorizer features achieve comparable or superior performance while being up to three orders of magnitude faster than tabular foundation models. The code is available at https://github.com/ContactSoftwareAI/TabEmbedBench.
Authors: Vincent-Daniel Yun, Junhyuk Jo, Sunwoo Lee
Abstract: Deep neural networks achieve outstanding performance in visual recognition tasks, yet their large number of parameters makes them less practical for real-world applications. Recently, one-shot pruning has emerged as an effective strategy for reducing model size without additional training. However, models trained with standard objective functions often suffer a significant drop in accuracy after aggressive pruning. Some existing pruning-robust optimizers, such as SAM, and CrAM, mitigate this accuracy drop by guiding the model toward flatter regions of the parameter space, but they inevitably incur non-negligible additional computations. We propose a Variance Amplifying Regularizer (VAR) that deliberately increases the variance of model parameters during training. Our study reveals an intriguing finding that parameters with higher variance exhibit greater pruning robustness. VAR exploits this property by promoting such variance in the weight distribution, thereby mitigating the adverse effects of pruning. We further provide a theoretical analysis of its convergence behavior, supported by extensive empirical results demonstrating the superior pruning robustness of VAR.
Authors: Chenyang Xu, Siming Li, Hao Wang
Abstract: Phonocardiogram (PCG) analysis is vital for cardiovascular disease diagnosis, yet the scarcity of labeled pathological data hinders the capability of AI systems. To bridge this, we introduce H-LDM, a Hierarchical Latent Diffusion Model for generating clinically accurate and controllable PCG signals from structured metadata. Our approach features: (1) a multi-scale VAE that learns a physiologically-disentangled latent space, separating rhythm, heart sounds, and murmurs; (2) a hierarchical text-to-biosignal pipeline that leverages rich clinical metadata for fine-grained control over 17 distinct conditions; and (3) an interpretable diffusion process guided by a novel Medical Attention module. Experiments on the PhysioNet CirCor dataset demonstrate state-of-the-art performance, achieving a Fr\'echet Audio Distance of 9.7, a 92% attribute disentanglement score, and 87.1% clinical validity confirmed by cardiologists. Augmenting diagnostic models with our synthetic data improves the accuracy of rare disease classification by 11.3\%. H-LDM establishes a new direction for data augmentation in cardiac diagnostics, bridging data scarcity with interpretable clinical insights.
Authors: Yuwen Zhang, Viet Tran, Paul Weng
Abstract: In clinical machine learning, the coexistence of multiple models with comparable performance -- a manifestation of the Rashomon Effect -- poses fundamental challenges for trustworthy deployment and evaluation. Small, imbalanced, and noisy datasets, coupled with high-dimensional and weakly identified clinical features, amplify this multiplicity and make conventional validation schemes unreliable. As a result, selecting among equally performing models becomes uncertain, particularly when resource constraints and operational priorities are not considered by conventional metrics like F1 score. To address these issues, we propose two complementary tools for robust model assessment and selection: Intervention Efficiency (IE) and the Perturbation Validation Framework (PVF). IE is a capacity-aware metric that quantifies how efficiently a model identifies actionable true positives when only limited interventions are feasible, thereby linking predictive performance with clinical utility. PVF introduces a structured approach to assess the stability of models under data perturbations, identifying models whose performance remains most invariant across noisy or shifted validation sets. Empirical results on synthetic and real-world healthcare datasets show that using these tools facilitates the selection of models that generalize more robustly and align with capacity constraints, offering a new direction for tackling the Rashomon Effect in clinical settings.
Authors: Aneesh Barthakur, Luiz F. O. Chamon
Abstract: As machine learning applications grow increasingly ubiquitous and complex, they face an increasing set of requirements beyond accuracy. The prevalent approach to handle this challenge is to aggregate a weighted combination of requirement violation penalties into the training objective. To be effective, this approach requires careful tuning of these hyperparameters (weights), involving trial-and-error and cross-validation, which becomes ineffective even for a moderate number of requirements. These issues are exacerbated when the requirements involve parities or equalities, as is the case in fairness and boundary value problems. An alternative technique uses constrained optimization to formulate these learning problems. Yet, existing approximation and generalization guarantees do not apply to problems involving equality constraints. In this work, we derive a generalization theory for equality-constrained statistical learning problems, showing that their solutions can be approximated using samples and rich parametrizations. Using these results, we propose a practical algorithm based on solving a sequence of unconstrained, empirical learning problems. We showcase its effectiveness and the new formulations enabled by equality constraints in fair learning, interpolating classifiers, and boundary value problems.
Authors: Nanxi Chen, Sifan Wang, Rujin Ma, Airong Chen, Chuanjie Cui
Abstract: Physics-informed neural networks (PINNs) represent a new paradigm for solving partial differential equations (PDEs) by integrating physical laws into the learning process of neural networks. However, despite their foundational role, the hidden irreversibility implied by the Second Law of Thermodynamics is often neglected during training, leading to unphysical solutions or even training failures in conventional PINNs. In this paper, we identify this critical gap and introduce a simple, generalized, yet robust irreversibility-regularized strategy that enforces hidden physical laws as soft constraints during training. This approach ensures that the learned solutions consistently respect the intrinsic one-way nature of irreversible physical processes. Across a wide range of benchmarks spanning traveling wave propagation, steady combustion, ice melting, corrosion evolution, and crack propagation, we demonstrate that our regularization scheme reduces predictive errors by more than an order of magnitude, while requiring only minimal modification to existing PINN frameworks. We believe that the proposed framework is broadly applicable to a wide class of PDE-governed physical systems and will have significant impact within the scientific machine learning community.
Authors: Bastien Vuillod, Pierre-Alain Moellic, Jean-Max Dutertre
Abstract: Large models adaptation through Federated Learning (FL) addresses a wide range of use cases and is enabled by Parameter-Efficient Fine-Tuning techniques such as Low-Rank Adaptation (LoRA). However, this distributed learning paradigm faces several security threats, particularly to its integrity, such as backdoor attacks that aim to inject malicious behavior during the local training steps of certain clients. We present the first analysis of the influence of LoRA on state-of-the-art backdoor attacks targeting model adaptation in FL. Specifically, we focus on backdoor lifespan, a critical characteristic in FL, that can vary depending on the attack scenario and the attacker's ability to effectively inject the backdoor. A key finding in our experiments is that for an optimally injected backdoor, the backdoor persistence after the attack is longer when the LoRA's rank is lower. Importantly, our work highlights evaluation issues of backdoor attacks against FL and contributes to the development of more robust and fair evaluations of backdoor attacks, enhancing the reliability of risk assessments for critical FL systems. Our code is publicly available.
Authors: Haobin Li, Mouxing Yang, Xi Peng
Abstract: Recently, the general-to-customized paradigm has emerged as the dominant approach for Cross-Modal Retrieval (CMR), which reconciles the distribution shift problem between the source domain and the target domain. However, existing general-to-customized CMR methods typically assume that the entire target-domain data is available, which is easily violated in real-world scenarios and thus inevitably suffer from the query shift (QS) problem. Specifically, query shift embraces the following two characteristics and thus poses new challenges to CMR. i) Online Shift: real-world queries always arrive in an online manner, rendering it impractical to access the entire query set beforehand for customization approaches; ii) Diverse Shift: even with domain customization, the CMR models struggle to satisfy queries from diverse users or scenarios, leaving an urgent need to accommodate diverse queries. In this paper, we observe that QS would not only undermine the well-structured common space inherited from the source model, but also steer the model toward forgetting the indispensable general knowledge for CMR. Inspired by the observations, we propose a novel method for achieving online and harmonious adaptation against QS, dubbed Robust adaptation with quEry ShifT (REST). To deal with online shift, REST first refines the retrieval results to formulate the query predictions and accordingly designs a QS-robust objective function on these predictions to preserve the well-established common space in an online manner. As for tackling the more challenging diverse shift, REST employs a gradient decoupling module to dexterously manipulate the gradients during the adaptation process, thus preventing the CMR model from forgetting the general knowledge. Extensive experiments on 20 benchmarks across three CMR tasks verify the effectiveness of our method against QS.
Authors: Xiaowei Xu, Justin Sonneck, Hongxiao Wang, Roman Burkard, Hendrik Wohrle, Anton Grabmasier, Matthias Gunzer, Jianxu Chen
Abstract: Autonomous migration is essential for the function of immune cells such as neutrophils and plays a pivotal role in diverse diseases. Recently, we introduced ComplexEye, a multi-lens array microscope comprising 16 independent aberration-corrected glass lenses arranged at the pitch of a 96-well plate, capable of capturing high-resolution movies of migrating cells. This architecture enables high-throughput live-cell video microscopy for migration analysis, supporting routine quantification of autonomous motility with strong potential for clinical translation. However, ComplexEye and similar high-throughput imaging platforms generate data at an exponential rate, imposing substantial burdens on storage and transmission. To address this challenge, we present FlowRoI, a fast optical-flow-based region of interest (RoI) extraction framework designed for high-throughput image compression in immune cell migration studies. FlowRoI estimates optical flow between consecutive frames and derives RoI masks that reliably cover nearly all migrating cells. The raw image and its corresponding RoI mask are then jointly encoded using JPEG2000 to enable RoI-aware compression. FlowRoI operates with high computational efficiency, achieving runtimes comparable to standard JPEG2000 and reaching an average throughput of about 30 frames per second on a modern laptop equipped with an Intel i7-1255U CPU. In terms of image quality, FlowRoI yields higher peak signal-to-noise ratio (PSNR) in cellular regions and achieves 2.0-2.2x higher compression rates at matched PSNR compared to standard JPEG2000.
Authors: Andrey Okhotin, Maksim Nakhodnov, Nikita Kazeev, Andrey E Ustyuzhanin, Dmitry Vetrov
Abstract: In recent years, diffusion-based models have demonstrated exceptional performance in searching for simultaneously stable, unique, and novel (S.U.N.) crystalline materials. However, most of these models don't have the ability to change the number of atoms in the crystal during the generation process, which limits the variability of model sampling trajectories. In this paper, we demonstrate the severity of this restriction and introduce a simple yet powerful technique, mirage infusion, which enables diffusion models to change the state of the atoms that make up the crystal from existent to non-existent (mirage) and vice versa. We show that this technique improves model quality by up to $\times2.5$ compared to the same model without this modification. The resulting model, Mirage Atom Diffusion (MiAD), is an equivariant joint diffusion model for de novo crystal generation that is capable of altering the number of atoms during the generation process. MiAD achieves an $8.2\%$ S.U.N. rate on the MP-20 dataset, which substantially exceeds existing state-of-the-art approaches. The source code can be found at \href{https://github.com/andrey-okhotin/miad.git}{\texttt{github.com/andrey-okhotin/miad}}.
Authors: Emanuele Palumbo, Sorawit Saengkyongam, Maria R. Cervera, Jens Behrmann, Andrew C. Miller, Guillermo Sapiro, Christina Heinze-Deml, Antoine Wehenkel
Abstract: Continuous cardiovascular monitoring can play a key role in precision health. However, some fundamental cardiac biomarkers of interest, including stroke volume and cardiac output, require invasive measurements, e.g., arterial pressure waveforms (APW). As a non-invasive alternative, photoplethysmography (PPG) measurements are routinely collected in hospital settings. Unfortunately, the prediction of key cardiac biomarkers from PPG instead of APW remains an open challenge, further complicated by the scarcity of annotated PPG measurements. As a solution, we propose a hybrid approach that uses hemodynamic simulations and unlabeled clinical data to estimate cardiovascular biomarkers directly from PPG signals. Our hybrid model combines a conditional variational autoencoder trained on paired PPG-APW data with a conditional density estimator of cardiac biomarkers trained on labeled simulated APW segments. As a key result, our experiments demonstrate that the proposed approach can detect fluctuations of cardiac output and stroke volume and outperform a supervised baseline in monitoring temporal changes in these biomarkers.
Authors: Nicola Rares Franco, Lorenzo Tedesco
Abstract: We introduce conditional push-forward neural networks (CPFN), a generative framework for conditional distribution estimation. Instead of directly modeling the conditional density $f_{Y|X}$, CPFN learns a stochastic map $\varphi=\varphi(x,u)$ such that $\varphi(x,U)$ and $Y|X=x$ follow approximately the same law, with $U$ a suitable random vector of pre-defined latent variables. This enables efficient conditional sampling and straightforward estimation of conditional statistics through Monte Carlo methods. The model is trained via an objective function derived from a Kullback-Leibler formulation, without requiring invertibility or adversarial training. We establish a near-asymptotic consistency result and demonstrate experimentally that CPFN can achieve performance competitive with, or even superior to, state-of-the-art methods, including kernel estimators, tree-based algorithms, and popular deep learning techniques, all while remaining lightweight and easy to train.
Authors: Cl\'ement Dumas
Abstract: Mechanistic interpretability research requires reliable tools for analyzing transformer internals across diverse architectures. Current approaches face a fundamental tradeoff: custom implementations like TransformerLens ensure consistent interfaces but require coding a manual adaptation for each architecture, introducing numerical mismatch with the original models, while direct HuggingFace access through NNsight preserves exact behavior but lacks standardization across models. To bridge this gap, we develop nnterp, a lightweight wrapper around NNsight that provides a unified interface for transformer analysis while preserving original HuggingFace implementations. Through automatic module renaming and comprehensive validation testing, nnterp enables researchers to write intervention code once and deploy it across 50+ model variants spanning 16 architecture families. The library includes built-in implementations of common interpretability methods (logit lens, patchscope, activation steering) and provides direct access to attention probabilities for models that support it. By packaging validation tests with the library, researchers can verify compatibility with custom models locally. nnterp bridges the gap between correctness and usability in mechanistic interpretability tooling.
Authors: Diego Armando P\'erez-Rosero, Danna Valentina Salazar-Dubois, Juan Camilo Lugo-Rojas, Andr\'es Marino \'Alvarez-Meza, Germ\'an Castellanos-Dominguez
Abstract: These notes provide a self-contained introduction to kernel methods and their geometric foundations in machine learning. Starting from the construction of Hilbert spaces, we develop the theory of positive definite kernels, reproducing kernel Hilbert spaces (RKHS), and Hilbert-Schmidt operators, emphasizing their role in statistical estimation and representation of probability measures. Classical concepts such as covariance, regression, and information measures are revisited through the lens of Hilbert space geometry. We also introduce kernel density estimation, kernel embeddings of distributions, and the Maximum Mean Discrepancy (MMD). The exposition is designed to serve as a foundation for more advanced topics, including Gaussian processes, kernel Bayesian inference, and functional analytic approaches to modern machine learning.
Authors: Jintao Zhang, Mingyue Cheng, Zirui Liu, Xianquan Wang, Yitong Zhou, Qi Liu
Abstract: Time series generation is critical for a wide range of applications, which greatly supports downstream analytical and decision-making tasks. However, the inherent temporal heterogeneous induced by localized perturbations present significant challenges for generating structurally consistent time series. While flow matching provides a promising paradigm by modeling temporal dynamics through trajectory-level supervision, it fails to adequately capture abrupt transitions in perturbed time series, as the use of globally shared parameters constrains the velocity field to a unified representation. To address these limitations, we introduce \textbf{PAFM}, a \textbf{P}erturbation-\textbf{A}ware \textbf{F}low \textbf{M}atching framework that models perturbed trajectories to ensure stable and structurally consistent time series generation. The framework incorporates perturbation-guided training to simulate localized disturbances and leverages a dual-path velocity field to capture trajectory deviations under perturbation, enabling refined modeling of perturbed behavior to enhance the structural coherence. In order to further improve sensitivity to trajectory perturbations while enhancing expressiveness, a mixture-of-experts decoder with flow routing dynamically allocates modeling capacity in response to different trajectory dynamics. Extensive experiments on both unconditional and conditional generation tasks demonstrate that PAFM consistently outperforms strong baselines. Code is available at https://anonymous.4open.science/r/PAFM-03B2.
Authors: Jiawei Yi, Ping Gong, Youhui Bai, Jiaqi Ruan, Shengnan Wang, Pengcheng Wang, Haibo Wang, Weiguang Wang, Xia Zhu, Feng Wu, Cheng Li
Abstract: The growth of million-token LLMs exposes the scalability limits of inference systems, where the KVCache dominates memory usage and data transfer overhead. Recent offloading systems migrate the KVCache to CPU memory and incorporate top-k attention to reduce the volume of data transferred from the CPU, while further applying system-level optimizations such as on-GPU caching and prefetching to lower transfer overhead. However, they overlook the CPU bottleneck in three aspects: (1) substantial overhead of fine-grained dynamic cache management performed on the CPU side, (2) significant transfer overhead from poor PCIe bandwidth utilization caused by heavy gathering operations at the CPU side, and (3) GPU runtime bubbles introduced by coarse-grained CPU-centric synchronization. To address these challenges, we propose CLO, a CPU-light KVCache offloading system via algorithm-system co-design. CLO features: (1) a coarse-grained head-wise approximate on-GPU caching strategy with negligible cache management cost, (2) seamless combination of data prefetching and on-GPU persistent caching for lower transfer overhead, (3) a zero-copy transfer engine to fully exploit PCIe bandwidth, and a GPU-centric synchronization method to eliminate GPU stalls. Evaluation on two widely-used LLMs demonstrates that CLO achieves comparable accuracy to state-of-the-art systems, while substantially minimizing CPU overhead, fully utilizing PCIe bandwidth, thus improving decoding throughput by 9.3%-66.6%. Our results highlight that algorithm-system co-design is essential for memory-constrained LLM inference on modern GPU platforms. We open source CLO at https://github.com/CommediaJW/CLO.
Authors: Hao Qian, Shikui Tu, Lei Xu
Abstract: Diffusion and flow matching models have recently emerged as promising approaches for peptide binder design. Despite their progress, these models still face two major challenges. First, categorical sampling of discrete residue types collapses their continuous parameters into onehot assignments, while continuous variables (e.g., atom positions) evolve smoothly throughout the generation process. This mismatch disrupts the update dynamics and results in suboptimal performance. Second, current models assume unimodal distributions for side-chain torsion angles, which conflicts with the inherently multimodal nature of side chain rotameric states and limits prediction accuracy. To address these limitations, we introduce PepBFN, the first Bayesian flow network for full atom peptide design that directly models parameter distributions in fully continuous space. Specifically, PepBFN models discrete residue types by learning their continuous parameter distributions, enabling joint and smooth Bayesian updates with other continuous structural parameters. It further employs a novel Gaussian mixture based Bayesian flow to capture the multimodal side chain rotameric states and a Matrix Fisher based Riemannian flow to directly model residue orientations on the $\mathrm{SO}(3)$ manifold. Together, these parameter distributions are progressively refined via Bayesian updates, yielding smooth and coherent peptide generation. Experiments on side chain packing, reverse folding, and binder design tasks demonstrate the strong potential of PepBFN in computational peptide design.
Authors: Youran Zhou, Mohamed Reda Bouadjenek, Sunil Aryal
Abstract: Incomplete data are common in real-world tabular applications, where numerical, categorical, and discrete attributes coexist within a single dataset. This heterogeneous structure presents significant challenges for existing diffusion-based imputation models, which typically assume a homogeneous feature space and rely on stochastic denoising trajectories. Such assumptions make it difficult to maintain conditional consistency, and they often lead to information collapse for categorical variables or instability when numerical variables require deterministic updates. These limitations indicate that a single diffusion process is insufficient for mixed-type tabular imputation. We propose a hybrid deterministic diffusion framework that separates heterogeneous features into two complementary generative channels. A continuous DDIM-based channel provides efficient and stable deterministic denoising for numerical variables, while a discrete latent-path diffusion channel, inspired by loopholing-based discrete diffusion, models categorical and discrete features without leaving their valid sample manifolds. The two channels are trained under a unified conditional imputation objective, enabling coherent reconstruction of mixed-type incomplete data. Extensive experiments on multiple real-world datasets show that the proposed framework achieves higher imputation accuracy, more stable sampling trajectories, and improved robustness across MCAR, MAR, and MNAR settings compared with existing diffusion-based and classical methods. These results demonstrate the importance of structure-aware diffusion processes for advancing deep learning approaches to incomplete tabular data.
Authors: Jaume Ros, Alessio Arleo, Fernando Paulovich
Abstract: Dimensionality Reduction (DR) techniques are commonly used for the visual exploration and analysis of high-dimensional data due to their ability to project datasets of high-dimensional points onto the 2D plane. However, projecting datasets in lower dimensions often entails some distortion, which is not necessarily easy to recognize but can lead users to misleading conclusions. Several Projection Quality Metrics (PQMs) have been developed as tools to quantify the goodness-of-fit of a DR projection; however, they mostly focus on measuring how well the projection captures the global or local structure of the data, without taking into account the visual distortion of the resulting plots, thus often ignoring the presence of outliers or artifacts that can mislead a visual analysis of the projection. In this work, we introduce the Warping Index (WI), a new metric for measuring the quality of DR projections onto the 2D plane, based on the assumption that the correct preservation of empty regions between points is of crucial importance towards a faithful visual representation of the data.
Authors: Adam Hazimeh, Alessandro Favero, Pascal Frossard
Abstract: Task arithmetic has recently emerged as a promising method for editing pre-trained \textit{open-vocabulary} models, offering a cost-effective alternative to standard multi-task fine-tuning. However, despite the abundance of \textit{closed-vocabulary} models that are not pre-trained with language supervision, applying task arithmetic to these models remains unexplored. In this paper, we deploy and study task addition in closed-vocabulary image classification models. We consider different pre-training schemes and find that \textit{weight disentanglement} -- the property enabling task arithmetic -- is a general consequence of pre-training, as it appears in different pre-trained closed-vocabulary models. In fact, we find that pre-trained closed-vocabulary vision transformers can also be edited with task arithmetic, achieving high task addition performance and enabling the efficient deployment of multi-task models. Finally, we demonstrate that simple linear probing is a competitive baseline to task addition. Overall, our findings expand the applicability of task arithmetic to a broader class of pre-trained models and open the way for more efficient use of pre-trained models in diverse settings.
Authors: Ankush Kadu, Ashwanth Krishnan
Abstract: Enabling agents to learn from experience and generalize across diverse tasks without task-specific training remains a fundamental challenge in reinforcement learning and decision-making. While recent approaches have explored episodic memory (Reflexion), gradient-based prompt optimization (TextGrad),and hierarchical task decomposition independently, their potential for synergistic integration remains unexplored. We introduce ReflexGrad, a novel architecture that tightly couples three complementary mechanisms: (1) LLM-based hierarchical TODO decomposition for strategic planning, (2) history-aware causal reflection that analyzes recent action patterns to identify failure root causes and enable within-trial learning, and (3) gradient-based optimization for systematic improvement. Unlike prior work relying on few-shot demonstrations, our system achieves true zero-shot generalization through pure LLM semantic reasoning,requiring no task-specific examples, fine-tuning, or hardcoded similarity metrics. Evaluated on ALFWorld benchmark tasks, ReflexGrad demonstrates 67% zero-shot success rate on Trial 0 without any prior task experience or demonstrations, establishing effective performance on first exposure. Through empirical analysis, we identify the architectural mechanisms underlying stable convergence (zero action loops) and effective cross-task transfer (67% to 78% improvement).Our work demonstrates that synergistic integration of complementary learning mechanisms enables robust zero-shot generalization that approaches few-shot baselines from prior work.
Authors: Marco Locatelli, Arjen Hommersom, Roberto Clemens Cerioli, Daniela Besozzi, Fabio Stella
Abstract: Learning the parameters of Partially Observable Markov Decision Processes (POMDPs) from limited data is a significant challenge. We introduce the Fuzzy MAP EM algorithm, a novel approach that incorporates expert knowledge into the parameter estimation process by enriching the Expectation Maximization (EM) framework with fuzzy pseudo-counts derived from an expert-defined fuzzy model. This integration naturally reformulates the problem as a Maximum A Posteriori (MAP) estimation, effectively guiding learning in environments with limited data. In synthetic medical simulations, our method consistently outperforms the standard EM algorithm under both low-data and high-noise conditions. Furthermore, a case study on Myasthenia Gravis illustrates the ability of the Fuzzy MAP EM algorithm to recover a clinically coherent POMDP, demonstrating its potential as a practical tool for data-efficient modeling in healthcare.
Authors: Ivy Yuqian Yang, David Yu Zhang
Abstract: Scientific idea generation and selection requires exploration following a target probability distribution. In contrast, current AI benchmarks have objectively correct answers, and training large language models (LLMs) via reinforcement learning against these benchmarks discourages probabilistic exploration. Here, we conducted systematic experiments requesting LLMs to produce outputs following simple probabilistic distributions, and found that all modern LLMs tested grossly fail to follow the distributions. For example, requesting a binary output of "1" 49% of the time produces an answer of "0" nearly 100% of the time. This step function-like behavior of near-exclusively generating the output with marginally highest probability even overrules even strong in-built LLM biases.
Authors: Yuchen Luo, Xinyu Li, Liuhua Peng, Mingming Gong
Abstract: In multivariate time series forecasting (MTSF), accurately modeling the intricate dependencies among multiple variables remains a significant challenge due to the inherent limitations of traditional approaches. Most existing models adopt either \textbf{channel-independent} (CI) or \textbf{channel-dependent} (CD) strategies, each presenting distinct drawbacks. CI methods fail to leverage the potential insights from inter-channel interactions, resulting in models that may not fully exploit the underlying statistical dependencies present in the data. Conversely, CD approaches often incorporate too much extraneous information, risking model overfitting and predictive inefficiency. To address these issues, we introduce the Adaptive Forecasting Transformer (\textbf{Adapformer}), an advanced Transformer-based framework that merges the benefits of CI and CD methodologies through effective channel management. The core of Adapformer lies in its dual-stage encoder-decoder architecture, which includes the \textbf{A}daptive \textbf{C}hannel \textbf{E}nhancer (\textbf{ACE}) for enriching embedding processes and the \textbf{A}daptive \textbf{C}hannel \textbf{F}orecaster (\textbf{ACF}) for refining the predictions. ACE enhances token representations by selectively incorporating essential dependencies, while ACF streamlines the decoding process by focusing on the most relevant covariates, substantially reducing noise and redundancy. Our rigorous testing on diverse datasets shows that Adapformer achieves superior performance over existing models, enhancing both predictive accuracy and computational efficiency, thus making it state-of-the-art in MTSF.
Authors: Vaskar Chakma, MD Jaheid Hasan Nerab, Abdur Rouf, Abu Sayed, Hossem MD Saim, Md. Nournabi Khan
Abstract: Smoking continues to be a major preventable cause of death worldwide, affecting millions through damage to the heart, metabolism, liver, and kidneys. However, current medical screening methods often miss the early warning signs of smoking-related health problems, leading to late-stage diagnoses when treatment options become limited. This study presents a systematic comparative evaluation of machine learning approaches for smoking-related health risk assessment, emphasizing clinical interpretability and practical deployment over algorithmic innovation. We analyzed health screening data from 55,691 individuals, examining various health indicators, including body measurements, blood tests, and demographic information. We tested three advanced prediction algorithms - Random Forest, XGBoost, and LightGBM - to determine which could most accurately identify people at high risk. This study employed a cross-sectional design to classify current smoking status based on health screening biomarkers, not to predict future disease development. Our Random Forest model performed best, achieving an Area Under the Curve (AUC) of 0.926, meaning it could reliably distinguish between high-risk and lower-risk individuals. Using SHAP (SHapley Additive exPlanations) analysis to understand what the model was detecting, we found that key health markers played crucial roles in prediction: blood pressure levels, triglyceride concentrations, liver enzyme readings, and kidney function indicators (serum creatinine) were the strongest signals of declining health in smokers.
Authors: Abolfazl Younesi, Leon Kiss, Zahra Najafabadi Samani, Juan Aznar Poveda, Thomas Fahringer
Abstract: Federated learning (FL) enables collaborative model training while preserving data privacy. However, it remains vulnerable to malicious clients who compromise model integrity through Byzantine attacks, data poisoning, or adaptive adversarial behaviors. Existing defense mechanisms rely on static thresholds and binary classification, failing to adapt to evolving client behaviors in real-world deployments. We propose FLARE, an adaptive reputation-based framework that transforms client reliability assessment from binary decisions to a continuous, multi-dimensional trust evaluation. FLARE integrates: (i) a multi-dimensional reputation score capturing performance consistency, statistical anomaly indicators, and temporal behavior, (ii) a self-calibrating adaptive threshold mechanism that adjusts security strictness based on model convergence and recent attack intensity, (iii) reputation-weighted aggregation with soft exclusion to proportionally limit suspicious contributions rather than eliminating clients outright, and (iv) a Local Differential Privacy (LDP) mechanism enabling reputation scoring on privatized client updates. We further introduce a highly evasive Statistical Mimicry (SM) attack, a benchmark adversary that blends honest gradients with synthetic perturbations and persistent drift to remain undetected by traditional filters. Extensive experiments with 100 clients on MNIST, CIFAR-10, and SVHN demonstrate that FLARE maintains high model accuracy and converges faster than state-of-the-art Byzantine-robust methods under diverse attack types, including label flipping, gradient scaling, adaptive attacks, ALIE, and SM. FLARE improves robustness by up to 16% and preserves model convergence within 30% of the non-attacked baseline, while achieving strong malicious-client detection performance with minimal computational overhead. https://github.com/Anonymous0-0paper/FLARE
Authors: Fu-Ming Guo, Yingfang Fan
Abstract: Adaptive optimizers with decoupled weight decay, such as AdamW, are the de facto standard for pre-training large transformer-based generative models. Yet the quadratic nature of the $\ell_2$ penalty embedded in weight decay drives all parameters toward the origin at the same rate, making the update vulnerable to rare but extreme gradient directions and often over-penalizing well-conditioned coordinates. We propose AdamHuberDecay, a drop-in replacement for AdamW that substitutes the $\ell_2$ penalty with a decoupled smooth Huber regularizer. The resulting update decays parameters quadratically while their magnitude remains below a threshold $\delta$, and linearly ($\ell_1$-like) once they exceed $\delta$, yielding (i) bounded regularization gradients, (ii) invariance to per-coordinate second-moment rescaling, and (iii) stronger sparsity pressure on overgrown weights. We derive the closed-form decoupled Huber decay step and show how to integrate it with any Adam-family optimizer at $O(1)$ extra cost. Extensive experiments on GPT-2 and GPT-3 pre-training demonstrate that AdamHuberDecay (a) converges 10-15% faster in wall-clock time, (b) reduces validation perplexity by up to 4 points, (c) delivers performance improvements of 2.5-4.7% across downstream tasks, and (d) yields visibly sparser weight histograms that translate into 20-30% memory savings after magnitude pruning, without tuning the decay coefficient beyond the default grid used for AdamW. Ablations confirm robustness to outlier gradients and large-batch regimes, together with theoretical analyses that bound the expected parameter norm under noisy updates. AdamHuberDecay therefore provides a simple, principled path toward more efficient and resilient training of next-generation foundational generative transformers.
Authors: Tzu-Hsuan Chou, Chun-Nan Chou
Abstract: Large language models (LLMs) have shown a remarkable ability to generalize beyond their pre-training data, and fine-tuning LLMs can elevate performance to human-level and beyond. However, in real-world scenarios, lacking labeled data often prevents practitioners from obtaining well-performing models, thereby forcing practitioners to highly rely on prompt-based approaches that are often tedious, inefficient, and driven by trial and error. To alleviate this issue of lacking labeled data, we present a learning framework integrating LLMs with active learning for unlabeled dataset (LAUD). LAUD mitigates the cold-start problem by constructing an initial label set with zero-shot learning. Experimental results show that LLMs derived from LAUD outperform LLMs with zero-shot or few-shot learning on commodity name classification tasks, demonstrating the effectiveness of LAUD.
Authors: Christof Naumzik, Abdurahman Maarouf, Stefan Feuerriegel, Markus Weinmann
Abstract: Online ratings influence customer decision-making, yet standard aggregation methods, such as the sample mean, fail to adapt to quality changes over time and ignore review heterogeneity (e.g., review sentiment, a review's helpfulness). To address these challenges, we demonstrate the value of using the Gaussian process (GP) framework for rating aggregation. Specifically, we present a tailored GP model that captures the dynamics of ratings over time while additionally accounting for review heterogeneity. Based on 121,123 ratings from Yelp, we compare the predictive power of different rating aggregation methods in predicting future ratings, thereby finding that the GP model is considerably more accurate and reduces the mean absolute error by 10.2% compared to the sample mean. Our findings have important implications for marketing practitioners and customers. By moving beyond means, designers of online reputation systems can display more informative and adaptive aggregated rating scores that are accurate signals of expected customer satisfaction.
Authors: Antonia Ebner, Christoph Bartmann, Sonja Topf, Sohvi Luukkonen, Johannes Schimunek, G\"unter Klambauer
Abstract: Deep learning's rise since the early 2010s has transformed fields like computer vision and natural language processing and strongly influenced biomedical research. For drug discovery specifically, a key inflection - akin to vision's "ImageNet moment" - arrived in 2015, when deep neural networks surpassed traditional approaches on the Tox21 Data Challenge. This milestone accelerated the adoption of deep learning across the pharmaceutical industry, and today most major companies have integrated these methods into their research pipelines. After the Tox21 Challenge concluded, its dataset was included in several established benchmarks, such as MoleculeNet and the Open Graph Benchmark. However, during these integrations, the dataset was altered and labels were imputed or manufactured, resulting in a loss of comparability across studies. Consequently, the extent to which bioactivity and toxicity prediction methods have improved over the past decade remains unclear. To this end, we introduce a reproducible leaderboard, hosted on Hugging Face with the original Tox21 Challenge dataset, together with a set of baseline and representative methods. The current version of the leaderboard indicates that the original Tox21 winner - the ensemble-based DeepTox method - and the descriptor-based self-normalizing neural networks introduced in 2017, continue to perform competitively and rank among the top methods for toxicity prediction, leaving it unclear whether substantial progress in toxicity prediction has been achieved over the past decade. As part of this work, we make all baselines and evaluated models publicly accessible for inference via standardized API calls to Hugging Face Spaces.
Authors: Haiqing Zhu, Tijana Zrnic, Celestine Mendler-D\"unner
Abstract: On many learning platforms, the optimization criteria guiding model training reflect the priorities of the designer rather than those of the individuals they affect. Consequently, users may act strategically to obtain more favorable outcomes, effectively contesting the platform's predictions. While past work has studied strategic user behavior on learning platforms, the focus has largely been on strategic responses to a deployed model, without considering the behavior of other users. In contrast, look-ahead reasoning takes into account that user actions are coupled, and -- at scale -- impact future predictions. Within this framework, we first formalize level-$k$ thinking, a concept from behavioral economics, where users aim to outsmart their peers by looking one step ahead. We show that, while convergence to an equilibrium is accelerated, the equilibrium remains the same, providing no benefit of higher-level reasoning for individuals in the long run. Then, we focus on collective reasoning, where users take coordinated actions by optimizing through their joint impact on the model. By contrasting collective with selfish behavior, we characterize the benefits and limits of coordination; a new notion of alignment between the learner's and the users' utilities emerges as a key concept. We discuss connections to several related mathematical frameworks, including strategic classification, performative prediction, and algorithmic collective action.
Authors: Junfeng Wu, Hadjer Benmeziane, Kaoutar El Maghraoui, Liu Liu, Yinan Wang
Abstract: Spatiotemporal data mining (STDM) has a wide range of applications in various complex physical systems (CPS), i.e., transportation, manufacturing, healthcare, etc. Among all the proposed methods, the Convolutional Long Short-Term Memory (ConvLSTM) has proved to be generalizable and extendable in different applications and has multiple variants achieving state-of-the-art performance in various STDM applications. However, ConvLSTM and its variants are computationally expensive, which makes them inapplicable in edge devices with limited computational resources. With the emerging need for edge computing in CPS, efficient AI is essential to reduce the computational cost while preserving the model performance. Common methods of efficient AI are developed to reduce redundancy in model capacity (i.e., model pruning, compression, etc.). However, spatiotemporal data mining naturally requires extensive model capacity, as the embedded dependencies in spatiotemporal data are complex and hard to capture, which limits the model redundancy. Instead, there is a fairly high level of data and feature redundancy that introduces an unnecessary computational burden, which has been largely overlooked in existing research. Therefore, we developed a novel framework SparseST, that pioneered in exploiting data sparsity to develop an efficient spatiotemporal model. In addition, we explore and approximate the Pareto front between model performance and computational efficiency by designing a multi-objective composite loss function, which provides a practical guide for practitioners to adjust the model according to computational resource constraints and the performance requirements of downstream tasks.
Authors: Ali Amin, Raichelle Aniceto, Ashwin Balakrishna, Kevin Black, Ken Conley, Grace Connors, James Darpinian, Karan Dhabalia, Jared DiCarlo, Danny Driess, Michael Equi, Adnan Esmail, Yunhao Fang, Chelsea Finn, Catherine Glossop, Thomas Godden, Ivan Goryachev, Lachy Groom, Hunter Hancock, Karol Hausman, Gashon Hussein, Brian Ichter, Szymon Jakubczak, Rowan Jen, Tim Jones, Ben Katz, Liyiming Ke, Chandra Kuchi, Marinda Lamb, Devin LeBlanc, Sergey Levine, Adrian Li-Bell, Yao Lu, Vishnu Mano, Mohith Mothukuri, Suraj Nair, Karl Pertsch, Allen Z. Ren, Charvi Sharma, Lucy Xiaoyang Shi, Laura Smith, Jost Tobias Springenberg, Kyle Stachowicz, Will Stoeckle, Alex Swerdlow, James Tanner, Marcel Torne, Quan Vuong, Anna Walling, Haohuan Wang, Blake Williams, Sukwon Yoo, Lili Yu, Ury Zhilinsky, Zhiyuan Zhou
Abstract: We study how vision-language-action (VLA) models can improve through real-world deployments via reinforcement learning (RL). We present a general-purpose method, RL with Experience and Corrections via Advantage-conditioned Policies (RECAP), that provides for RL training of VLAs via advantage conditioning. Our method incorporates heterogeneous data into the self-improvement process, including demonstrations, data from on-policy collection, and expert teleoperated interventions provided during autonomous execution. RECAP starts by pre-training a generalist VLA with offline RL, which we call $\pi^{*}_{0.6}$, that can then be specialized to attain high performance on downstream tasks through on-robot data collection. We show that the $\pi^{*}_{0.6}$ model trained with the full RECAP method can fold laundry in real homes, reliably assemble boxes, and make espresso drinks using a professional espresso machine. On some of the hardest tasks, RECAP more than doubles task throughput and roughly halves the task failure rate.
Authors: Sparsh Gupta, Kamalavasan Kamalakkannan, Maxim Moraru, Galen Shipman, Patrick Diehl
Abstract: Scientific applications continue to rely on legacy Fortran codebases originally developed for homogeneous, CPU-based systems. As High-Performance Computing (HPC) shifts toward heterogeneous GPU-accelerated architectures, many accelerators lack native Fortran bindings, creating an urgent need to modernize legacy codes for portability. Frameworks like Kokkos provide performance portability and a single-source C++ abstraction, but manual Fortran-to-Kokkos porting demands significant expertise and time. Large language models (LLMs) have shown promise in source-to-source code generation, yet their use in fully autonomous workflows for translating and optimizing parallel code remains largely unexplored, especially for performance portability across diverse hardware. This paper presents an agentic AI workflow where specialized LLM "agents" collaborate to translate, validate, compile, run, test, debug, and optimize Fortran kernels into portable Kokkos C++ programs. Results show the pipeline modernizes a range of benchmark kernels, producing performance-portable Kokkos codes across hardware partitions. Paid OpenAI models such as GPT-5 and o4-mini-high executed the workflow for only a few U.S. dollars, generating optimized codes that surpassed Fortran baselines, whereas open-source models like Llama4-Maverick often failed to yield functional codes. This work demonstrates the feasibility of agentic AI for Fortran-to-Kokkos transformation and offers a pathway for autonomously modernizing legacy scientific applications to run portably and efficiently on diverse supercomputers. It further highlights the potential of LLM-driven agentic systems to perform structured, domain-specific reasoning tasks in scientific and systems-oriented applications.
Authors: Aris Daniilidis, Alberto Dom\'inguez Corella, Philipp Wissgott
Abstract: We analyze an algorithm for assigning weights prior to scalarization in discrete multi-objective problems arising from data analysis. The algorithm evolves the weights (the relevance of features) by a replicator-type dynamic on the standard simplex, with update indices computed from a normalized data matrix. We prove that the resulting sequence converges globally to a unique interior equilibrium, yielding non-degenerate limiting weights. The method, originally inspired by evolutionary game theory, differs from standard weighting schemes in that it is analytically tractable with provable convergence.
Authors: Omkar Desai (Syracuse University), Ziyang Jiao (Syracuse University), Shuyi Pei (Samsung Semiconductor Inc.), Janki Bhimani (Florida International University), Bryan S. Kim (Syracuse University)
Abstract: Input data preprocessing is a common bottleneck when concurrently training multimedia machine learning (ML) models in modern systems. To alleviate these bottlenecks and reduce the training time for concurrent jobs, we present Seneca, a data loading system that optimizes cache partitioning and data sampling for the data storage and ingestion (DSI) pipeline. The design of Seneca contains two key techniques. First, Seneca uses a performance model for the data pipeline to optimally partition the cache for three different forms of data (encoded, decoded, and augmented). Second, Seneca opportunistically serves cached data over uncached ones during random batch sampling so that concurrent jobs benefit from each other. We implement Seneca by modifying PyTorch and demonstrate its effectiveness by comparing it against several state-of-the-art caching systems for DNN training. Seneca reduces the makespan by 45.23% compared to PyTorch and increases data processing throughput by up to 3.45x compared to the next best dataloader.
Authors: Huseyin Goksu
Abstract: Graph Neural Networks (GNNs) based on spectral filters, such as the Adaptive Orthogonal Polynomial Filter (AOPF) class (e.g., LaguerreNet), have shown promise in unifying the solutions for heterophily and over-smoothing. However, these single-filter models suffer from a "compromise" problem, as their single adaptive parameter (e.g., alpha) must learn a suboptimal, averaged response across the entire graph spectrum. In this paper, we propose DualLaguerreNet, a novel GNN architecture that solves this by introducing "Decoupled Spectral Flexibility." DualLaguerreNet splits the graph Laplacian into two operators, L_low (low-frequency) and L_high (high-frequency), and learns two independent, adaptive Laguerre polynomial filters, parameterized by alpha_1 and alpha_2, respectively. This work, however, uncovers a deeper finding. While our experiments show DualLaguerreNet's flexibility allows it to achieve state-of-the-art results on complex heterophilic tasks (outperforming LaguerreNet), it simultaneously underperforms on simpler, homophilic tasks. We identify this as a fundamental "Flexibility-Stability Trade-off". The increased parameterization (2x filter parameters and 2x model parameters) leads to overfitting on simple tasks, demonstrating that the "compromise" of simpler models acts as a crucial regularizer. This paper presents a new SOTA architecture for heterophily while providing a critical analysis of the bias-variance trade-off inherent in adaptive GNN filter design.
Authors: Huseyin Goksu
Abstract: Spectral Graph Neural Networks (GNNs) operating in the canonical [-1, 1] domain (like ChebyNet and its adaptive generalization, L-JacobiNet) face a fundamental Flexibility-Stability Trade-off. Our previous work revealed a critical puzzle: the 2-parameter adaptive L-JacobiNet often suffered from high variance and was surprisingly outperformed by the 0-parameter, stabilized-static S-JacobiNet. This suggested that stabilization was more critical than adaptation in this domain. In this paper, we propose \textbf{GegenbauerNet}, a novel GNN filter based on the Gegenbauer polynomials, to find the Optimal Compromise in this trade-off. By enforcing symmetry (alpha=beta) but allowing a single shape parameter (lambda) to be learned, GegenbauerNet limits flexibility (variance) while escaping the fixed bias of S-JacobiNet. We demonstrate that GegenbauerNet (1-parameter) achieves superior performance in the key local filtering regime (K=2 on heterophilic graphs) where overfitting is minimal, validating the hypothesis that a controlled, symmetric degree of freedom is optimal. Furthermore, our comprehensive K-ablation study across homophilic and heterophilic graphs, using 7 diverse datasets, clarifies the domain's behavior: the fully adaptive L-JacobiNet maintains the highest performance on high-K filtering tasks, showing the value of maximum flexibility when regularization is managed. This study provides crucial design principles for GNN developers, showing that in the [-1, 1] spectral domain, the optimal filter depends critically on the target locality (K) and the acceptable level of design bias.
Authors: Moran Yanuka, Paul Dixon, Eyal Finkelshtein, Daniel Rotman, Raja Giryes
Abstract: Speculative decoding accelerates autoregressive speech generation by letting a fast draft model propose tokens that a larger target model verifies. However, for speech LLMs that generate acoustic tokens, exact token matching is overly restrictive: many discrete tokens are acoustically or semantically interchangeable, reducing acceptance rates and limiting speedups. We introduce Principled Coarse-Graining (PCG), which verifies proposals at the level of Acoustic Similarity Groups (ASGs) derived from the target model's embedding space. By splitting each token's probability mass across the overlapping groups that contain it, we define an overlap-aware coarse-grained distribution and perform rejection sampling on the resulting group variable. This yields an exactness guarantee at the group level while allowing the accepted draft token to stand in for any member of the group in practice. On LibriTTS, PCG increases acceptance and throughput relative to standard speculative decoding and prior speech-specific relaxations while maintaining intelligibility and speaker similarity. These results suggest acoustically aware, group-level acceptance as a simple and general way to accelerate speech token generation while maintaining speech quality.
Authors: Wenchao Yang, Weidong Yan, Wenkang Liu, Yulan Ma, Yang Li
Abstract: Large-scale pre-trained models hold significant potential for learning universal EEG representations. However, most existing methods, particularly autoregressive (AR) frameworks, primarily rely on straightforward temporal sequencing of multi-channel EEG data, which fails to capture the rich physiological characteristics inherent to EEG signals. Moreover, their time-centered modeling approach also limits the effective representation of the dynamic spatial topology of brain activity. To address these challenges and fully exploit the potential of large-scale EEG models, we propose a novel Topology Hierarchical Derived Brain Autoregressive Modeling (THD-BAR) for EEG generic representations. The core innovation of THD-BAR lies in the introduction of the Brain Topology Hierarchy (BTH), which establishes a multi-scale spatial order for EEG channels. This hierarchical structure enables a redefinition of autoregressive learning as a "next-scale-time prediction" problem, effectively capturing both spatial and temporal dynamics. Based on BTH, we design a Topology-Hierarchical Vector Quantized-Variational Autoencoder (THVQ-VAE) for multi-scale tokenization and develop an enhanced Brain Autoregressive (BAR) module with specialized masking strategies for prediction. Through extensive large-scale pre-training on 17 datasets, followed by rigorous validation on 10 downstream datasets spanning 5 distinct tasks, THD-BAR consistently outperforms existing methods. These results highlight the superior generalization and modeling capabilities of our proposed approach.
Authors: Seid H. Pourtakdoust, Amir H. Khodabakhsh
Abstract: This study presents a novel deep learning approach aimed at enhancing stochastic Gust Load Alleviation (GLA) specifically for compliant wings. The approach incorporates the concept of smooth wing camber variation, where the camber of the wing's chord is actively adjusted during flight using a control signal to achieve the desired aerodynamic loading. The proposed method employs a deep learning-based model predictive controller designed for probability density shaping. This controller effectively solves the probability density evolution equation through a custom Physics-Informed Neural Network (PINN) and utilizes Automatic Differentiation for Model Predictive Control (MPC) optimization. Comprehensive numerical simulations were conducted on a compliant wing (CW) model, evaluating performance of the proposed approach against stochastic gust profiles. The evaluation involved stochastic aerodynamic loads generated from Band-Limited White Noise (BLWN) and Dryden gust models. The evaluation were conducted for two distinct Compliant Chord Fractions (CCF). The results demonstrate the effectiveness of the proposed probability density shaping model predictive control in alleviating stochastic gust load and reducing wing tip deflection.
Authors: Juntang Yang, Mohamed Khalil Ben-Larbi
Abstract: This paper implements deep reinforcement learning (DRL) for spacecraft reorientation control with a single pointing keep-out zone. The Soft Actor-Critic (SAC) algorithm is adopted to handle continuous state and action space. A new state representation is designed to explicitly include a compact representation of the attitude constraint zone. The reward function is formulated to achieve the control objective while enforcing the attitude constraint. A curriculum learning approach is used for the agent training. Simulation results demonstrate the effectiveness of the proposed DRL-based method for spacecraft pointing-constrained attitude control.
Authors: Alexander Acker, Soeren Becker, Sasho Nedelkoski, Dominik Scheinert, Odej Kao, Philipp Wiesner
Abstract: Although LLM training is typically centralized with high-bandwidth interconnects and large compute budgets, emerging methods target communication-constrained training in distributed environments. The model trade-offs introduced by this shift remain underexplored, and our goal is to study them. We use the open-source nanochat project, a compact 8K-line full-stack ChatGPT-like implementation containing tokenization, pretraining, fine-tuning, and serving, as a controlled baseline. We implement the DiLoCo algorithm as a lightweight wrapper over nanochat's training loop, performing multiple local steps per worker before synchronization with an outer optimizer, effectively reducing communication by orders of magnitude. This inner-outer training is compared against a standard data-parallel (DDP) setup. Because nanochat is small and inspectable, it enables controlled pipeline adaptations and allows direct comparison with the conventional centralized baseline. DiLoCo achieves stable convergence and competitive loss in pretraining but yields worse MMLU, GSM8K, and HumanEval scores after mid-training and SFT. We discover that using DiLoCo-pretrained weights and running mid- and post-training with DDP fails to recover performance, revealing irreversible representation drift from asynchronous updates that impairs downstream alignment. We provide this implementation as an official fork of nanochat on GitHub.
Authors: Anthony Kiggundu, Bin Han, Hans D. Schotten
Abstract: We study how two information feeds, a closed-form Markov estimator of residual sojourn and an online trained actor-critic, affect reneging and jockeying in a dual M/M/1 system. Analytically, for unequal service rates and total-time patience, we show that total wait grows linearly so abandonment is inevitable and the probability of a successful jockey vanishes as the backlog approaches towards infinity. Furthermore, under a mild sub-linear error condition both information models yield the same asymptotic limits (robustness). We empirically validate these limits and quantify finite backlog differences. Our findings show that learned and analytic feeds produce different delays, reneging rates and transient jockeying behavior at practical sizes, but converge to the same asymptotic outcome implied by our theory. The results characterize when value-of-information matters (finite regimes) and when it does not (asymptotics), informing lightweight telemetry and decision-logic design for low-cost, jockeying-aware systems.
Authors: Dongdong Zhao, Ranxin Fang, Changtian Song, Zhihui Liu, Jianwen Xiang
Abstract: Open Set Recognition (OSR) requires models not only to accurately classify known classes but also to effectively reject unknown samples. However, when unknown samples are semantically similar to known classes, inter-class overlap in the feature space often causes models to assign unjustifiably high confidence to them, leading to misclassification as known classes -- a phenomenon known as overconfidence. This overconfidence undermines OSR by blurring the decision boundary between known and unknown classes. To address this issue, we propose a framework that explicitly mitigates overconfidence caused by inter-class overlap. The framework consists of two components: a perturbation-based uncertainty estimation module, which applies controllable parameter perturbations to generate diverse predictions and quantify predictive uncertainty, and an unknown detection module with distinct learning-based classifiers, implemented as a two-stage procedure, which leverages the estimated uncertainty to improve discrimination between known and unknown classes, thereby enhancing OSR performance. Experimental results on three public datasets show that the proposed framework achieves superior performance over existing OSR methods.
Authors: Mohammad Abdi, Francesca Meneghello, Francesco Restuccia
Abstract: Mobile devices increasingly require the parallel execution of several computing tasks offloaded at the wireless edge. Existing communication systems only support parallel transmissions at the bit level, which fundamentally limits the number of tasks that can be concurrently processed. To address this bottleneck, this paper introduces the new concept of Semantic Multiplexing. Our approach shifts stream multiplexing from bits to tasks by merging multiple task-related compressed representations into a single semantic representation. As such, Semantic Multiplexing can multiplex more tasks than the number of physical channels without adding antennas or widening bandwidth by extending the effective degrees of freedom at the semantic layer, without contradicting Shannon capacity rules. We have prototyped Semantic Multiplexing on an experimental testbed with Jetson Orin Nano and millimeter-wave software-defined radios and tested its performance on image classification and sentiment analysis while comparing to several existing baselines in semantic communications. Our experiments demonstrate that Semantic Multiplexing allows jointly processing multiple tasks at the semantic level while maintaining sufficient task accuracy. For example, image classification accuracy drops by less than 4% when increasing from 2 to 8 the number of tasks multiplexed over a 4$\times$4 channel. Semantic Multiplexing reduces latency, energy consumption, and communication load respectively by up to 8$\times$, 25$\times$, and 54$\times$ compared to the baselines while keeping comparable performance. We pledge to publicly share the complete software codebase and the collected datasets for reproducibility.
Authors: Yue Ling, Peiqi Zhang, Zhenyi Zhang, Peijie Zhou
Abstract: Single-cell RNA sequencing (scRNA-seq), especially temporally resolved datasets, enables genome-wide profiling of gene expression dynamics at single-cell resolution across discrete time points. However, current technologies provide only sparse, static snapshots of cell states and are inherently influenced by technical noise, complicating the inference and representation of continuous transcriptional dynamics. Although embedding methods can reduce dimensionality and mitigate technical noise, the majority of existing approaches typically treat trajectory inference separately from embedding construction, often neglecting temporal structure. To address this challenge, here we introduce CellStream, a novel deep learning framework that jointly learns embedding and cellular dynamics from single-cell snapshot data by integrating an autoencoder with unbalanced dynamical optimal transport. Compared to existing methods, CellStream generates dynamics-informed embeddings that robustly capture temporal developmental processes while maintaining high consistency with the underlying data manifold. We demonstrate CellStream's effectiveness on both simulated datasets and real scRNA-seq data, including spatial transcriptomics. Our experiments indicate significant quantitative improvements over state-of-the-art methods in representing cellular trajectories with enhanced temporal coherence and reduced noise sensitivity. Overall, CellStream provides a new tool for learning and representing continuous streams from the noisy, static snapshots of single-cell gene expression.
Authors: Pratik Chakraborty, Aryan Bhargava
Abstract: Proteins perform essential biological functions, and accurate classification of their sequences is critical for understanding structure-function relationships, enzyme mechanisms, and molecular interactions. This study presents a deep learning-based framework for functional group classification of protein sequences derived from the Protein Data Bank (PDB). Four architectures were implemented: Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM), CNN-BiLSTM hybrid, and CNN with Attention. Each model was trained using k-mer integer encoding to capture both local and long-range dependencies. Among these, the CNN achieved the highest validation accuracy of 91.8%, demonstrating the effectiveness of localized motif detection. Explainable AI techniques, including Grad-CAM and Integrated Gradients, were applied to interpret model predictions and identify biologically meaningful sequence motifs. The discovered motifs, enriched in histidine, aspartate, glutamate, and lysine, represent amino acid residues commonly found in catalytic and metal-binding regions of transferase enzymes. These findings highlight that deep learning models can uncover functionally relevant biochemical signatures, bridging the gap between predictive accuracy and biological interpretability in protein sequence analysis.
Authors: Zhaoxuan Wang, Weichen Kang, Yutian Han, Lingyuan Zhao, Bo Li
Abstract: Human Immunodeficiency Virus (HIV) has posed a major global health challenge for decades, and forecasting HIV diagnoses continues to be a critical area of research. However, capturing the complex spatial and temporal dependencies of HIV transmission remains challenging. Conventional Message Passing Neural Network (MPNN) models rely on a fixed binary adjacency matrix that only encodes geographic adjacency, which is unable to represent interactions between non-contiguous counties. Our study proposes a deep learning architecture Mobility-Aware Transformer-Message Passing Neural Network (MAT-MPNN) framework to predict county-level HIV diagnosis rates across California, Florida, and the New England region. The model combines temporal features extracted by a Transformer encoder with spatial relationships captured through a Mobility Graph Generator (MGG). The MGG improves conventional adjacency matrices by combining geographic and demographic information. Compared with the best-performing hybrid baseline, the Transformer MPNN model, MAT-MPNN reduced the Mean Squared Prediction Error (MSPE) by 27.9% in Florida, 39.1% in California, and 12.5% in New England, and improved the Predictive Model Choice Criterion (PMCC) by 7.7%, 3.5%, and 3.9%, respectively. MAT-MPNN also achieved better results than the Spatially Varying Auto-Regressive (SVAR) model in Florida and New England, with comparable performance in California. These results demonstrate that applying mobility-aware dynamic spatial structures substantially enhances predictive accuracy and calibration in spatiotemporal epidemiological prediction.
Authors: Mohammad Reza Shafie, Morteza Hajiabadi, Hamed Khosravi, Mobina Noori, Imtiaz Ahmed
Abstract: Microbial Fuel Cells (MFCs) offer a promising pathway for sustainable energy generation by converting organic matter into electricity through microbial processes. A key factor influencing MFC performance is the anode structure, where design and material properties play a crucial role. Existing predictive models struggle to capture the complex geometric dependencies necessary to optimize these structures. To solve this problem, we propose KANGURA: Kolmogorov-Arnold Network-Based Geometry-Aware Learning with Unified Representation Attention. KANGURA introduces a new approach to three-dimensional (3D) machine learning modeling. It formulates prediction as a function decomposition problem, where Kolmogorov-Arnold Network (KAN)- based representation learning reconstructs geometric relationships without a conventional multi- layer perceptron (MLP). To refine spatial understanding, geometry-disentangled representation learning separates structural variations into interpretable components, while unified attention mechanisms dynamically enhance critical geometric regions. Experimental results demonstrate that KANGURA outperforms over 15 state-of-the-art (SOTA) models on the ModelNet40 benchmark dataset, achieving 92.7% accuracy, and excels in a real-world MFC anode structure problem with 97% accuracy. This establishes KANGURA as a robust framework for 3D geometric modeling, unlocking new possibilities for optimizing complex structures in advanced manufacturing and quality-driven engineering applications.
Authors: Edward Raff, Ryan R. Curtin, Derek Everett, Robert J. Joyce, James Holt
Abstract: A classifier using byte n-grams as features is the only approach we have found fast enough to meet requirements in size (sub 2 MB), speed (multiple GB/s), and latency (sub 10 ms) for deployment in numerous malware detection scenarios. However, we've consistently found that 6-8 grams achieve the best accuracy on our production deployments but have been unable to deploy regularly updated models due to the high cost of finding the top-k most frequent n-grams over terabytes of executable programs. Because the Zipfian distribution well models the distribution of n-grams, we exploit its properties to develop a new top-k n-gram extractor that is up to $35\times$ faster than the previous best alternative. Using our new Zipf-Gramming algorithm, we are able to scale up our production training set and obtain up to 30\% improvement in AUC at detecting new malware. We show theoretically and empirically that our approach will select the top-k items with little error and the interplay between theory and engineering required to achieve these results.
Authors: Julian Soltes
Abstract: High-dimensional numerical optimization presents a persistent challenge. This paper introduces Quasi-Adaptive Search with Asymptotic Reinitialization (QUASAR), an evolutionary algorithm to accelerate convergence in complex, non-differentiable problems afflicted by the curse of dimensionality. Evaluated on the notoriously difficult CEC2017 benchmark suite of 29 functions, QUASAR achieved the lowest overall rank sum (150) using the Friedman test, significantly outperforming L-SHADE (229) and standard DE (305) in the dimension-variant trials. QUASAR also proves computationally efficient, with run times averaging $1.4 \text{x}$ faster than DE and $7.8 \text{x}$ faster than L-SHADE ($p \ll 0.001$) in the population-variant trials. Building upon Differential Evolution (DE), QUASAR introduces a highly stochastic architecture to dynamically balance exploration and exploitation. Inspired by the probabilistic behavior of quantum particles in a stellar core, the algorithm implements three primary components that augment standard DE mechanisms: 1) probabilistically selected mutation strategies and scaling factors; 2) rank-based crossover rates; 3) asymptotically decaying reinitialization that leverages a covariance matrix of the best solutions to introduce high-quality genetic diversity. QUASAR's performance establishes it as an effective, user-friendly optimizer for complex high-dimensional problems.
Authors: Abdul Rehman, Iqra Rasool, Ayesha Imran, Mohsen Ali, Waqas Sultani
Abstract: Digital hematopathology requires cell-level analysis across diverse disease categories, including malignant disorders (e.g., leukemia), infectious conditions (e.g., malaria), and non-malignant red blood cell disorders (e.g., sickle cell disease). Whether single-task, vision-language, WSI-optimized, or single-cell hematology models, these approaches share a key limitation, they cannot provide unified, multi-task, multi-modal reasoning across the complexities of digital hematopathology. To overcome these limitations, we propose Uni-Hema, a multi-task, unified model for digital hematopathology integrating detection, classification, segmentation, morphology prediction, and reasoning across multiple diseases. Uni-Hema leverages 46 publicly available datasets, encompassing over 700K images and 21K question-answer pairs, and is built upon Hema-Former, a multimodal module that bridges visual and textual representations at the hierarchy level for the different tasks (detection, classification, segmentation, morphology, mask language modeling and visual question answer) at different granularity. Extensive experiments demonstrate that Uni-Hema achieves comparable or superior performance to train on a single-task and single dataset models, across diverse hematological tasks, while providing interpretable, morphologically relevant insights at the single-cell level. Our framework establishes a new standard for multi-task and multi-modal digital hematopathology. The code will be made publicly available.
Authors: Chengrui Li, Yunmiao Wang, Yule Wang, Weihan Li, Dieter Jaeger, Anqi Wu
Abstract: Low-rank recurrent neural networks (lrRNNs) are a class of models that uncover low-dimensional latent dynamics underlying neural population activity. Although their functional connectivity is low-rank, it lacks disentanglement interpretations, making it difficult to assign distinct computational roles to different latent dimensions. To address this, we propose the Disentangled Recurrent Neural Network (DisRNN), a generative lrRNN framework that assumes group-wise independence among latent dynamics while allowing flexible within-group entanglement. These independent latent groups allow latent dynamics to evolve separately, but are internally rich for complex computation. We reformulate the lrRNN under a variational autoencoder (VAE) framework, enabling us to introduce a partial correlation penalty that encourages disentanglement between groups of latent dimensions. Experiments on synthetic, monkey M1, and mouse voltage imaging data show that DisRNN consistently improves the disentanglement and interpretability of learned neural latent trajectories in low-dimensional space and low-rank connectivity over baseline lrRNNs that do not encourage partial disentanglement.
Authors: Vasiliki Tassopoulou, Charis Stamouli, Haochang Shou, George J. Pappas, Christos Davatzikos
Abstract: Despite recent progress in predicting biomarker trajectories from real clinical data, uncertainty in the predictions poses high-stakes risks (e.g., misdiagnosis) that limit their clinical deployment. To enable safe and reliable use of such predictions in healthcare, we introduce a conformal method for uncertainty-calibrated prediction of biomarker trajectories resulting from randomly-timed clinical visits of patients. Our approach extends conformal prediction to the setting of randomly-timed trajectories via a novel nonconformity score that produces prediction bands guaranteed to cover the unknown biomarker trajectories with a user-prescribed probability. We apply our method across a wide range of standard and state-of-the-art predictors for two well-established brain biomarkers of Alzheimer's disease, using neuroimaging data from real clinical studies. We observe that our conformal prediction bands consistently achieve the desired coverage, while also being tighter than baseline prediction bands. To further account for population heterogeneity, we develop group-conditional conformal bands and test their coverage guarantees across various demographic and clinically relevant subpopulations. Moreover, we demonstrate the clinical utility of our conformal bands in identifying subjects at high risk of progression to Alzheimer's disease. Specifically, we introduce an uncertainty-calibrated risk score that enables the identification of 17.5% more high-risk subjects compared to standard risk scores, highlighting the value of uncertainty calibration in real-world clinical decision making. Our code is available at github.com/vatass/ConformalBiomarkerTrajectories.
Authors: Xiaoyu Zhang, Mingtao Hu, Sen Lu, Soohyeon Kim, Eric Yeu-Jer Lee, Yuyang Liu, Wei D. Lu
Abstract: State space models (SSMs) have recently emerged as a powerful framework for long sequence processing, outperforming traditional methods on diverse benchmarks. Fundamentally, SSMs can generalize both recurrent and convolutional networks and have been shown to even capture key functions of biological systems. Here we report an approach to implement SSMs in energy-efficient compute-in-memory (CIM) hardware to achieve real-time, event-driven processing. Our work re-parameterizes the model to function with real-valued coefficients and shared decay constants, reducing the complexity of model mapping onto practical hardware systems. By leveraging device dynamics and diagonalized state transition parameters, the state evolution can be natively implemented in crossbar-based CIM systems combined with memristors exhibiting short-term memory effects. Through this algorithm and hardware co-design, we show the proposed system offers both high accuracy and high energy efficiency while supporting fully asynchronous processing for event-based vision and audio tasks.
Authors: Rongsheng Qian, Chi Xu, Xiaoqiang Ma, Hao Fang, Yili Jin, William I. Atlas, Jiangchuan Liu
Abstract: Real-time imaging sonar has become an important tool for underwater monitoring in environments where optical sensing is unreliable. Its broader use is constrained by two coupled challenges: highly limited uplink bandwidth and severe sonar-specific artifacts (speckle, motion blur, reverberation, acoustic shadows) that affect up to 98% of frames. We present SCOPE, a self-supervised framework that jointly performs compression and artifact correction without clean-noise pairs or synthetic assumptions. SCOPE combines (i) Adaptive Codebook Compression (ACC), which learns frequency-encoded latent representations tailored to sonar, with (ii) Frequency-Aware Multiscale Segmentation (FAMS), which decomposes frames into low-frequency structure and sparse high-frequency dynamics while suppressing rapidly fluctuating artifacts. A hedging training strategy further guides frequency-aware learning using low-pass proxy pairs generated without labels. Evaluated on months of in-situ ARIS sonar data, SCOPE achieves a structural similarity index (SSIM) of 0.77, representing a 40% improvement over prior self-supervised denoising baselines, at bitrates down to <= 0.0118 bpp. It reduces uplink bandwidth by more than 80% while improving downstream detection. The system runs in real time, with 3.1 ms encoding on an embedded GPU and 97 ms full multi-layer decoding on the server end. SCOPE has been deployed for months in three Pacific Northwest rivers to support real-time salmon enumeration and environmental monitoring in the wild. Results demonstrate that learning frequency-structured latents enables practical, low-bitrate sonar streaming with preserved signal details under real-world deployment conditions.
Authors: Harold D. Chiang, Yukitoshi Matsushita, Taisuke Otsu
Abstract: We develop an empirical likelihood (EL) framework for random forests and related ensemble methods, providing a likelihood-based approach to quantify their statistical uncertainty. Exploiting the incomplete $U$-statistic structure inherent in ensemble predictions, we construct an EL statistic that is asymptotically chi-squared when subsampling induced by incompleteness is not overly sparse. Under sparser subsampling regimes, the EL statistic tends to over-cover due to loss of pivotality; we therefore propose a modified EL that restores pivotality through a simple adjustment. Our method retains key properties of EL while remaining computationally efficient. Theory for honest random forests and simulations demonstrate that modified EL achieves accurate coverage and practical reliability relative to existing inference methods.
Authors: Aaron Broukhim, Yiran Shen, Prithviraj Ammanabrolu, Nadir Weibel
Abstract: Despite the parallel challenges that audio and text domains face in evaluating generative model outputs, preference learning remains remarkably underexplored in audio applications. Through a PRISMA-guided systematic review of approximately 500 papers, we find that only 30 (6%) apply preference learning to audio tasks. Our analysis reveals a field in transition: pre-2021 works focused on emotion recognition using traditional ranking methods (rankSVM), while post-2021 studies have pivoted toward generation tasks employing modern RLHF frameworks. We identify three critical patterns: (1) the emergence of multi-dimensional evaluation strategies combining synthetic, automated, and human preferences; (2) inconsistent alignment between traditional metrics (WER, PESQ) and human judgments across different contexts; and (3) convergence on multi-stage training pipelines that combine reward signals. Our findings suggest that while preference learning shows promise for audio, particularly in capturing subjective qualities like naturalness and musicality, the field requires standardized benchmarks, higher-quality datasets, and systematic investigation of how temporal factors unique to audio impact preference learning frameworks.
Authors: Stuart H. Sul, Simran Arora, Benjamin F. Spector, Christopher R\'e
Abstract: Inter-GPU communication has become a major bottleneck for modern AI workloads as models scale and improvements in hardware compute throughput outpace improvements in interconnect bandwidth. Existing systems mitigate this through compute-communication overlap but often fail to meet theoretical peak performance across heterogeneous workloads and new accelerators. Instead of operator-specific techniques, we ask whether a small set of simple, reusable principles can systematically guide the design of optimal multi-GPU kernels. We present ParallelKittens (PK), a minimal CUDA framework that drastically simplifies the development of overlapped multi-GPU kernels. PK extends the ThunderKittens framework and embodies the principles of multi-GPU kernel design through eight core primitives and a unified programming template, derived from a comprehensive analysis of the factors that govern multi-GPU performance$\unicode{x2014}$data-transfer mechanisms, resource scheduling, and design overheads. We validate PK on both Hopper and Blackwell architectures. With fewer than 50 lines of device code, PK achieves up to $2.33 \times$ speedup for data- and tensor-parallel workloads, $4.08 \times$ for sequence-parallel workloads, and $1.22 \times$ for expert-parallel workloads.
Authors: Kevin I. Ruiz Vargas, Gabriel G. Galdino, Tsang Ing Ren, Alexandre L. Cunha
Abstract: We investigate image segmentation of cells under the lens of scalar fields. Our goal is to learn a continuous scalar field on image domains such that its segmentation produces robust instances for cells present in images. This field is a function parameterized by the trained network, and its segmentation is realized by the watershed method. The fields we experiment with are solutions to the Poisson partial differential equation and a diffusion mimicking the steady-state solution of the heat equation. These solutions are obtained by minimizing just the field residuals, no regularization is needed, providing a robust regression capable of diminishing the adverse impacts of outliers in the training data and allowing for sharp cell boundaries. A single tensor is all that is needed to train a \unet\ thus simplifying implementation, lowering training and inference times, hence reducing energy consumption, and requiring a small memory footprint, all attractive features in edge computing. We present competitive results on public datasets from the literature and show that our novel, simple yet geometrically insightful approach can achieve excellent cell segmentation results.
Authors: Matin Daghyani, Lyuyang Wang, Nima Hashemi, Bassant Medhat, Baraa Abdelsamad, Eros Rojas Velez, XiaoXiao Li, Michael Y. C. Tsang, Christina Luong, Teresa S. M. Tsang, Purang Abolmaesumi
Abstract: Purpose: Echocardiographic interpretation requires video-level reasoning and guideline-based measurement analysis, which current deep learning models for cardiac ultrasound do not support. We present EchoAgent, a framework that enables structured, interpretable automation for this domain. Methods: EchoAgent orchestrates specialized vision tools under Large Language Model (LLM) control to perform temporal localization, spatial measurement, and clinical interpretation. A key contribution is a measurement-feasibility prediction model that determines whether anatomical structures are reliably measurable in each frame, enabling autonomous tool selection. We curated a benchmark of diverse, clinically validated video-query pairs for evaluation. Results: EchoAgent achieves accurate, interpretable results despite added complexity of spatiotemporal video analysis. Outputs are grounded in visual evidence and clinical guidelines, supporting transparency and traceability. Conclusion: This work demonstrates the feasibility of agentic, guideline-aligned reasoning for echocardiographic video analysis, enabled by task-specific tools and full video-level automation. EchoAgent sets a new direction for trustworthy AI in cardiac ultrasound.
Authors: Nilay Kumar, Priyansh Bhandari, G. Maragatham
Abstract: Human emotions are difficult to convey through words and are often abstracted in the process; however, electroencephalogram (EEG) signals can offer a more direct lens into emotional brain activity. Recent studies show that deep learning models can process these signals to perform emotion recognition with high accuracy. However, many existing approaches overlook the dynamic interplay between distinct brain regions, which can be crucial to understanding how emotions unfold and evolve over time, potentially aiding in more accurate emotion recognition. To address this, we propose RBTransformer, a Transformer-based neural network architecture that models inter-cortical neural dynamics of the brain in latent space to better capture structured neural interactions for effective EEG-based emotion recognition. First, the EEG signals are converted into Band Differential Entropy (BDE) tokens, which are then passed through Electrode Identity embeddings to retain spatial provenance. These tokens are processed through successive inter-cortical multi-head attention blocks that construct an electrode x electrode attention matrix, allowing the model to learn the inter-cortical neural dependencies. The resulting features are then passed through a classification head to obtain the final prediction. We conducted extensive experiments, specifically under subject-dependent settings, on the SEED, DEAP, and DREAMER datasets, over all three dimensions, Valence, Arousal, and Dominance (for DEAP and DREAMER), under both binary and multi-class classification settings. The results demonstrate that the proposed RBTransformer outperforms all previous state-of-the-art methods across all three datasets, over all three dimensions under both classification settings. The source code is available at: https://github.com/nnilayy/RBTransformer.
Authors: Chengpeng Li, Farnaz Behrang, August Shi, Peng Liu
Abstract: Flaky tests that non-deterministically pass or fail waste developer time and slow release cycles. While large language models (LLMs) show promise for automatically repairing flaky tests, existing approaches like FlakyDoctor fail in industrial settings due to the context problem: providing either too little context (missing critical production code) or too much context (overwhelming the LLM with irrelevant information). We present FlakyGuard, which addresses this problem by treating code as a graph structure and using selective graph exploration to find only the most relevant context. Evaluation on real-world flaky tests from industrial repositories shows that FlakyGuard repairs 47.6 % of reproducible flaky tests with 51.8 % of the fixes accepted by developers. Besides it outperforms state-of-the-art approaches by at least 22 % in repair success rate. Developer surveys confirm that 100 % find FlakyGuard's root cause explanations useful.
Authors: SangHyuk Kim, Daniel Haehn, Sumientra Rampersad
Abstract: Humans can easily identify anatomical planes (axial, coronal, and sagittal) on a 2D MRI slice, but automated systems struggle with this task. Missing plane orientation metadata can complicate analysis, increase domain shift when merging heterogeneous datasets, and reduce accuracy of diagnostic classifiers. This study develops a classifier that accurately generates plane orientation metadata. We adopt a 2.5D context-aware model that leverages multi-slice information to avoid ambiguity from isolated slices and enable robust feature learning. We train the 2.5D model on both 3D slice sequences and static 2D images. While our 2D reference model achieves 98.74% accuracy, our 2.5D method raises this to 99.49%, reducing errors by 60%, highlighting the importance of 2.5D context. We validate the utility of our generated metadata in a brain tumor detection task. A gated strategy selectively uses metadata-enhanced predictions based on uncertainty scores, boosting accuracy from 97.0% with an image-only model to 98.0%, reducing misdiagnoses by 33.3%. We integrate our plane orientation model into an interactive web application and provide it open-source.
Authors: Pradeep Kumar Sharma, Ishaan Puri, Mantinder Jit Singh, Swapnil Shivaprasad, Hritvik Shrivastava
Abstract: Modern codebases evolve continuously: files are renamed or deleted; public APIs drift; behavior shifts within otherwise familiar modules. A model trained yesterday to map a developer's natural-language question to the exact set of repository file paths that matter will degrade tomorrow, even if the questions themselves look unchanged. In this paper we study, at system scale and across several widely used repositories, how to keep such a model fresh without surrendering retention on earlier code. We frame freshness as a form of domain drift between a base snapshot and the current HEAD, and we compare three families of update strategies: (A) Full Refresh, retraining the entire model at the new snapshot; (B) In-Context Learning (ICL) that injects recent deltas (raw git diffs or concise English summaries) at inference; and (C) Incremental Fine-Tuning (Inc-FT) on delta-derived training sets, with carefully controlled NEW:OLD mixing to mitigate catastrophic forgetting. We contribute an alias-aware evaluation protocol that credits rename while never rewarding deleted paths, and a practical Forgetting Probe that quantifies residual emissions of obsolete paths. Across Flask, SQLAlchemy, Pandas, and Poetry, Inc-FT with old-aware mixes delivers the best overall balance on mixed sets, ICL with English delta summaries delivers the fastest new-code lift when training is not feasible, and Full Refresh remains the ceiling when maximum NEW accuracy matters. We also compare Git-diff Inc-FT to full-file Inc-FT, showing that diffs excel in rename/delete-heavy windows while full-file context wins in behavior-change-heavy windows.
Authors: Sungik Choi, Hankook Lee, Moontae Lee
Abstract: AI-generated image detection has become crucial with the rapid advancement of vision-generative models. Instead of training detectors tailored to specific datasets, we study a training-free approach leveraging self-supervised models without requiring prior data knowledge. These models, pre-trained with augmentations like RandomResizedCrop, learn to produce consistent representations across varying resolutions. Motivated by this, we propose WaRPAD, a training-free AI-generated image detection algorithm based on self-supervised models. Since neighborhood pixel differences in images are highly sensitive to resizing operations, WaRPAD first defines a base score function that quantifies the sensitivity of image embeddings to perturbations along high-frequency directions extracted via Haar wavelet decomposition. To simulate robustness against cropping augmentation, we rescale each image to a multiple of the models input size, divide it into smaller patches, and compute the base score for each patch. The final detection score is then obtained by averaging the scores across all patches. We validate WaRPAD on real datasets of diverse resolutions and domains, and images generated by 23 different generative models. Our method consistently achieves competitive performance and demonstrates strong robustness to test-time corruptions. Furthermore, as invariance to RandomResizedCrop is a common training scheme across self-supervised models, we show that WaRPAD is applicable across self-supervised models.
Authors: Mara Daniels, Philippe Rigollet
Abstract: We introduce a highly expressive class of function approximators called Splat Regression Models. Model outputs are mixtures of heterogeneous and anisotropic bump functions, termed splats, each weighted by an output vector. The power of splat modeling lies in its ability to locally adjust the scale and direction of each splat, achieving both high interpretability and accuracy. Fitting splat models reduces to optimization over the space of mixing measures, which can be implemented using Wasserstein-Fisher-Rao gradient flows. As a byproduct, we recover the popular Gaussian Splatting methodology as a special case, providing a unified theoretical framework for this state-of-the-art technique that clearly disambiguates the inverse problem, the model, and the optimization algorithm. Through numerical experiments, we demonstrate that the resulting models and algorithms constitute a flexible and promising approach for solving diverse approximation, estimation, and inverse problems involving low-dimensional data.
Authors: Cutter Beck, Evan Smith, Khagendra Katuwal, Rudra Kafle, Jacob Whitehill
Abstract: Coronal holes (CHs) are low-activity, low-density solar coronal regions with open magnetic field lines (Cranmer 2009). In the extreme ultraviolet (EUV) spectrum, CHs appear as dark patches. Using daily hand-drawn maps from the Space Weather Prediction Center (SWPC), we developed a semi-automated pipeline to digitize the SWPC maps into binary segmentation masks. The resulting masks constitute the CHASM-SWPC dataset, a high-quality dataset to train and test automated CH detection models, which is released with this paper. We developed CHASM (Coronal Hole Annotation using Semi-automatic Methods), a software tool for semi-automatic annotation that enables users to rapidly and accurately annotate SWPC maps. The CHASM tool enabled us to annotate 1,111 CH masks, comprising the CHASM-SWPC-1111 dataset. We then trained multiple CHRONNOS (Coronal Hole RecOgnition Neural Network Over multi-Spectral-data) architecture (Jarolim et al. 2021) neural networks using the CHASM-SWPC dataset and compared their performance. Training the CHRONNOS neural network on these data achieved an accuracy of 0.9805, a True Skill Statistic (TSS) of 0.6807, and an intersection-over-union (IoU) of 0.5668, which is higher than the original pretrained CHRONNOS model Jarolim et al. (2021) achieved an accuracy of 0.9708, a TSS of 0.6749, and an IoU of 0.4805, when evaluated on the CHASM-SWPC-1111 test set.
Authors: Zifan Wang, Georgios Pantazis, Sergio Grammatico, Michael M. Zavlanos, Karl H. Johansson
Abstract: We study a class of distributionally robust games where agents are allowed to heterogeneously choose their risk aversion with respect to distributional shifts of the uncertainty. In our formulation, heterogeneous Wasserstein ball constraints on each distribution are enforced through a penalty function leveraging a Lagrangian formulation. We then formulate the distributionally robust Nash equilibrium problem and show that under certain assumptions it is equivalent to a finite-dimensional variational inequality problem with a strongly monotone mapping. We then design an approximate Nash equilibrium seeking algorithm and prove convergence of the average regret to a quantity that diminishes with the number of iterations, thus learning the desired equilibrium up to an a priori specified accuracy. Numerical simulations corroborate our theoretical findings.
Authors: Shenglin Zhang, Ziang Chen, Zijing Que, Yilun Liu, Yongqian Sun, Sicheng Wei, Dan Pei, Hailin Li
Abstract: Log anomaly detection, which is critical for identifying system failures and preempting security breaches, detects irregular patterns within large volumes of log data, and impacts domains such as service reliability, performance optimization, and database log analysis. Modern log anomaly detection methods rely on training deep learning models on clean, anomaly-free log sequences. However, obtaining such clean log data requires costly and tedious human labeling, and existing automatic cleaning methods fail to fully integrate the specific characteristics and actual semantics of logs in their purification process. In this paper, we propose a cost-aware, rule-enhanced purification framework, LogPurge, that automatically selects a sufficient subset of normal log sequences from contamination log sequences to train a anomaly detection model. Our approach involves a two-stage filtering algorithm: In the first stage, we use a large language model (LLM) to remove clustered anomalous patterns and enhance system rules to improve LLM's understanding of system logs; in the second stage, we utilize a divide-and-conquer strategy that decomposes the remaining contaminated regions into smaller subproblems, allowing each to be effectively purified through the first stage procedure. Our experiments, conducted on two public datasets and one industrial dataset, show that our method significantly removes an average of 98.74% of anomalies while retaining 82.39% of normal samples. Compared to the latest unsupervised log sample selection algorithms, our method achieves F-1 score improvements of 35.7% and 84.11% on the public datasets, and an impressive 149.72% F-1 improvement on the private dataset, demonstrating the effectiveness of our approach.
Authors: Ajesh Koyatan Chathoth, Stephen Lee
Abstract: Sensor data-based recognition systems are widely used in various applications, such as gait-based authentication and human activity recognition (HAR). Modern wearable and smart devices feature various built-in Inertial Measurement Unit (IMU) sensors, and such sensor-based measurements can be fed to a machine learning-based model to train and classify human activities. While deep learning-based models have proven successful in classifying human activity and gestures, they pose various security risks. In our paper, we discuss a novel dynamic trigger-generation technique for performing black-box adversarial attacks on sensor data-based IoT systems. Our empirical analysis shows that the attack is successful on various datasets and classifier models with minimal perturbation on the input data. We also provide a detailed comparative analysis of performance and stealthiness to various other poisoning techniques found in backdoor attacks. We also discuss some adversarial defense mechanisms and their impact on the effectiveness of our trigger-generation technique.
Authors: Sithmini Ranasingha, Agasthi Haputhanthri, Hansa Marasinghe, Nima Wickramasinghe, Kithmin Wickremasinghe, Jithangi Wanigasinghe, Chamira U. S. Edussooriya, Joshua P. Kulasingham
Abstract: Neonates are highly susceptible to seizures, often leading to short or long-term neurological impairments. However, clinical manifestations of neonatal seizures are subtle and often lead to misdiagnoses. This increases the risk of prolonged, untreated seizure activity and subsequent brain injury. Continuous video electroencephalogram (cEEG) monitoring is the gold standard for seizure detection. However, this is an expensive evaluation that requires expertise and time. In this study, we propose a convolutional neural network-based model for early prediction of neonatal seizures by distinguishing between interictal and preictal states of the EEG. Our model is patient-independent, enabling generalization across multiple subjects, and utilizes mel-frequency cepstral coefficient matrices extracted from multichannel EEG and electrocardiogram (ECG) signals as input features. Trained and validated on the Helsinki neonatal EEG dataset with 10-fold cross-validation, the proposed model achieved an average accuracy of 97.52%, sensitivity of 98.31%, specificity of 96.39%, and F1-score of 97.95%, enabling accurate seizure prediction up to 30 minutes before onset. The inclusion of ECG alongside EEG improved the F1-score by 1.42%, while the incorporation of an attention mechanism yielded an additional 0.5% improvement. To enhance transparency, we incorporated SHapley Additive exPlanations (SHAP) as an explainable artificial intelligence method to interpret the model and provided localization of seizure focus using scalp plots. The overall results demonstrate the model's potential for minimally supervised deployment in neonatal intensive care units, enabling timely and reliable prediction of neonatal seizures, while demonstrating strong generalization capability across unseen subjects through transfer learning.
Authors: Sabiha Afroz, Redwan Ibne Seraj Khan, Hadeel Albahar, Jingoo Han, Ali R. Butt
Abstract: Training large language models (LLMs) in the cloud faces growing memory bottlenecks due to the limited capacity and high cost of GPUs. While GPU memory offloading to CPU and NVMe has made large-scale training more feasible, existing approaches suffer from high tensor migration latency and suboptimal device memory utilization, ultimately increasing training time and cloud costs. To address these challenges, we present 10Cache, a resource-aware tensor caching and migration system that accelerates LLM training by intelligently coordinating memory usage across GPU, CPU, and NVMe tiers. 10Cache profiles tensor execution order to construct prefetch policies, allocates memory buffers in pinned memory based on tensor size distributions, and reuses memory buffers to minimize allocation overhead. Designed for cloud-scale deployments, 10Cache improves memory efficiency and reduces reliance on high-end GPUs. Across diverse LLM workloads, it achieves up to 2x speedup in training time, improves GPU cache hit rate by up to 86.6x, and increases CPU/GPU memory utilization by up to 2.15x and 1.33x, respectively, compared to state-of-the-art offloading methods. These results demonstrate that 10Cache is a practical and scalable solution for optimizing LLM training throughput and resource efficiency in cloud environments.
Authors: Xiang Luo, Chang Liu, Gang Xiong, Chen Yang, Gaopeng Gou, Yaochen Ren, Zhen Li
Abstract: Fine-grained identification of IDS-flagged suspicious traffic is crucial in cybersecurity. In practice, cyber threats evolve continuously, making the discovery of novel malicious traffic a critical necessity as well as the identification of known classes. Recent studies have advanced this goal with deep models, but they often rely on task-specific architectures that limit transferability and require per-dataset tuning. In this paper we introduce MalRAG, the first LLM driven retrieval-augmented framework for open-set malicious traffic identification. MalRAG freezes the LLM and operates via comprehensive traffic knowledge construction, adaptive retrieval, and prompt engineering. Concretely, we construct a multi-view traffic database by mining prior malicious traffic from content, structural, and temporal perspectives. Furthermore, we introduce a Coverage-Enhanced Retrieval Algorithm that queries across these views to assemble the most probable candidates, thereby improving the inclusion of correct evidence. We then employ Traffic-Aware Adaptive Pruning to select a variable subset of these candidates based on traffic-aware similarity scores, suppressing incorrect matches and yielding reliable retrieved evidence. Moreover, we develop a suite of guidance prompts where task instruction, evidence referencing, and decision guidance are integrated with the retrieved evidence to improve LLM performance. Across diverse real-world datasets and settings, MalRAG delivers state-of-the-art results in both fine-grained identification of known classes and novel malicious traffic discovery. Ablation and deep-dive analyses further show that MalRAG effective leverages LLM capabilities yet achieves open-set malicious traffic identification without relying on a specific LLM.
Authors: Omkar Mahesh Kashyap, Padegal Amit, Madhav Kashyap, Ashwini M Joshi, Shylaja SS
Abstract: Aspect-Based Sentiment Analysis (ABSA) predicts sentiment polarity for specific aspect terms, a task made difficult by conflicting sentiments across aspects and the sparse context of short texts. Prior graph-based approaches model only pairwise dependencies, forcing them to construct multiple graphs for different relational views. These introduce redundancy, parameter overhead, and error propagation during fusion, limiting robustness in short-text, low-resource settings. We present HyperABSA, a dynamic hypergraph framework that induces aspect-opinion structures through sample-specific hierarchical clustering. To construct these hyperedges, we introduce a novel acceleration-fallback cutoff for hierarchical clustering, which adaptively determines the level of granularity. Experiments on three benchmarks (Lap14, Rest14, MAMS) show consistent improvements over strong graph baselines, with substantial gains when paired with RoBERTa backbones. These results position dynamic hypergraph construction as an efficient, powerful alternative for ABSA, with potential extensions to other short-text NLP tasks.
Authors: Renjie Chen, Viet Anh Nguyen, Huifu Xu
Abstract: We propose a distributionally robust formulation for simultaneously estimating the covariance matrix and the precision matrix of a random vector.The proposed model minimizes the worst-case weighted sum of the Frobenius loss of the covariance estimator and Stein's loss of the precision matrix estimator against all distributions from an ambiguity set centered at the nominal distribution. The radius of the ambiguity set is measured via convex spectral divergence. We demonstrate that the proposed distributionally robust estimation model can be reduced to a convex optimization problem, thereby yielding quasi-analytical estimators. The joint estimators are shown to be nonlinear shrinkage estimators. The eigenvalues of the estimators are shrunk nonlinearly towards a positive scalar, where the scalar is determined by the weight coefficient of the loss terms. By tuning the coefficient carefully, the shrinkage corrects the spectral bias of the empirical covariance/precision matrix estimator. By this property, we call the proposed joint estimator the Spectral concentrated COvariance and Precision matrix Estimator (SCOPE). We demonstrate that the shrinkage effect improves the condition number of the estimator. We provide a parameter-tuning scheme that adjusts the shrinkage target and intensity that is asymptotically optimal. Numerical experiments on synthetic and real data show that our shrinkage estimators perform competitively against state-of-the-art estimators in practical applications.
Authors: Alexander Christie, Matan Leibovich, Miguel Moscoso, Alexei Novikov, George Papanicolaou, Chrysoula Tsogka
Abstract: We develop an imaging algorithm that exploits strong scattering to achieve super-resolution in changing random media. The method processes large and diverse array datasets using sparse dictionary learning, clustering, and multidimensional scaling. Starting from random initializations, the algorithm reliably extracts the unknown medium properties necessary for accurate imaging using back-propagation, $\ell_2$ or $\ell_1$ methods. Remarkably, scattering enhances resolution beyond homogeneous medium limits. When abundant data are available, the algorithm allows the realization of super-resolution in imaging.
Authors: Yuhua Jiang, Shuang Cheng, Yan Ding, Feifei Gao, Biqing Qi
Abstract: Vision-language-action (VLA) models have recently emerged as a powerful paradigm for building generalist robots. However, traditional VLA models that generate actions through flow matching (FM) typically rely on rigid and uniform time schedules, i.e., synchronous FM (SFM). Without action context awareness and asynchronous self-correction, SFM becomes unstable in long-horizon tasks, where a single action error can cascade into failure. In this work, we propose asynchronous flow matching VLA (AsyncVLA), a novel framework that introduces temporal flexibility in asynchronous FM (AFM) and enables self-correction in action generation. AsyncVLA breaks from the vanilla SFM in VLA models by generating the action tokens in a non-uniform time schedule with action context awareness. Besides, our method introduces the confidence rater to extract confidence of the initially generated actions, enabling the model to selectively refine inaccurate action tokens before execution. Moreover, we propose a unified training procedure for SFM and AFM that endows a single model with both modes, improving KV-cache utilization. Extensive experiments on robotic manipulation benchmarks demonstrate that AsyncVLA is data-efficient and exhibits self-correction ability. AsyncVLA achieves state-of-the-art results across general embodied evaluations due to its asynchronous generation in AFM. Our code is available at https://github.com/YuhuaJiang2002/AsyncVLA.
Authors: Alessio Zanga, Marco Scutari, Fabio Stella
Abstract: Causal discovery combines data with knowledge provided by experts to learn the DAG representing the causal relationships between a given set of variables. When data are scarce, bagging is used to measure our confidence in an average DAG obtained by aggregating bootstrapped DAGs. However, the aggregation step has received little attention from the specialized literature: the average DAG is constructed using only the confidence in the individual edges of the bootstrapped DAGs, thus disregarding complex higher-order edge structures. In this paper, we introduce a novel theoretical framework based on higher-order structures and describe a new DAG aggregation algorithm. We perform a simulation study, discussing the advantages and limitations of the proposed approach. Our proposal is both computationally efficient and effective, outperforming state-of-the-art solutions, especially in low sample size regimes and under high dimensionality settings.
Authors: N Dinesh Reddy, Sudeep Pillai
Abstract: We introduce Orion, a visual agent framework that can take in any modality and generate any modality. Using an agentic framework with multiple tool-calling capabilities, Orion is designed for visual AI tasks and achieves state-of-the-art results. Unlike traditional vision-language models that produce descriptive outputs, Orion orchestrates a suite of specialized computer vision tools, including object detection, keypoint localization, panoptic segmentation, Optical Character Recognition, and geometric analysis, to execute complex multi-step visual workflows. The system achieves competitive performance on MMMU, MMBench, DocVQA, and MMLongBench while extending monolithic vision-language models to production-grade visual intelligence. By combining neural perception with symbolic execution, Orion enables autonomous visual reasoning, marking a transition from passive visual understanding to active, tool-driven visual intelligence.
Authors: Pattaraphon Kenny Wongchamcharoen, Paul Glasserman
Abstract: Large language models (LLMs) are increasingly used in finance and economics, where prompt-based attempts against look-ahead bias implicitly assume that models understand chronology. We test this fundamental question with a series of chronological ordering tasks with increasing complexities over facts the model already knows from pre-training. Our tasks cover (1) chronological ordering, (2) conditional sorting (filter, then order), and (3) anachronism detection. We evaluate GPT-4.1, Claude-3.7 Sonnet, with and without Extended Thinking (ET), and GPT-5 across multiple reasoning-effort settings. Across models, Exact match rate drops sharply as sequences lengthen even while rank correlations stay high as LLMs largely preserve local order but struggle to maintain a single globally consistent timeline. In conditional sorting, most failures stem from the filtering step rather than the ordering step, but GPT-5 and Claude-3.7 Sonnet with Extended Thinking outshine normal models significantly. Lastly, anachronism detection is found to be the easiest task for the LLMs but performance still declines with increasingly overlapping timelines or entities. Overall, our main contribution is showing that allocating explicit reasoning budget helps with chronological ordering with GPT-5 at medium/high reasoning effort achieving flawless ordering at all lengths and perfect conditional sorting (both self-filtered and given-subset), whereas low/minimal effort degrades with longer lists, mirroring earlier models. Our findings delineate limits of current LLMs on chronological tasks, providing insights into task complexity, and demonstrate scenarios in which reasoning helps. These patterns are important for the real-time application of LLMs in finance. We release all code and evaluation templates to support full reproducibility.
Authors: Yuxiang Wang, Siwen Wang, Haowei Han, Ao Wang, Boya Liu, Yong Zhao, Chengbo Wu, Bin Zhu, Bin Qin, Xiaokai Zhou, Xiao Yan, Jiawei Jiang, Bo Du
Abstract: Operation recommendation for IoT devices refers to generating personalized device operations for users based on their context, such as historical operations, environment information, and device status. This task is crucial for enhancing user satisfaction and corporate profits. Existing recommendation models struggle with complex operation logic, diverse user preferences, and sensitive to suboptimal suggestions, limiting their applicability to IoT device operations. To address these issues, we propose DevPiolt, a LLM-based recommendation model for IoT device operations. Specifically, we first equip the LLM with fundamental domain knowledge of IoT operations via continual pre-training and multi-task fine-tuning. Then, we employ direct preference optimization to align the fine-tuned LLM with specific user preferences. Finally, we design a confidence-based exposure control mechanism to avoid negative user experiences from low-quality recommendations. Extensive experiments show that DevPiolt significantly outperforms baselines on all datasets, with an average improvement of 69.5% across all metrics. DevPiolt has been practically deployed in Xiaomi Home app for one quarter, providing daily operation recommendations to 255,000 users. Online experiment results indicate a 21.6% increase in unique visitor device coverage and a 29.1% increase in page view acceptance rates.
Authors: Yan Huang, Yongyi Su, Xin Lin, Le Zhang, Xun Xu
Abstract: The emergence of foundation models has substantially advanced zero-shot generalization in monocular depth estimation (MDE), as exemplified by the Depth Anything series. However, given access to some data from downstream tasks, a natural question arises: can the performance of these models be further improved? To this end, we propose WeSTAR, a parameter-efficient framework that performs Weakly supervised Self-Training Adaptation with Regularization, designed to enhance the robustness of MDE foundation models in unseen and diverse domains. We first adopt a dense self-training objective as the primary source of structural self-supervision. To further improve robustness, we introduce semantically-aware hierarchical normalization, which exploits instance-level segmentation maps to perform more stable and multi-scale structural normalization. Beyond dense supervision, we introduce a cost-efficient weak supervision in the form of pairwise ordinal depth annotations to further guide the adaptation process, which enforces informative ordinal constraints to mitigate local topological errors. Finally, a weight regularization loss is employed to anchor the LoRA updates, ensuring training stability and preserving the model's generalizable knowledge. Extensive experiments on both realistic and corrupted out-of-distribution datasets under diverse and challenging scenarios demonstrate that WeSTAR consistently improves generalization and achieves state-of-the-art performance across a wide range of benchmarks.
Authors: Jonathan Yaffe, Ben Maman, Meinard M\"uller, Amit H. Bermano
Abstract: Automatic Music Transcription (AMT) converts audio recordings into symbolic musical representations. Training deep neural networks (DNNs) for AMT typically requires strongly aligned training pairs with precise frame-level annotations. Since creating such datasets is costly and impractical for many musical contexts, weakly aligned approaches using segment-level annotations have gained traction. However, existing methods often rely on Dynamic Time Warping (DTW) or soft alignment loss functions, both of which still require local semantic correspondences, making them error-prone and computationally expensive. In this article, we introduce CountEM, a novel AMT framework that eliminates the need for explicit local alignment by leveraging note event histograms as supervision, enabling lighter computations and greater flexibility. Using an Expectation-Maximization (EM) approach, CountEM iteratively refines predictions based solely on note occurrence counts, significantly reducing annotation efforts while maintaining high transcription accuracy. Experiments on piano, guitar, and multi-instrument datasets demonstrate that CountEM matches or surpasses existing weakly supervised methods, improving AMT's robustness, scalability, and efficiency. Our project page is available at https://yoni-yaffe.github.io/count-the-notes.
Authors: Zhenchuan Ma, Qizhi Teng, Pengcheng Yan, Lindong Li, Kirill M. Gerke, Marina V. Karsanina, Xiaohai He
Abstract: Heterogeneous porous materials play a crucial role in various engineering systems. Microstructure characterization and reconstruction provide effective means for modeling these materials, which are critical for conducting physical property simulations, structure-property linkage studies, and enhancing their performance across different applications. To achieve superior controllability and applicability with small sample sizes, we propose a statistically controllable microstructure reconstruction framework that integrates neural networks with sliced-Wasserstein metric. Specifically, our approach leverages local pattern distribution for microstructure characterization and employs a controlled sampling strategy to generate target distributions that satisfy given conditional parameters. A neural network-based model establishes the mapping from the input distribution to the target local pattern distribution, enabling microstructure reconstruction. Combinations of sliced-Wasserstein metric and gradient optimization techniques minimize the distance between these distributions, leading to a stable and reliable model. Our method can perform stochastic and controllable reconstruction tasks even with small sample sizes. Additionally, it can generate large-size (e.g. 512 and 1024) 3D microstructures using a chunking strategy. By introducing spatial location masks, our method excels at generating spatially heterogeneous and complex microstructures. We conducted experiments on stochastic reconstruction, controllable reconstruction, heterogeneous reconstruction, and large-size microstructure reconstruction across various materials. Comparative analysis through visualization, statistical measures, and physical property simulations demonstrates the effectiveness, providing new insights and possibilities for research on structure-property linkage and material inverse design.
Authors: Zi-Chen Xi, Jiahui Huang, Hao-Xiang Chen, Francis Williams, Qun-Ce Xu, Tai-Jiang Mu, Shi-Min Hu
Abstract: We proposed a generalized method, NeuralSSD, for reconstructing a 3D implicit surface from the widely-available point cloud data. NeuralSSD is a solver-based on the neural Galerkin method, aimed at reconstructing higher-quality and accurate surfaces from input point clouds. Implicit method is preferred due to its ability to accurately represent shapes and its robustness in handling topological changes. However, existing parameterizations of implicit fields lack explicit mechanisms to ensure a tight fit between the surface and input data. To address this, we propose a novel energy equation that balances the reliability of point cloud information. Additionally, we introduce a new convolutional network that learns three-dimensional information to achieve superior optimization results. This approach ensures that the reconstructed surface closely adheres to the raw input points and infers valuable inductive biases from point clouds, resulting in a highly accurate and stable surface reconstruction. NeuralSSD is evaluated on a variety of challenging datasets, including the ShapeNet and Matterport datasets, and achieves state-of-the-art results in terms of both surface reconstruction accuracy and generalizability.
Authors: Marcel Gibier, Rapha\"el Duroselle, Pierre Serrano, Olivier Boeffard, Jean-Fran\c{c}ois Bonastre
Abstract: Recent audio-language models have shown impressive performance across a wide range of audio tasks and are increasingly capable of handling long audio inputs. However, the computing costs in these models heavily depend on sequence length, which can become very large given the nature of audio data. In the vision-language domain, token pruning methods have proven effective in reducing token counts while preserving strong performance on standard benchmarks. In this work, we investigate the relevance and effectiveness of such token selection strategies in the context of audio-language models. We also improve them by proposing a lightweight strategy that takes the time dimension into account. While retaining only a quarter of the initial tokens, our approach results in a relative maximum decrease of 2% in CIDEr on Clotho v2 and a relative maximum decrease of 4% in accuracy on MMAU.
Authors: Mohammad Zbib, Hasan Abed Al Kader Hammoud, Sina Mukalled, Nadine Rizk, Fatima Karnib, Issam Lakkis, Ammar Mohanna, Bernard Ghanem
Abstract: We present AraLingBench: a fully human annotated benchmark for evaluating the Arabic linguistic competence of large language models (LLMs). The benchmark spans five core categories: grammar, morphology, spelling, reading comprehension, and syntax, through 150 expert-designed multiple choice questions that directly assess structural language understanding. Evaluating 35 Arabic and bilingual LLMs reveals that current models demonstrate strong surface level proficiency but struggle with deeper grammatical and syntactic reasoning. AraLingBench highlights a persistent gap between high scores on knowledge-based benchmarks and true linguistic mastery, showing that many models succeed through memorization or pattern recognition rather than authentic comprehension. By isolating and measuring fundamental linguistic skills, AraLingBench provides a diagnostic framework for developing Arabic LLMs. The full evaluation code is publicly available on GitHub.
Authors: Eric Xue, Ruiyi Zhang, Zijun Zhang, Pengtao Xie
Abstract: Transformer models are foundational to natural language processing (NLP) applications, yet remain vulnerable to backdoor attacks introduced through poisoned data, which implant hidden behaviors during training. To strengthen the ability to prevent such compromises, recent research has focused on designing increasingly stealthy attacks to stress-test existing defenses, pairing backdoor behaviors with stylized artifact or token-level perturbation triggers. However, this trend diverts attention from the harder and more realistic case: making the model respond to semantic triggers such as specific names or entities, where a successful backdoor could manipulate outputs tied to real people or events in deployed systems. Motivated by this growing disconnect, we introduce SteganoBackdoor, bringing stealth techniques back into line with practical threat models. Leveraging innocuous properties from natural-language steganography, SteganoBackdoor applies a gradient-guided data optimization process to transform semantic trigger seeds into steganographic carriers that embed a high backdoor payload, remain fluent, and exhibit no representational resemblance to the trigger. Across diverse experimental settings, SteganoBackdoor achieves over 99% attack success at an order-of-magnitude lower data-poisoning rate than prior approaches while maintaining unparalleled evasion against a comprehensive suite of data-level defenses. By revealing this practical and covert attack, SteganoBackdoor highlights an urgent blind spot in current defenses and demands immediate attention to adversarial data defenses and real-world threat modeling.
Authors: Marcel Gibier, Nolwenn Celton, Rapha\"el Duroselle, Pierre Serrano, Olivier Boeffard, Jean-Fran\c{c}ois Bonastre
Abstract: In this report, we describe our submission to Track 5 of the DCASE 2025 Challenge for the task of Audio Question Answering(AQA). Our system leverages the SSL backbone BEATs to extract frame-level audio features, which are then processed by a classification head to generate segment-level predictions of acoustic events, following the Audioset ontology. These segment-level predictions are subsequently calibrated before producing event-level predictions. Finally, these predictions are incorporated into a structured prompt, along with the question and candidate answers. This prompt is then fed to a fine-tuned version of Qwen2.5-7B-Instruct, trained using the GRPO algorithm with a simple reward function. Our method achieves an accuracy of 62.6 % on the development set, demonstrating the effectiveness of combining acoustic event reasoning with instruction-tuned large language models for AQA.
Authors: Alessio Pellegrino, Jacopo Mauro
Abstract: One of the long-standing goals in optimisation and constraint programming is to describe a problem in natural language and automatically obtain an executable, efficient model. Large language models appear to bring this vision closer, showing impressive results in automatically generating models for classical benchmarks. However, much of this apparent success may derive from data contamination rather than genuine reasoning: many standard CP problems are likely included in the training data of these models. To examine this hypothesis, we systematically rephrased and perturbed a set of well-known CSPLib problems to preserve their structure while modifying their context and introducing misleading elements. We then compared the models produced by three representative LLMs across original and modified descriptions. Our qualitative analysis shows that while LLMs can produce syntactically valid and semantically plausible models, their performance drops sharply under contextual and linguistic variation, revealing shallow understanding and sensitivity to wording.
Authors: Rishi Gupta, Mukilan Karuppasamy, Shyam Marjit, Aditay Tripathi, Anirban Chakraborty
Abstract: While Large Vision Language Models (LVLMs) are increasingly deployed in real-world applications, their ability to interpret abstract visual inputs remains limited. Specifically, they struggle to comprehend hand-drawn sketches, a modality that offers an intuitive means of expressing concepts that are difficult to describe textually. We identify the primary bottleneck as the absence of a large-scale dataset that jointly models sketches, photorealistic images, and corresponding natural language instructions. To address this, we present two key contributions: (1) a new, large-scale dataset of image-sketch-instruction triplets designed to facilitate both pretraining and instruction tuning, and (2) O3SLM, an LVLM trained on this dataset. Comprehensive evaluations on multiple sketch-based tasks: (a) object localization, (b) counting, (c) image retrieval i.e., (SBIR and fine-grained SBIR), and (d) visual question answering (VQA); while incorporating the three existing sketch datasets, namely QuickDraw!, Sketchy, and Tu Berlin, along with our generated SketchVCL dataset, show that O3SLM achieves state-of-the-art performance, substantially outperforming existing LVLMs in sketch comprehension and reasoning.
Authors: Zhengchunmin Dai, Jiaxiong Tang, Peng Sun, Honglong Chen, Liantao Wu
Abstract: In decentralized machine learning paradigms such as Split Federated Learning (SFL) and its variant U-shaped SFL, the server's capabilities are severely restricted. Although this enhances client-side privacy, it also leaves the server highly vulnerable to model theft by malicious clients. Ensuring intellectual property protection for such capability-limited servers presents a dual challenge: watermarking schemes that depend on client cooperation are unreliable in adversarial settings, whereas traditional server-side watermarking schemes are technically infeasible because the server lacks access to critical elements such as model parameters or labels. To address this challenge, this paper proposes Sigil, a mandatory watermarking framework designed specifically for capability-limited servers. Sigil defines the watermark as a statistical constraint on the server-visible activation space and embeds the watermark into the client model via gradient injection, without requiring any knowledge of the data. Besides, we design an adaptive gradient clipping mechanism to ensure that our watermarking process remains both mandatory and stealthy, effectively countering existing gradient anomaly detection methods and a specifically designed adaptive subspace removal attack. Extensive experiments on multiple datasets and models demonstrate Sigil's fidelity, robustness, and stealthiness.
Authors: Rickmer Krohn, Vignesh Prasad, Gabriele Tiboni, Georgia Chalvatzaki
Abstract: Effective contact-rich manipulation requires robots to synergistically leverage vision, force, and proprioception. However, Reinforcement Learning agents struggle to learn in such multisensory settings, especially amidst sensory noise and dynamic changes. We propose MultiSensory Dynamic Pretraining (MSDP), a novel framework for learning expressive multisensory representations tailored for task-oriented policy learning. MSDP is based on masked autoencoding and trains a transformer-based encoder by reconstructing multisensory observations from only a subset of sensor embeddings, leading to cross-modal prediction and sensor fusion. For downstream policy learning, we introduce a novel asymmetric architecture, where a cross-attention mechanism allows the critic to extract dynamic, task-specific features from the frozen embeddings, while the actor receives a stable pooled representation to guide its actions. Our method demonstrates accelerated learning and robust performance under diverse perturbations, including sensor noise, and changes in object dynamics. Evaluations in multiple challenging, contact-rich robot manipulation tasks in simulation and the real world showcase the effectiveness of MSDP. Our approach exhibits strong robustness to perturbations and achieves high success rates on the real robot with as few as 6,000 online interactions, offering a simple yet powerful solution for complex multisensory robotic control.
Authors: Daniel Klippert, Alexander Marx
Abstract: To distinguish Markov equivalent graphs in causal discovery, it is necessary to restrict the structural causal model. Crucially, we need to be able to distinguish cause $X$ from effect $Y$ in bivariate models, that is, distinguish the two graphs $X \to Y$ and $Y \to X$. Location-scale noise models (LSNMs), in which the effect $Y$ is modeled based on the cause $X$ as $Y = f(X) + g(X)N$, form a flexible class of models that is general and identifiable in most cases. Estimating these models for arbitrary noise terms $N$, however, is challenging. Therefore, practical estimators are typically restricted to symmetric distributions, such as the normal distribution. As we showcase in this paper, when $N$ is a skewed random variable, which is likely in real-world domains, the reliability of these approaches decreases. To approach this limitation, we propose SkewD, a likelihood-based algorithm for bivariate causal discovery under LSNMs with skewed noise distributions. SkewD extends the usual normal-distribution framework to the skew-normal setting, enabling reliable inference under symmetric and skewed noise. For parameter estimation, we employ a combination of a heuristic search and an expectation conditional maximization algorithm. We evaluate SkewD on novel synthetically generated datasets with skewed noise as well as established benchmark datasets. Throughout our experiments, SkewD exhibits a strong performance and, in comparison to prior work, remains robust under high skewness.
Authors: Trishala Jayesh Ahalpara
Abstract: We present Tell Me, a mental well-being system that leverages advances in large language models to provide accessible, context-aware support for users and researchers. The system integrates three components: (i) a retrieval-augmented generation (RAG) assistant for personalized, knowledge-grounded dialogue; (ii) a synthetic client-therapist dialogue generator conditioned on client profiles to facilitate research on therapeutic language and data augmentation; and (iii) a Well-being AI crew, implemented with CrewAI, that produces weekly self-care plans and guided meditation audio. The system is designed as a reflective space for emotional processing rather than a substitute for professional therapy. It illustrates how conversational assistants can lower barriers to support, complement existing care, and broaden access to mental health resources. To address the shortage of confidential therapeutic data, we introduce synthetic client-therapist dialogue generation conditioned on client profiles. Finally, the planner demonstrates an innovative agentic workflow for dynamically adaptive, personalized self-care, bridging the limitations of static well-being tools. We describe the architecture, demonstrate its functionalities, and report evaluation of the RAG assistant in curated well-being scenarios using both automatic LLM-based judgments and a human-user study. This work highlights opportunities for interdisciplinary collaboration between NLP researchers and mental health professionals to advance responsible innovation in human-AI interaction for well-being.
Authors: Tim Schwabe, Maribel Acosta
Abstract: Join ordering is the NP-hard problem of selecting the most efficient sequence in which to evaluate joins (conjunctive, binary operators) in a database query. As the performance of query execution critically depends on this choice, join ordering lies at the core of query optimization. Traditional approaches cast this problem as a discrete combinatorial search over binary trees guided by a cost model, but they often suffer from high computational complexity and limited scalability. We show that, when the cost model is differentiable, the query plans can be continuously relaxed into a soft adjacency matrix representing a superposition of plans. This continuous relaxation, together with a Gumbel-Softmax parameterization of the adjacency matrix and differentiable constraints enforcing plan validity, enables gradient-based search for plans within this relaxed space. Using a learned Graph Neural Network as the cost model, we demonstrate that this gradient-based approach can find comparable and even lower-cost plans compared to traditional discrete local search methods on two different graph datasets. Furthermore, we empirically show that the runtime of this approach scales linearly with query size, in contrast to quadratic or exponential runtimes of classical approaches. We believe this first step towards gradient-based join ordering can lead to more effective and efficient query optimizers in the future.
Authors: Abdurakhmon Sadiev, Yury Demidovich, Igor Sokolov, Grigory Malinovsky, Sarit Khirirat, Peter Richt\'arik
Abstract: Communication compression is essential for scalable distributed training of modern machine learning models, but it often degrades convergence due to the noise it introduces. Error Feedback (EF) mechanisms are widely adopted to mitigate this issue of distributed compression algorithms. Despite their popularity and training efficiency, existing distributed EF algorithms often require prior knowledge of problem parameters (e.g., smoothness constants) to fine-tune stepsizes. This limits their practical applicability especially in large-scale neural network training. In this paper, we study normalized error feedback algorithms that combine EF with normalized updates, various momentum variants, and parameter-agnostic, time-varying stepsizes, thus eliminating the need for problem-dependent tuning. We analyze the convergence of these algorithms for minimizing smooth functions, and establish parameter-agnostic complexity bounds that are close to the best-known bounds with carefully-tuned problem-dependent stepsizes. Specifically, we show that normalized EF21 achieve the convergence rate of near ${O}(1/T^{1/4})$ for Polyak's heavy-ball momentum, ${O}(1/T^{2/7})$ for Iterative Gradient Transport (IGT), and ${O}(1/T^{1/3})$ for STORM and Hessian-corrected momentum. Our results hold with decreasing stepsizes and small mini-batches. Finally, our empirical experiments confirm our theoretical insights.
Authors: Xiangchen Yin, Jiahui Yuan, Zhangchi Hu, Wenzhang Sun, Jie Chen, Xiaozhen Qiao, Hao Li, Xiaoyan Sun
Abstract: Existing video Variational Autoencoders (VAEs) generally overlook the similarity between frame contents, leading to redundant latent modeling. In this paper, we propose decoupled VAE (DeCo-VAE) to achieve compact latent representation. Instead of encoding RGB pixels directly, we decompose video content into distinct components via explicit decoupling: keyframe, motion and residual, and learn dedicated latent representation for each. To avoid cross-component interference, we design dedicated encoders for each decoupled component and adopt a shared 3D decoder to maintain spatiotemporal consistency during reconstruction. We further utilize a decoupled adaptation strategy that freezes partial encoders while training the others sequentially, ensuring stable training and accurate learning of both static and dynamic features. Extensive quantitative and qualitative experiments demonstrate that DeCo-VAE achieves superior video reconstruction performance.
Authors: Haorui Ma, Dennis Frauen, Stefan Feuerriegel
Abstract: Structural nested mean models (SNMMs) are a principled approach to estimate the treatment effects over time. A particular strength of SNMMs is to break the joint effect of treatment sequences over time into localized, time-specific ``blip effects''. This decomposition promotes interpretability through the incremental effects and enables the efficient offline evaluation of optimal treatment policies without re-computation. However, neural frameworks for SNMMs are lacking, as their inherently sequential g-estimation scheme prevents end-to-end, gradient-based training. Here, we propose DeepBlip, the first neural framework for SNMMs, which overcomes this limitation with a novel double optimization trick to enable simultaneous learning of all blip functions. Our DeepBlip seamlessly integrates sequential neural networks like LSTMs or transformers to capture complex temporal dependencies. By design, our method correctly adjusts for time-varying confounding to produce unbiased estimates, and its Neyman-orthogonal loss function ensures robustness to nuisance model misspecification. Finally, we evaluate our DeepBlip across various clinical datasets, where it achieves state-of-the-art performance.
Authors: Mohammad Romani
Abstract: Deepfakes generated by advanced GANs and autoencoders severely threaten information integrity and societal stability. Single-stream CNNs fail to capture multi-scale forgery artifacts across spatial, texture, and frequency domains, limiting robustness and generalization. We introduce the ForensicFlow, a tri-modal forensic framework that synergistically fuses RGB, texture, and frequency evidence for video Deepfake detection. The RGB branch (ConvNeXt-tiny) extracts global visual inconsistencies; the texture branch (Swin Transformer-tiny) detects fine-grained blending artifacts; the frequency branch (CNN + SE) identifies periodic spectral noise. Attention-based temporal pooling dynamically prioritizes high-evidence frames, while adaptive attention fusion balances branch contributions.Trained on Celeb-DF (v2) with Focal Loss, ForensicFlow achieves AUC 0.9752, F1-Score 0.9408, and accuracy 0.9208, outperforming single-stream baselines. Ablation validates branch synergy; Grad-CAM confirms forensic focus. This comprehensive feature fusion provides superior resilience against subtle forgeries.
Authors: Xinzhe Zheng, Shiyu Jiang, Gustavo Seabra, Chenglong Li, Yanjun Li
Abstract: Deep generative models are rapidly advancing structure-based drug design, offering substantial promise for generating small molecule ligands that bind to specific protein targets. However, most current approaches assume a rigid protein binding pocket, neglecting the intrinsic flexibility of proteins and the conformational rearrangements induced by ligand binding, limiting their applicability in practical drug discovery. Here, we propose Apo2Mol, a diffusion-based generative framework for 3D molecule design that explicitly accounts for conformational flexibility in protein binding pockets. To support this, we curate a dataset of over 24,000 experimentally resolved apo-holo structure pairs from the Protein Data Bank, enabling the characterization of protein structure changes associated with ligand binding. Apo2Mol employs a full-atom hierarchical graph-based diffusion model that simultaneously generates 3D ligand molecules and their corresponding holo pocket conformations from input apo states. Empirical studies demonstrate that Apo2Mol can achieve state-of-the-art performance in generating high-affinity ligands and accurately capture realistic protein pocket conformational changes.
Authors: Hugo Frezat, Thomas Gastine, Alexandre Fournier
Abstract: The use of machine learning to represent subgrid-scale (SGS) dynamics is now well established in weather forecasting and climate modelling. Recent advances have demonstrated that SGS models trained via ``online'' end-to-end learning -- where the dynamical solver operating on the filtered equations participates in the training -- can outperform traditional physics-based approaches. Most studies, however, have focused on idealised periodic domains, neglecting the mechanical boundaries present e.g. in planetary interiors. To address this issue, we consider two-dimensional quasi-geostrophic turbulent flow in an axisymmetric bounded domain that we model using a pseudo-spectral differentiable solver, thereby enabling online learning. We examine three configurations, varying the geometry (between an exponential container and a spherical shell) and the rotation rate. Flow is driven by a prescribed analytical forcing, allowing for precise control over the energy injection scale and an exact estimate of the power input. We evaluate the accuracy of the online-trained SGS model against the reference direct numerical simulation using integral quantities and spectral diagnostics. In all configurations, we show that an SGS model trained on data spanning only one turnover time remains stable and accurate over integrations at least a hundred times longer than the training period. Moreover, we demonstrate the model's remarkable ability to reproduce slow processes occurring on time scales far exceeding the training duration, such as the inward drift of jets in the spherical shell. These results suggest a promising path towards developing SGS models for planetary and stellar interior dynamics, including dynamo processes.
Authors: Yilu Fang, Jordan G. Nestor, Casey N. Ta, Jerard Z. Kneifati-Hayek, Chunhua Weng
Abstract: Patients with acute kidney injury (AKI) are at high risk of developing chronic kidney disease (CKD), but identifying those at greatest risk remains challenging. We used electronic health record (EHR) data to dynamically track AKI patients' clinical evolution and characterize AKI-to-CKD progression. Post-AKI clinical states were identified by clustering patient vectors derived from longitudinal medical codes and creatinine measurements. Transition probabilities between states and progression to CKD were estimated using multi-state modeling. After identifying common post-AKI trajectories, CKD risk factors in AKI subpopulations were identified through survival analysis. Of 20,699 patients with AKI at admission, 3,491 (17%) developed CKD. We identified fifteen distinct post-AKI states, each with different probabilities of CKD development. Most patients (75%, n=15,607) remained in a single state or made only one transition during the study period. Both established (e.g., AKI severity, diabetes, hypertension, heart failure, liver disease) and novel CKD risk factors, with their impact varying across these clinical states. This study demonstrates a data-driven approach for identifying high-risk AKI patients, supporting the development of decision-support tools for early CKD detection and intervention.
Authors: Shreya Adrita Banik, Niaz Nafi Rahman, Tahsina Moiukh, Farig Sadeque
Abstract: Detecting political bias in news media is a complex task that requires interpreting subtle linguistic and contextual cues. Although recent advances in Natural Language Processing (NLP) have enabled automatic bias classification, the extent to which large language models (LLMs) align with human judgment still remains relatively underexplored and not yet well understood. This study aims to present a comparative framework for evaluating the detection of political bias across human annotations and multiple LLMs, including GPT, BERT, RoBERTa, and FLAN. We construct a manually annotated dataset of news articles and assess annotation consistency, bias polarity, and inter-model agreement to quantify divergence between human and model perceptions of bias. Experimental results show that among traditional transformer-based models, RoBERTa achieves the highest alignment with human labels, whereas generative models such as GPT demonstrate the strongest overall agreement with human annotations in a zero-shot setting. Among all transformer-based baselines, our fine-tuned RoBERTa model acquired the highest accuracy and the strongest alignment with human-annotated labels. Our findings highlight systematic differences in how humans and LLMs perceive political slant, underscoring the need for hybrid evaluation frameworks that combine human interpretability with model scalability in automated media bias detection.
Authors: Ruoyu Qin, Weiran He, Weixiao Huang, Yangkun Zhang, Yikai Zhao, Bo Pang, Xinran Xu, Yingdi Shan, Yongwei Wu, Mingxing Zhang
Abstract: Reinforcement Learning (RL) has become critical for advancing modern Large Language Models (LLMs), yet existing synchronous RL systems face severe performance bottlenecks. The rollout phase, which dominates end-to-end iteration time, suffers from substantial long-tail latency and poor resource utilization due to inherent workload imbalance. We present Seer, a novel online context learning system that addresses these challenges by exploiting previously overlooked similarities in output lengths and generation patterns among requests sharing the same prompt. Seer introduces three key techniques: divided rollout for dynamic load balancing, context-aware scheduling, and adaptive grouped speculative decoding. Together, these mechanisms substantially reduce long-tail latency and improve resource efficiency during rollout. Evaluations on production-grade RL workloads demonstrate that Seer improves end-to-end rollout throughput by 74% to 97% and reduces long-tail latency by 75% to 93% compared to state-of-the-art synchronous RL systems, significantly accelerating RL training iterations.
Authors: Avi Bagchi, Dwight Hutchenson
Abstract: Radio spectrum monitoring in contested environments motivates the need for reliable automatic signal classification technology. Prior work highlights deep learning as a promising approach, but existing models depend on brute-force Doppler augmentation to achieve real-world generalization, which undermines both training efficiency and interpretability. In this paper, we propose a convolutional neural network (CNN) architecture with complex-valued layers that exploits convolutional shift equivariance in the frequency domain. To establish provable frequency bin shift invariance, we use adaptive polyphase sampling (APS) as pooling layers followed by a global average pooling layer at the end of the network. Using a synthetic dataset of common interference signals, experimental results demonstrate that unlike a vanilla CNN, our model maintains consistent classification accuracy with and without random Doppler shifts despite being trained on no Doppler-shifted examples. Overall, our method establishes an invariance-driven framework for signal classification that offers provable robustness against real-world effects.
Authors: Jan Naumann
Abstract: Recently developed applications in the field of machine learning and computational physics rely on automatic differentiation techniques, that require stable and efficient linear algebra gradient computations. This technical note provides a comprehensive and detailed discussion of the derivative of the truncated singular and eigenvalue decomposition. It summarizes previous work and builds on them with an extensive description of how to derive the relevant terms. A main focus is correctly expressing the derivative in terms of the truncated part, despite lacking knowledge of the full decomposition.
Authors: Rui Zhu, Xiaopu Zhou, Haixu Tang, Stephen W. Scherer, Lucila Ohno-Machado
Abstract: Trained on massive cross-species DNA corpora, DNA large language models (LLMs) learn the fundamental "grammar" and evolutionary patterns of genomic sequences. This makes them powerful priors for DNA sequence modeling, particularly over long ranges. However, two major constraints hinder their use in practice: the quadratic computational cost of self-attention and the growing memory required for key-value (KV) caches during autoregressive decoding. These constraints force the use of heuristics such as fixed-window truncation or sliding windows, which compromise fidelity on ultra-long sequences by discarding distant information. We introduce FOCUS (Feature-Oriented Compression for Ultra-long Self-attention), a progressive context-compression module that can be plugged into pretrained DNA LLMs. FOCUS combines the established k-mer representation in genomics with learnable hierarchical compression: it inserts summary tokens at k-mer granularity and progressively compresses attention key and value activations across multiple Transformer layers, retaining only the summary KV states across windows while discarding ordinary-token KV. A shared-boundary windowing scheme yields a stationary cross-window interface that propagates long-range information with minimal loss. We validate FOCUS on an Evo-2-based DNA LLM fine-tuned on GRCh38 chromosome 1 with self-supervised training and randomized compression schedules to promote robustness across compression ratios. On held-out human chromosomes, FOCUS achieves near-lossless fidelity: compressing a 1 kb context into only 10 summary tokens (about 100x) shifts the average per-nucleotide probability by only about 0.0004. Compared to a baseline without compression, FOCUS reduces KV-cache memory and converts effective inference scaling from O(N^2) to near-linear O(N), enabling about 100x longer inference windows on commodity GPUs with near-lossless fidelity.
Authors: Sriram Srinivasan, Srinivasan Aruchamy, Siva Ram Krisha Vadali
Abstract: Seismic sensing has emerged as a promising solution for border surveillance and monitoring; the seismic sensors that are often buried underground are small and cannot be noticed easily, making them difficult for intruders to detect, avoid, or vandalize. This significantly enhances their effectiveness compared to highly visible cameras or fences. However, accurately detecting and distinguishing between overlapping activities that are happening simultaneously, such as human intrusions, animal movements, and vehicle rumbling, remains a major challenge due to the complex and noisy nature of seismic signals. Correctly identifying simultaneous activities is critical because failing to separate them can lead to misclassification, missed detections, and an incomplete understanding of the situation, thereby reducing the reliability of surveillance systems. To tackle this problem, we propose HyMAD (Hybrid Multi-Activity Detection), a deep neural architecture based on spatio-temporal feature fusion. The framework integrates spectral features extracted with SincNet and temporal dependencies modeled by a recurrent neural network (RNN). In addition, HyMAD employs self-attention layers to strengthen intra-modal representations and a cross-modal fusion module to achieve robust multi-label classification of seismic events. e evaluate our approach on a dataset constructed from real-world field recordings collected in the context of border surveillance and monitoring, demonstrating its ability to generalize to complex, simultaneous activity scenarios involving humans, animals, and vehicles. Our method achieves competitive performance and offers a modular framework for extending seismic-based activity recognition in real-world security applications.
Authors: Zonghao Chen, Atsushi Nitanda, Arthur Gretton, Taiji Suzuki
Abstract: We establish the first global convergence result of neural networks for two stage least squares (2SLS) approach in nonparametric instrumental variable regression (NPIV). This is achieved by adopting a lifted perspective through mean-field Langevin dynamics (MFLD), unlike standard MFLD, however, our setting of 2SLS entails a \emph{bilevel} optimization problem in the space of probability measures. To address this challenge, we leverage the penalty gradient approach recently developed for bilevel optimization which formulates bilevel optimization as a Lagrangian problem. This leads to a novel fully first-order algorithm, termed \texttt{F$^2$BMLD}. Apart from the convergence bound, we further provide a generalization bound, revealing an inherent trade-off in the choice of the Lagrange multiplier between optimization and statistical guarantees. Finally, we empirically validate the effectiveness of the proposed method on an offline reinforcement learning benchmark.
Authors: Albert Lin, Alessandro Pinto, Somil Bansal
Abstract: As perception-based controllers for autonomous systems become increasingly popular in the real world, it is important that we can formally verify their safety and performance despite perceptual uncertainty. Unfortunately, the verification of such systems remains challenging, largely due to the complexity of the controllers, which are often nonlinear, nonconvex, learning-based, and/or black-box. Prior works propose verification algorithms that are based on approximate reachability methods, but they often restrict the class of controllers and systems that can be handled or result in overly conservative analyses. Hamilton-Jacobi (HJ) reachability analysis is a popular formal verification tool for general nonlinear systems that can compute optimal reachable sets under worst-case system uncertainties; however, its application to perception-based systems is currently underexplored. In this work, we propose RoVer-CoRe, a framework for the Robust Verification of Controllers via HJ Reachability. To the best of our knowledge, RoVer-CoRe is the first HJ reachability-based framework for the verification of perception-based systems under perceptual uncertainty. Our key insight is to concatenate the system controller, observation function, and the state estimation modules to obtain an equivalent closed-loop system that is readily compatible with existing reachability frameworks. Within RoVer-CoRe, we propose novel methods for formal safety verification and robust controller design. We demonstrate the efficacy of the framework in case studies involving aircraft taxiing and NN-based rover navigation. Code is available at the link in the footnote.
Authors: Keya Hu, Ali Cy, Linlu Qiu, Xiaoman Delores Ding, Runqian Wang, Yeyin Eva Zhu, Jacob Andreas, Kaiming He
Abstract: The Abstraction and Reasoning Corpus (ARC) is designed to promote research on abstract reasoning, a fundamental aspect of human intelligence. Common approaches to ARC treat it as a language-oriented problem, addressed by large language models (LLMs) or recurrent reasoning models. However, although the puzzle-like tasks in ARC are inherently visual, existing research has rarely approached the problem from a vision-centric perspective. In this work, we formulate ARC within a vision paradigm, framing it as an image-to-image translation problem. To incorporate visual priors, we represent the inputs on a "canvas" that can be processed like natural images. It is then natural for us to apply standard vision architectures, such as a vanilla Vision Transformer (ViT), to perform image-to-image mapping. Our model is trained from scratch solely on ARC data and generalizes to unseen tasks through test-time training. Our framework, termed Vision ARC (VARC), achieves 60.4% accuracy on the ARC-1 benchmark, substantially outperforming existing methods that are also trained from scratch. Our results are competitive with those of leading LLMs and close the gap to average human performance.
Authors: Erkao Bao, Jingcheng Lu, Linqi Song, Nathan Hart-Hodgson, William Parson, Yanheng Zhou
Abstract: Equivariant neural networks are a class of neural networks designed to preserve symmetries inherent in the data. In this paper, we introduce a general method for modifying a neural network to enforce equivariance, a process we refer to as equivarification. We further show that group convolutional neural networks (G-CNNs) arise as a special case of our framework.
Authors: Kenta Oono, Nontawat Charoenphakdee, Kotatsu Bito, Zhengyan Gao, Hideyoshi Igata, Masashi Yoshikawa, Yoshiaki Ota, Hiroki Okui, Kei Akita, Shoichiro Yamaguchi, Yohei Sugawara, Shin-ichi Maeda, Kunihiko Miyoshi, Yuki Saito, Koki Tsuda, Hiroshi Maruyama, Kohei Hayashi
Abstract: Virtual Human Generative Model (VHGM) is a generative model that approximates the joint probability over more than 2000 human healthcare-related attributes. This paper presents the core algorithm, VHGM-MAE, a masked autoencoder (MAE) tailored for handling high-dimensional, sparse healthcare data. VHGM-MAE tackles four key technical challenges: (1) heterogeneity of healthcare data types, (2) probability distribution modeling, (3) systematic missingness in the training dataset arising from multiple data sources, and (4) the high-dimensional, small-$n$-large-$p$ problem. To address these challenges, VHGM-MAE employs a likelihood-based approach to model distributions with heterogeneous types, a transformer-based MAE to capture complex dependencies among observed and missing attributes, and a novel training scheme that effectively leverages available samples with diverse missingness patterns to mitigate the small-n-large-p problem. Experimental results demonstrate that VHGM-MAE outperforms existing methods in both missing value imputation and synthetic data generation.
Authors: Wenyu Liu, Tianqiang Huang, Pengfei Zhang, Zong Ke, Minghui Min, Puning Zhao
Abstract: Adversarial attacks pose a major challenge to distributed learning systems, prompting the development of numerous robust learning methods. However, most existing approaches suffer from the curse of dimensionality, i.e. the error increases with the number of model parameters. In this paper, we make a progress towards high dimensional problems, under arbitrary number of Byzantine attackers. The cornerstone of our design is a direct high dimensional semi-verified mean estimation method. The idea is to identify a subspace with large variance. The components of the mean value perpendicular to this subspace are estimated using corrupted gradient vectors uploaded from worker machines, while the components within this subspace are estimated using auxiliary dataset. As a result, a combination of large corrupted dataset and small clean dataset yields significantly better performance than using them separately. We then apply this method as the aggregator for distributed learning problems. The theoretical analysis shows that compared with existing solutions, our method gets rid of $\sqrt{d}$ dependence on the dimensionality, and achieves minimax optimal statistical rates. Numerical results validate our theory as well as the effectiveness of the proposed method.
Authors: Shivam Gupta, Aditya Parulekar, Eric Price, Zhiyang Xun
Abstract: Diffusion models have become the most popular approach to deep generative modeling of images, largely due to their empirical performance and reliability. From a theoretical standpoint, a number of recent works have studied the iteration complexity of sampling, assuming access to an accurate diffusion model. In this work, we focus on understanding the sample complexity of training such a model; how many samples are needed to learn an accurate diffusion model using a sufficiently expressive neural network? Prior work showed bounds polynomial in the dimension, desired Total Variation error, and Wasserstein error. We show an exponential improvement in the dependence on Wasserstein error and depth, along with improved dependencies on other relevant parameters.
Authors: Jiashuo Jiang, Yinyu Ye
Abstract: We consider the reinforcement learning problem for the constrained Markov decision process (CMDP), which plays a central role in satisfying safety or resource constraints in sequential learning and decision-making. In this problem, we are given finite resources and a MDP with unknown transition probabilities. At each stage, we take an action, collecting a reward and consuming some resources, all assumed to be unknown and need to be learned over time. In this work, we take the first step towards deriving optimal problem-dependent guarantees for the CMDP problems. We derive a logarithmic regret bound, which translates into a $O(\frac{1}{\Delta\cdot\epsilon}\cdot\log^2(1/\epsilon))$ sample complexity bound, with $\Delta$ being a problem-dependent parameter, yet independent of $\epsilon$. Our sample complexity bound improves upon the state-of-art $O(1/\epsilon^2)$ sample complexity for CMDP problems established in the previous literature, in terms of the dependency on $\epsilon$. To achieve this advance, we develop a new framework for analyzing CMDP problems. To be specific, our algorithm operates in the primal space and we resolve the primal LP for the CMDP problem at each period in an online manner, with adaptive remaining resource capacities. The key elements of our algorithm are: i) a characterization of the instance hardness via LP basis, ii) an eliminating procedure that identifies one optimal basis of the primal LP, and; iii) a resolving procedure that is adaptive to the remaining resources and sticks to the characterized optimal basis.
Authors: Andreas Lutz, Gabriele Steidl, Karsten M\"uller, Wojciech Samek
Abstract: Although deep learning has revolutionized domains such as natural language processing and computer vision, its dependence on centralized datasets raises serious privacy concerns. Federated learning addresses this issue by enabling multiple clients to collaboratively train a global deep learning model without compromising their data privacy. However, the performance of such a model degrades under label skew, where the label distribution differs between clients. To overcome this issue, a novel method called FedEntOpt is proposed. In each round, it selects clients to maximize the entropy of the aggregated label distribution, ensuring that the global model is exposed to data from all available classes. Extensive experiments on multiple benchmark datasets show that the proposed method outperforms several state-of-the-art algorithms by up to 6% in classification accuracy under standard settings regardless of the model size, while achieving gains of over 30% in scenarios with low participation rates and client dropout. In addition, FedEntOpt offers the flexibility to be combined with existing algorithms, enhancing their classification accuracy by more than 40%. Importantly, its performance remains unaffected even when differential privacy is applied.
Authors: Arnav M. Das, Chi Ian Tang, Fahim Kawsar, Mohammad Malekzadeh
Abstract: Sensing human motions through Inertial Measurement Units (IMUs) embedded in personal devices has enabled significant applications in health and wellness. Labeled IMU data is scarce, however, unlabeled or weakly labeled IMU data can be used to model human motions. For video or text modalities, the "pretrain and adapt" approach utilizes large volumes of unlabeled or weakly labeled data to build a strong feature extractor, followed by adaptation to specific tasks using limited labeled data. However, pretraining methods are poorly understood for IMU data, and pipelines are rarely evaluated on out-of-domain tasks. We propose PRIMUS: a method for PRetraining IMU encoderS that uses a novel pretraining objective that is empirically validated based on downstream performance on both in-domain and out-of-domain datasets. The PRIMUS objective effectively enhances downstream performance by combining self-supervision, multimodal, and nearest-neighbor supervision. With fewer than 500 labeled samples per class, PRIMUS improves test accuracy by up to 15%, compared to state-of-the-art baselines. To benefit the broader community, we have open-sourced our code at github.com/nokia-bell-labs/pretrained-imu-encoders.
Authors: Soroush Omranpour, Guillaume Rabusseau, Reihaneh Rabbany
Abstract: Modern datasets are increasingly high-dimensional and multiway, often represented as tensor-valued data with multi-indexed variables. While Transformers excel in sequence modeling and high-dimensional tasks, their direct application to multiway data is computationally prohibitive due to the quadratic cost of dot-product attention and the need to flatten inputs, which disrupts tensor structure and cross-dimensional dependencies. We propose the Higher-Order Transformer (HOT), a novel factorized attention framework that represents multiway attention as sums of Kronecker products or sums of mode-wise attention matrices. HOT efficiently captures dense and sparse relationships across dimensions while preserving tensor structure. Theoretically, HOT retains the expressiveness of full high-order attention and allows complexity control via factorization rank. Experiments on 2D and 3D datasets show that HOT achieves competitive performance in multivariate time series forecasting and image classification, with significantly reduced computational and memory costs. Visualizations of mode-wise attention matrices further reveal interpretable high-order dependencies learned by HOT, demonstrating its versatility for complex multiway data across diverse domains. The implementation of our proposed method is publicly available at https://github.com/s-omranpour/HOT.
Authors: Peter Lalor, Henry Adams, Alex Hagen
Abstract: Machine learning has the potential to improve the speed and reliability of radioisotope identification using gamma spectroscopy. However, meticulously labeling an experimental dataset for training is often prohibitively expensive, while training models purely on synthetic data is risky due to the domain gap between simulated and experimental measurements. In this research, we demonstrate that supervised domain adaptation can substantially improve the performance of radioisotope identification models by transferring knowledge between synthetic and experimental data domains. We consider two domain adaptation scenarios: (1) a simulation-to-simulation adaptation, where we perform multi-label proportion estimation using simulated high-purity germanium detectors, and (2) a simulation-to-experimental adaptation, where we perform multi-class, single-label classification using measured spectra from handheld lanthanum bromide (LaBr) and sodium iodide (NaI) detectors. We begin by pretraining a spectral classifier on synthetic data using a custom transformer-based neural network. After subsequent fine-tuning on just 64 labeled experimental spectra, we achieve a test accuracy of 96% in the sim-to-real scenario with a LaBr detector, far surpassing a synthetic-only baseline model (75%) and a model trained from scratch (80%) on the same 64 spectra. Furthermore, we demonstrate that domain-adapted models learn more human-interpretable features than experiment-only baseline models. Overall, our results highlight the potential for supervised domain adaptation techniques to bridge the sim-to-real gap in radioisotope identification, enabling the development of accurate and explainable classifiers even in real-world scenarios where access to experimental data is limited.
Authors: Chang Zou, Evelyn Zhang, Runlin Guo, Haohang Xu, Conghui He, Xuming Hu, Linfeng Zhang
Abstract: Diffusion Transformers (DiT) have become the dominant methods in image and video generation yet still suffer substantial computational costs. As an effective approach for DiT acceleration, feature caching methods are designed to cache the features of DiT in previous timesteps and reuse them in the next timesteps, allowing us to skip the computation in the next timesteps. Among them, token-wise feature caching has been introduced to perform different caching ratios for different tokens in DiTs, aiming to skip the computation for unimportant tokens while still computing the important ones. In this paper, we propose to carefully check the effectiveness in token-wise feature caching with the following two questions: (1) Is it really necessary to compute the so-called "important" tokens in each step? (2) Are so-called important tokens really important? Surprisingly, this paper gives some counter-intuition answers, demonstrating that consistently computing the selected ``important tokens'' in all steps is not necessary. The selection of the so-called ``important tokens'' is often ineffective, and even sometimes shows inferior performance than random selection. Based on these observations, this paper introduces dual feature caching referred to as DuCa, which performs aggressive caching strategy and conservative caching strategy iteratively and selects the tokens for computing randomly. Extensive experimental results demonstrate the effectiveness of our method in DiT, PixArt, FLUX, and OpenSora, demonstrating significant improvements than the previous token-wise feature caching.
Authors: Dylan Sam, Marc Finzi, J. Zico Kolter
Abstract: As large language models (LLMs) are increasingly relied on in AI systems, predicting when they make mistakes is crucial. While a great deal of work in the field uses internal representations to interpret model behavior, these representations are inaccessible when given solely black-box access through an API. In this paper, we extract features of LLMs in a black-box manner by using follow-up prompts and taking the probabilities of different responses as representations to train reliable predictors of model behavior. We demonstrate that training a linear model on these low-dimensional representations produces reliable and generalizable predictors of model performance at the instance level (e.g., if a particular generation correctly answers a question). Remarkably, these can often outperform white-box linear predictors that operate over a model's hidden state or the full distribution over its vocabulary. In addition, we demonstrate that these extracted features can be used to evaluate more nuanced aspects of a language model's state. For instance, they can be used to distinguish between a clean version of GPT-4o-mini and a version that has been influenced via an adversarial system prompt that answers question-answering tasks incorrectly or introduces bugs into generated code. Furthermore, they can reliably distinguish between different model architectures and sizes, enabling the detection of misrepresented models provided through an API (e.g., identifying if GPT-3.5 is supplied instead of GPT-4o-mini).
Authors: Jonathan Ethier, Mathieu Chateauvert, Ryan G. Dempsey, Alexis Bose
Abstract: Wireless communications rely on path loss modeling, which is most effective when it includes the physical details of the propagation environment. Acquiring this data has historically been challenging, but geographic information systems data is becoming increasingly available with higher resolution and accuracy. Access to such details enables propagation models to more accurately predict coverage and account for interference in wireless deployments. Machine learning-based modeling can significantly support this effort, with feature based approaches allowing for accurate, efficient, and scalable propagation modeling. Building on previous work, we introduce an extended set of features that improves prediction accuracy while, most importantly, proving model generalization through rigorous statistical assessment and the use of test set holdouts.
Authors: Takuro Kutsuna
Abstract: Importance sampling is widely used to improve the efficiency of deep neural network (DNN) training by reducing the variance of gradient estimators. However, efficiently assessing the variance reduction relative to uniform sampling remains challenging due to computational overhead. This paper proposes a method for estimating variance reduction during DNN training using only minibatches sampled under importance sampling. By leveraging the proposed method, the paper also proposes an effective minibatch size to enable automatic learning rate adjustment. An absolute metric to quantify the efficiency of importance sampling is also introduced as well as an algorithm for real-time estimation of importance scores based on moving gradient statistics. Theoretical analysis and experiments on benchmark datasets demonstrated that the proposed algorithm consistently reduces variance, improves training efficiency, and enhances model accuracy compared with current importance-sampling approaches while maintaining minimal computational overhead.
Authors: Robert O'Shea, Bipin Rajendran
Abstract: State-of-the-art methods for backpropagation-free learning employ local error feedback to direct iterative optimisation via gradient descent. In this study, we examine the more restrictive setting where retrograde communication from neuronal outputs is unavailable for pre-synaptic weight optimisation. To address this challenge, we propose Forward Projection (FP). This randomised closed-form training method requires only a single forward pass over the entire dataset for model fitting, without retrograde communication. Our method generates target values for pre-activation membrane potentials at each layer through randomised nonlinear projections of pre-synaptic inputs and the labels, thereby encoding information from both sources. Local loss functions are optimised over pre-synaptic inputs using closed-form regression, without feedback from neuronal outputs or downstream layers. Interpretability is a key advantage of FP training; membrane potentials of hidden neurons in FP-trained networks encode information which are interpretable layer-wise as label predictions. We demonstrate the effectiveness of FP across four biomedical datasets, comparing it with backpropagation and local learning techniques such as Forward-Forward training and Local Supervision in multi-layer perceptron and convolutional architectures. In some few-shot learning tasks, FP yielded more generalisable models than those optimised via backpropagation. In large-sample tasks, FP-based models achieve generalisation comparable to gradient descent-based local learning methods while requiring only a single forward propagation step, achieving significant speed up for training.
Authors: Zheng Chen, Yushi Feng, Jisheng Dang, Yue Deng, Changyang He, Hongxi Pu, Haoxuan Li, Bo Li
Abstract: Large Language Models (LLMs) have attained human-level fluency in text generation, which complicates the distinguishing between human-written and LLM-generated texts. This increases the risk of misuse and highlights the need for reliable detectors. Yet, existing detectors exhibit poor robustness on out-of-distribution (OOD) data and attacked data, which is critical for real-world scenarios. Also, they struggle to provide interpretable evidence to support their decisions, thus undermining the reliability. In light of these challenges, we propose IPAD (Inverse Prompt for AI Detection), a novel framework consisting of a Prompt Inverter that identifies predicted prompts that could have generated the input text, and two Distinguishers that examine the probability that the input texts align with the predicted prompts. Empirical evaluations demonstrate that IPAD outperforms the strongest baselines by 9.05% (Average Recall) on in-distribution data, 12.93% (AUROC) on out-of-distribution data, and 5.48% (AUROC) on attacked data. IPAD also performs robustly on structured datasets. Furthermore, an interpretability assessment is conducted to illustrate that IPAD enhances the AI detection trustworthiness by allowing users to directly examine the decision-making evidence, which provides interpretable support for its state-of-the-art detection results.
Authors: Varun Babbar, Hayden McTavish, Cynthia Rudin, Margo Seltzer
Abstract: Decision tree optimization is fundamental to interpretable machine learning. The most popular approach is to greedily search for the best feature at every decision point, which is fast but provably suboptimal. Recent approaches find the global optimum using branch and bound with dynamic programming, showing substantial improvements in accuracy and sparsity at great cost to scalability. An ideal solution would have the accuracy of an optimal method and the scalability of a greedy method. We introduce a family of algorithms called SPLIT (SParse Lookahead for Interpretable Trees) that moves us significantly forward in achieving this ideal balance. We demonstrate that not all sub-problems need to be solved to optimality to find high quality trees; greediness suffices near the leaves. Since each depth adds an exponential number of possible trees, this change makes our algorithms orders of magnitude faster than existing optimal methods, with negligible loss in performance. We extend this algorithm to allow scalable computation of sets of near-optimal trees (i.e., the Rashomon set).
Authors: Marius F. R. Juston, Ramavarapu S. Sreenivas, William R. Norris, Dustin Nottage, Ahmet Soylemezoglu
Abstract: This paper discusses the weight parametrization of two standard 1-Lipschitz network architectures, the Almost-Orthogonal-Layers (AOL) and the SDP-based Lipschitz Layers (SLL). It examines their impact on initialization for deep 1-Lipschitz feedforward networks, and discusses underlying issues surrounding this initialization. These networks are mainly used in certifiably robust classification applications to combat adversarial attacks by limiting the impact of perturbations on the classification output. Exact and upper bounds for the parameterized weight variance were calculated assuming a standard Normal distribution initialization; additionally, an upper bound was computed assuming a Generalized Normal Distribution, generalizing the proof for Uniform, Laplace, and Normal distribution weight initializations. It is demonstrated that the weight variance holds no bearing on the output variance distribution and that only the dimension of the weight matrices matters. Additionally, this paper demonstrates that the weight initialization always causes deep 1-Lipschitz networks to decay to zero.
Authors: Greg d'Eon, Hala Murad, Kevin Leyton-Brown, James R. Wright
Abstract: Behavioral game theory models serve two purposes: yielding insights into how human decision-making works, and predicting how people would behave in novel strategic settings. A system called GameNet represents the state of the art for predicting human behavior in the setting of unrepeated simultaneous-move games, combining a simple "level-k" model of strategic reasoning with a complex neural network model of non-strategic "level-0" behavior. Although this reliance on well-established ideas from cognitive science ought to make GameNet interpretable, the flexibility of its level-0 model raises the possibility that it is able to emulate strategic reasoning. In this work, we prove that GameNet's level-0 model is indeed too general. We then introduce ElementaryNet, a novel neural network that is provably incapable of expressing strategic behavior. We show that these additional restrictions are empirically harmless, with ElementaryNet and GameNet having statistically indistinguishable performance. We then show how it is possible to derive insights about human behavior by varying ElementaryNet's features and interpreting its parameters, finding evidence of iterative reasoning, learning about the depth of this reasoning process, and showing the value of a rich level-0 specification.
Authors: Haihong Zhao, Zhixun Li, Chenyi Zi, Aochuan Chen, Fugee Tsung, Jia Li, Jeffrey Xu Yu
Abstract: Graph learning plays a vital role in mining and analyzing complex relationships within graph data and has been widely applied to real-world scenarios such as social, citation, and e-commerce networks. Foundation models in computer vision (CV) and natural language processing (NLP) have demonstrated remarkable cross-domain capabilities that are equally significant for graph data. However, existing graph learning approaches often struggle to generalize across domains. Motivated by recent advances in CV and NLP, cross-domain graph learning (CDGL) has gained renewed attention as a promising step toward realizing true graph foundation models. In this survey, we provide a comprehensive review and analysis of existing works on CDGL. We propose a new taxonomy that categorizes existing approaches according to the type of transferable knowledge learned across domains: structure-oriented, feature-oriented, and mixture-oriented. Based on this taxonomy, we systematically summarize representative methods in each category, discuss the key challenges and limitations of current studies, and outline promising directions for future research. A continuously updated collection of related works is available at: https://github.com/cshhzhao/Awesome-Cross-Domain-Graph-Learning.
URLs: https://github.com/cshhzhao/Awesome-Cross-Domain-Graph-Learning.
Authors: Avrim Blum, Steve Hanneke, Chirag Pabbaraju, Donya Saless
Abstract: We consider a model for explainable AI in which an explanation for a prediction $h(x)=y$ consists of a subset $S'$ of the training data (if it exists) such that all classifiers $h' \in H$ that make at most $b$ mistakes on $S'$ predict $h'(x)=y$. Such a set $S'$ serves as a proof that $x$ indeed has label $y$ under the assumption that (1) the target function $h^\star$ belongs to $H$, and (2) the set $S$ contains at most $b$ corrupted points. For example, if $b=0$ and $H$ is the family of linear classifiers in $\mathbb{R}^d$, and if $x$ lies inside the convex hull of the positive data points in $S$ (and hence every consistent linear classifier labels $x$ as positive), then Carath\'eodory's theorem states that $x$ lies inside the convex hull of $d+1$ of those points. So, a set $S'$ of size $d+1$ could be released as an explanation for a positive prediction, and would serve as a short proof of correctness of the prediction under the assumption of realizability. In this work, we consider this problem more generally, for general hypothesis classes $H$ and general values $b\geq 0$. We define the notion of the robust hollow star number of $H$ (which generalizes the standard hollow star number), and show that it precisely characterizes the worst-case size of the smallest certificate achievable, and analyze its size for natural classes. We also consider worst-case distributional bounds on certificate size, as well as distribution-dependent bounds that we show tightly control the sample size needed to get a certificate for any given test example. In particular, we define a notion of the certificate coefficient $\varepsilon_x$ of an example $x$ with respect to a data distribution $D$ and target function $h^\star$, and prove matching upper and lower bounds on sample size as a function of $\varepsilon_x$, $b$, and the VC dimension $d$ of $H$.
Authors: Hanyu Zhang, Zhen Xing, Ruian He, Wenxuan Yang, Chenxi Ma, Weimin Tan, Bo Yan
Abstract: Coreset selection aims to identify a small yet highly informative subset of data, thereby enabling more efficient model training while reducing storage overhead. Recently, this capability has been leveraged to tackle the challenges of fine-tuning large foundation models, offering a direct pathway to their efficient and practical deployment. However, most existing methods are class-agnostic, causing them to overlook significant difficulty variations among classes. This leads them to disproportionately prune samples from either overly easy or hard classes, resulting in a suboptimal allocation of the data budget that ultimately degrades the final coreset performance. To address this limitation, we propose Non-Uniform Class-Wise Coreset Selection (NUCS), a novel framework that both integrates class-level and sample-level difficulty. We propose a robust metric for global class difficulty, quantified as the winsorized average of per-sample difficulty scores. Guided by this metric, our method performs a theoretically-grounded, non-uniform allocation of data selection budgets inter-class, while adaptively selecting samples intra-class with optimal difficulty ranges. Extensive experiments on a wide range of visual classification tasks demonstrate that NUCS consistently outperforms state-of-the-art methods across 10 diverse datasets and pre-trained models, achieving both superior accuracy and computational efficiency, highlighting the promise of non-uniform class-wise selection strategy for advancing the efficient fine-tuning of large foundation models.
Authors: G\'er\^ome Andry, Sacha Lewin, Fran\c{c}ois Rozet, Omer Rochman, Victor Mangeleer, Matthias Pirlet, Elise Faulx, Marilaure Gr\'egoire, Gilles Louppe
Abstract: Deep learning has advanced weather forecasting, but accurate predictions first require identifying the current state of the atmosphere from observational data. In this work, we introduce Appa, a score-based data assimilation model generating global atmospheric trajectories at 0.25\si{\degree} resolution and 1-hour intervals. Powered by a 565M-parameter latent diffusion model trained on ERA5, Appa can be conditioned on arbitrary observations to infer plausible trajectories, without retraining. Our probabilistic framework handles reanalysis, filtering, and forecasting, within a single model, producing physically consistent reconstructions from various inputs. Results establish latent score-based data assimilation as a promising foundation for future global atmospheric modeling systems.
Authors: Jialin Mao, Itay Griniasty, Yan Sun, Mark K. Transtrum, James P. Sethna, Pratik Chaudhari
Abstract: Recent experiments have shown that training trajectories of multiple deep neural networks with different architectures, optimization algorithms, hyper-parameter settings, and regularization methods evolve on a remarkably low-dimensional "hyper-ribbon-like" manifold in the space of probability distributions. Inspired by the similarities in the training trajectories of deep networks and linear networks, we analytically characterize this phenomenon for the latter. We show, using tools in dynamical systems theory, that the geometry of this low-dimensional manifold is controlled by (i) the decay rate of the eigenvalues of the input correlation matrix of the training data, (ii) the relative scale of the ground-truth output to the weights at the beginning of training, and (iii) the number of steps of gradient descent. By analytically computing and bounding the contributions of these quantities, we characterize phase boundaries of the region where hyper-ribbons are to be expected. We also extend our analysis to kernel machines and linear models that are trained with stochastic gradient descent.
Authors: Chris Cundy, Adam Gleave
Abstract: As AI systems become more capable, deceptive behaviors can undermine evaluation and mislead users at deployment. Recent work has shown that lie detectors can accurately classify deceptive behavior, but they are not typically used in the training pipeline due to concerns around contamination and objective hacking. We examine these concerns by incorporating a lie detector into the labelling step of LLM post-training and evaluating whether the learned policy is genuinely more honest, or instead learns to fool the lie detector while remaining deceptive. Using DolusChat, a novel 65k-example dataset with paired truthful/deceptive responses, we identify three key factors that determine the honesty of learned policies: amount of exploration during preference learning, lie detector accuracy, and KL regularization strength. We find that preference learning with lie detectors and GRPO can lead to policies which evade lie detectors, with deception rates of over 85\%. However, if the lie detector true positive rate (TPR) or KL regularization is sufficiently high, GRPO learns honest policies. In contrast, off-policy algorithms (DPO) consistently lead to deception rates under 25\% for realistic TPRs. Our results illustrate a more complex picture than previously assumed: depending on the context, lie-detector-enhanced training can be a powerful tool for scalable oversight, or a counterproductive method encouraging undetectable misalignment.
Authors: Roberto L. Castro, Andrei Panferov, Soroush Tabesh, Oliver Sieberling, Jiale Chen, Mahdi Nikdan, Saleh Ashkboos, Dan Alistarh
Abstract: Training large language models (LLMs) models directly in low-precision offers a way to address computational costs by improving both throughput and energy efficiency. For those purposes, NVIDIA's recent Blackwell architecture facilitates very low-precision operations using FP4 variants. Yet, current algorithms for training LLMs in FP4 precision face significant accuracy degradation and often rely on mixed-precision fallbacks. In this paper, we investigate hardware-supported FP4 training and introduce a new approach for accurate, end-to-end FP4 training with all the major computations (i.e., linear layers) in low precision. Through extensive evaluations on Llama-type models, we reveal a new low-precision scaling law that quantifies performance trade-offs across bit-widths and training setups. Guided by this investigation, we design an "optimal" technique in terms of accuracy-vs-computation, called Quartet. We implement Quartet using optimized CUDA kernels tailored for Blackwell, demonstrating that fully FP4-based training is a competitive alternative to FP16 half-precision and to FP8 training. Our code is available at https://github.com/IST-DASLab/Quartet.
Authors: Fabrizio Boncoraglio, Emanuele Troiani, Vittorio Erba, Lenka Zdeborov\'a
Abstract: We introduce the attention-indexed model (AIM), a theoretical framework for analyzing learning in deep attention layers. Inspired by multi-index models, AIM captures how token-level outputs emerge from layered bilinear interactions over high-dimensional embeddings. Unlike prior tractable attention models, AIM allows full-width key and query matrices, aligning more closely with practical transformers. Using tools from statistical mechanics and random matrix theory, we derive closed-form predictions for Bayes-optimal generalization error and identify sharp phase transitions as a function of sample complexity, model width, and sequence length. We propose a matching approximate message passing algorithm and show that gradient descent can reach optimal performance. AIM offers a solvable playground for understanding learning in self-attention layers, that are key components of modern architectures.
Authors: Woosung Kim, Jinho Lee, Jongmin Lee, Byung-Jun Lee
Abstract: Multi-objective reinforcement learning (MORL) aims to optimize policies in the presence of conflicting objectives, where linear scalarization is commonly used to reduce vector-valued returns into scalar signals. While effective for certain preferences, this approach cannot capture fairness-oriented goals such as Nash social welfare or max-min fairness, which require nonlinear and non-additive trade-offs. Although several online algorithms have been proposed for specific fairness objectives, a unified approach for optimizing nonlinear welfare criteria in the offline setting-where learning must proceed from a fixed dataset-remains unexplored. In this work, we present FairDICE, the first offline MORL framework that directly optimizes nonlinear welfare objective. FairDICE leverages distribution correction estimation to jointly account for welfare maximization and distributional regularization, enabling stable and sample-efficient learning without requiring explicit preference weights or exhaustive weight search. Across multiple offline benchmarks, FairDICE demonstrates strong fairness-aware performance compared to existing baselines.
Authors: Zitong Huang, Mansooreh Montazerin, Ajitesh Srivastava
Abstract: Designing neural networks typically relies on manual trial and error or a neural architecture search (NAS) followed by weight training. The former is time-consuming and labor-intensive, while the latter often discretizes architecture search and weight optimization. In this paper, we propose a fundamentally different approach that simultaneously optimizes both the architecture and the weights of a neural network. Our framework first trains a universal multi-scale autoencoder that embeds both architectural and parametric information into a continuous latent space, where functionally similar neural networks are mapped closer together. Given a dataset, we then randomly initialize a point in the embedding space and update it via gradient descent to obtain the optimal neural network, jointly optimizing its structure and weights. The optimization process incorporates sparsity and compactness penalties to promote efficient models. Experiments on synthetic regression tasks demonstrate that our method effectively discovers sparse and compact neural networks with strong performance.
Authors: Denis Gudovskiy, Wenzhao Zheng, Tomoyuki Okuno, Yohei Nakata, Kurt Keutzer
Abstract: Continuous normalizing flows (CNFs) and diffusion models (DMs) generate high-quality data from a noise distribution. However, their sampling process demands multiple iterations to solve an ordinary differential equation (ODE) with high computational complexity. State-of-the-art methods focus on reducing the number of discrete time steps during sampling to improve efficiency. In this work, we explore a complementary direction in which the quality-complexity tradeoff can also be controlled in terms of the neural network length. We achieve this by rewiring the blocks in the transformer-based architecture to solve an inner discretized ODE w.r.t. its depth. Then, we apply a length consistency term during flow matching training, and as a result, the sampling can be performed with an arbitrary number of time steps and transformer blocks. Unlike others, our ODE$_t$(ODE$_l$) approach is solver-agnostic in time dimension and reduces both latency and, importantly, memory usage. CelebA-HQ and ImageNet generation experiments show a latency reduction of up to $2\times$ in the most efficient sampling mode, and FID improvement of up to $2.8$ points for high-quality sampling when applied to prior methods. We open-source our code and checkpoints at github.com/gudovskiy/odelt.
Authors: Youkang Wang, Jian Wang, Rubing Chen, Xiao-Yong Wei
Abstract: Inference-time scaling has emerged as a powerful technique for enhancing the reasoning performance of Large Language Models (LLMs). However, existing approaches often rely on heuristic strategies for parallel sampling, lacking a principled foundation. To address this gap, we propose a probabilistic framework that formalizes the optimality of inference-time scaling under the assumption that parallel samples are independently and identically distributed (i.i.d.), and where the Best-of-N selection strategy follows a probability distribution that can be estimated. Within this framework, we derive a theoretical lower bound on the required number of samples to achieve a target performance level, providing the first principled guidance for compute-efficient scaling. Leveraging this insight, we develop \textsc{OptScale}, a practical algorithm that dynamically determines the optimal number of sampled responses. \textsc{OptScale} employs a language model-based predictor to estimate probabilistic prior parameters, enabling the decision of the minimal number of samples needed that satisfy predefined performance thresholds and confidence levels. Extensive experiments on representative reasoning benchmarks (including MATH-500, GSM8K, AIME, and AMC) demonstrate that \textsc{OptScale} significantly reduces sampling overhead while remaining better or on par with state-of-the-art reasoning performance. Our work offers both a theoretical foundation and a practical solution for principled inference-time scaling, addressing a critical gap in the efficient deployment of LLMs for complex reasoning. The source code is publicly available at https://github.com/Albertwyk/OptScale.
Authors: F\'elix Lefebvre, Ga\"el Varoquaux
Abstract: Many machine learning tasks can benefit from external knowledge. Large knowledge graphs store such knowledge, and embedding methods can be used to distill it into ready-to-use vector representations for downstream applications. For this purpose, current models have however two limitations: they are primarily optimized for link prediction, via local contrastive learning, and their application to the largest graphs requires significant engineering effort due to GPU memory limits. To address these, we introduce SEPAL: a Scalable Embedding Propagation ALgorithm for large knowledge graphs designed to produce high-quality embeddings for downstream tasks at scale. The key idea of SEPAL is to ensure global embedding consistency by optimizing embeddings only on a small core of entities, and then propagating them to the rest of the graph with message passing. We evaluate SEPAL on 7 large-scale knowledge graphs and 46 downstream machine learning tasks. Our results show that SEPAL significantly outperforms previous methods on downstream tasks. In addition, SEPAL scales up its base embedding model, enabling fitting huge knowledge graphs on commodity hardware.
Authors: Ajesh Koyatan Chathoth, Shuhao Yu, Stephen Lee
Abstract: User-controllable privacy is important in modern sensing systems, as privacy preferences can vary significantly from person to person and may evolve over time. This is especially relevant in devices equipped with Inertial Measurement Unit (IMU) sensors, such as smartphones and wearables, which continuously collect rich time-series data that can inadvertently expose sensitive user behaviors. While prior work has proposed privacy-preserving methods for sensor data, most rely on static, predefined privacy labels or require large quantities of private training data, limiting their adaptability and user agency. In this work, we introduce PrivCLIP, a dynamic, user-controllable, few-shot privacy-preserving sensing framework. PrivCLIP allows users to specify and modify their privacy preferences by categorizing activities as sensitive (black-listed), non-sensitive (white-listed), or neutral (gray-listed). Leveraging a multimodal contrastive learning approach, PrivCLIP aligns IMU sensor data with natural language activity descriptions in a shared embedding space, enabling few-shot detection of sensitive activities. When a privacy-sensitive activity is identified, the system uses a language-guided activity sanitizer and a motion generation module (IMU-GPT) to transform the original data into a privacy-compliant version that semantically resembles a non-sensitive activity. We evaluate PrivCLIP on multiple human activity recognition datasets and demonstrate that it significantly outperforms baseline methods in terms of both privacy protection and data utility.
Authors: Jyotirmai Singh, Samar Khanna, James Burgess
Abstract: Diffusion models typically inject isotropic Gaussian noise, disregarding structure in the data. Motivated by the way quantum squeezed states redistribute uncertainty according to the Heisenberg uncertainty principle, we introduce Squeezed Diffusion Models (SDM), which scale noise anisotropically along the principal component of the training distribution. As squeezing enhances the signal-to-noise ratio in physics, we hypothesize that scaling noise in a data-dependent manner can better assist diffusion models in learning important data features. We study two configurations: (i) a Heisenberg diffusion model that compensates the scaling on the principal axis with inverse scaling on orthogonal directions and (ii) a standard SDM variant that scales only the principal axis. Counterintuitively, on CIFAR-10/100 and CelebA-64, mild antisqueezing - i.e. increasing variance on the principal axis - consistently improves FID by up to 15% and shifts the precision-recall frontier toward higher recall. Our results demonstrate that simple, data-aware noise shaping can deliver robust generative gains without architectural changes.
Authors: Cheol Woo Kim, Shresth Verma, Mauricio Tec, Milind Tambe
Abstract: We study an LLM fine-tuning task for designing reward functions for sequential resource allocation problems in public health, guided by human preferences expressed in natural language. This setting presents a challenging testbed for alignment due to complex and ambiguous objectives and limited data availability. We propose DPO-PRO, a robust fine-tuning algorithm based on Direct Preference Optimization (DPO), which accounts for uncertainty in the preference distribution using a lightweight Distributionally Robust Optimization (DRO) formulation. Unlike prior DRO-based DPO methods, DPO-PRO is significantly less conservative. We evaluate DPO-PRO on a real-world maternal mobile health program operated by the non-profit organization ARMMAN, as well as on standard alignment benchmarks. Experimental results demonstrate that our method consistently improves robustness to noisy preference signals compared to existing DPO variants. Moreover, DPO-PRO achieves comparable performance to prior self-reflection-based baseline for reward function design, while requiring significantly lower inference-time cost.
Authors: Delphine Longuet, Amira Elouazzani, Alejandro Penacho Riveiros, Nicola Bastianello
Abstract: Failures in satellite components are costly and challenging to address, often requiring significant human and material resources. Embedding a hybrid AI-based system for fault detection directly in the satellite can greatly reduce this burden by allowing earlier detection. However, such systems must operate with extremely high reliability. To ensure this level of dependability, we employ the formal verification tool Marabou to verify the local robustness of the neural network models used in the AI-based algorithm. This tool allows us to quantify how much a model's input can be perturbed before its output behavior becomes unstable, thereby improving trustworthiness with respect to its performance under uncertainty.
Authors: Ainhize Barrainkua, Giovanni De Toni, Jose Antonio Lozano, Novi Quadrianto
Abstract: Machine learning based predictions are increasingly used in sensitive decision-making applications that directly affect our lives. This has led to extensive research into ensuring the fairness of classifiers. Beyond just fair classification, emerging legislation now mandates that when a classifier delivers a negative decision, it must also offer actionable steps an individual can take to reverse that outcome. This concept is known as algorithmic recourse. Nevertheless, many researchers have expressed concerns about the fairness guarantees within the recourse process itself. In this work, we provide a holistic theoretical characterization of unfairness in algorithmic recourse, formally linking fairness guarantees in recourse and classification, and highlighting limitations of the standard equal cost paradigm. We then introduce a novel fairness framework based on social burden, along with a practical algorithm (MISOB), broadly applicable under real-world conditions. Empirical results on real-world datasets show that MISOB reduces the social burden across all groups without compromising overall classifier accuracy.
Authors: Matan Avitan, Moran Baruch, Nir Drucker, Itamar Zimerman, Yoav Goldberg
Abstract: Large language models (LLMs) power modern AI applications, but processing sensitive data on untrusted servers raises privacy concerns. Homomorphic encryption (HE) enables computation on encrypted data for secure inference. However, neural text generation requires decoding methods like argmax and sampling, which are non-polynomial and thus computationally expensive under encryption, creating a significant performance bottleneck. We introduce cutmax, an HE-friendly argmax algorithm that reduces ciphertext operations compared to prior methods, enabling practical greedy decoding under encryption. We also propose the first HE-compatible nucleus (top-p) sampling method, leveraging cutmax for efficient stochastic decoding with provable privacy guarantees. Both techniques are polynomial, supporting efficient inference in privacy-preserving settings. Moreover, their differentiability facilitates gradient-based sequence-level optimization as a polynomial alternative to straight-through estimators. We further provide strong theoretical guarantees for cutmax, proving its convergence via exponential amplification of the gap ratio between the maximum and runner-up elements. Evaluations on realistic LLM outputs show latency reductions of 24x-35x over baselines, advancing secure text generation.
Authors: Spencer King, Zhilu Zhang, Ruofan Yu, Baris Coskun, Wei Ding, Qian Cui
Abstract: Anomaly detection is critical in domains such as cybersecurity and finance, especially when working with large-scale tabular data. Yet, unsupervised anomaly detection-where no labeled anomalies are available-remains challenging because traditional deep learning methods model a single global distribution, assuming all samples follow the same behavior. In contrast, real-world data often contain heterogeneous contexts (e.g., different users, accounts, or devices), where globally rare events may be normal within specific conditions. We introduce a contextual learning framework that explicitly models how normal behavior varies across contexts by learning conditional data distributions $P(\mathbf{Y} \mid \mathbf{C})$ rather than a global joint distribution $P(\mathbf{X})$. The framework encompasses (1) a probabilistic formulation for context-conditioned learning, (2) a principled bilevel optimization strategy for automatically selecting informative context features using early validation loss, and (3) theoretical grounding through variance decomposition and discriminative learning principles. We instantiate this framework using a novel conditional Wasserstein autoencoder as a simple yet effective model for tabular anomaly detection. Extensive experiments across eight benchmark datasets demonstrate that contextual learning consistently outperforms global approaches-even when the optimal context is not intuitively obvious-establishing a new foundation for anomaly detection in heterogeneous tabular data.
Authors: Mianchu Wang, Giovanni Montana
Abstract: Retrosynthesis planning aims to decompose target molecules into available building blocks, forming a synthetic tree where each internal node represents an intermediate compound and each leaf ideally corresponds to a purchasable reactant. However, this tree becomes invalid if any leaf node is not a valid building block, making the planning process vulnerable to the "weakest link" in the synthetic route. Existing methods often optimise for average performance across branches, failing to account for this worst-case sensitivity. In this paper, we reframe retrosynthesis as a worst-path optimisation problem within tree-structured Markov Decision Processes (MDPs). We prove that this formulation admits a unique optimal solution and provides monotonic improvement guarantees. Building on this insight, we introduce Interactive Retrosynthesis Planning (InterRetro), a method that interacts with the tree MDP, learns a value function for worst-path outcomes, and improves its policy through self-imitation, preferentially reinforcing past decisions with high estimated advantage. Empirically, InterRetro achieves state-of-the-art results - solving 100% of targets on the Retro*-190 benchmark, shortening synthetic routes by 4.9%, and achieving promising performance using only 10% of the training data.
Authors: Zongxin Shen, Yanyong Huang, Bin Wang, Jinyuan Chang, Shiyu Liu, Tianrui Li
Abstract: Multi-view unsupervised feature selection (MUFS) has recently received increasing attention for its promising ability in dimensionality reduction on multi-view unlabeled data. Existing MUFS methods typically select discriminative features by capturing correlations between features and clustering labels. However, an important yet underexplored question remains: \textit{Are such correlations sufficiently reliable to guide feature selection?} In this paper, we analyze MUFS from a causal perspective by introducing a novel structural causal model, which reveals that existing methods may select irrelevant features because they overlook spurious correlations caused by confounders. Building on this causal perspective, we propose a novel MUFS method called CAusal multi-view Unsupervised feature Selection leArning (CAUSA). Specifically, we first employ a generalized unsupervised spectral regression model that identifies informative features by capturing dependencies between features and consensus clustering labels. We then introduce a causal regularization module that can adaptively separate confounders from multi-view data and simultaneously learn view-shared sample weights to balance confounder distributions, thereby mitigating spurious correlations. Thereafter, integrating both into a unified learning framework enables CAUSA to select causally informative features. Comprehensive experiments demonstrate that CAUSA outperforms several state-of-the-art methods. To our knowledge, this is the first in-depth study of causal multi-view feature selection in the unsupervised setting.
Authors: Kin G. Olivares, Malcolm Wolff, Tatiana Konstantinova, Shankar Ramasubramanian, Boris Oreshkin, Andrew Gordon Wilson, Andres Potapczynski, Willa Potosnak, Michael W. Mahoney, Mengfei Cao, Dmitry Efimov
Abstract: Cross-frequency transfer learning (CFTL) has emerged as a popular framework for curating large-scale time series datasets to pre-train foundation forecasting models (FFMs). Although CFTL has shown promise, current benchmarking practices fall short of accurately assessing its performance. This shortcoming stems from many factors: an over-reliance on small-scale evaluation datasets; inadequate treatment of sample size when computing summary statistics; reporting of suboptimal statistical models; and failing to account for non-negligible risks of overlap between pre-training and test datasets. To address these limitations, we introduce a unified reimplementation of widely-adopted neural forecasting networks, adapting them for the CFTL setup; we pre-train only on proprietary and synthetic data, being careful to prevent test leakage; and we evaluate on 15 large, diverse public forecast competition datasets. Our empirical analysis reveals that statistical models' accuracy is frequently underreported. Notably, we confirm that statistical models and their ensembles consistently outperform existing FFMs by more than 8.2% in sCRPS, and by more than 20% MASE, across datasets. However, we also find that synthetic dataset pre-training does improve the accuracy of a FFM by 7% percent.
Authors: Adam J. Stewart, Caleb Robinson, Arindam Banerjee
Abstract: Recent advances in machine learning have been supported by the emergence of domain-specific software libraries, enabling streamlined workflows and increased reproducibility. For geospatial machine learning (GeoML), the availability of Earth observation data has outpaced the development of domain libraries to handle its unique challenges, such as varying spatial resolutions, spectral properties, temporal cadence, data coverage, coordinate systems, and file formats. This chapter presents a comprehensive overview of GeoML libraries, analyzing their evolution, core functionalities, and the current ecosystem. It also introduces popular GeoML libraries such as TorchGeo, eo-learn, and Raster Vision, detailing their architecture, supported data types, and integration with ML frameworks. Additionally, it discusses common methodologies for data preprocessing, spatial--temporal joins, benchmarking, and the use of pretrained models. Through a case study in crop type mapping, it demonstrates practical applications of these tools. Best practices in software design, licensing, and testing are highlighted, along with open challenges and future directions, particularly the rise of foundation models and the need for governance in open-source geospatial software. Our aim is to guide practitioners, developers, and researchers in navigating and contributing to the rapidly evolving GeoML landscape.
Authors: Kenechukwu Ogbuagu, Sepehr Maleki, Giuseppe Bruni, Senthil Krishnababu
Abstract: The accurate prediction of flow fields around airfoils is crucial for aerodynamic design and optimisation. Computational Fluid Dynamics (CFD) models are effective but computationally expensive, thus inspiring the development of surrogate models to enable quicker predictions. These surrogate models can be based on deep learning architectures, such as Convolutional Neural Networks (CNNs), Graph Neural Networks (GNNs), and Diffusion Models (DMs). Diffusion models have shown significant promise in predicting complex flow fields. In this work, we propose FoilDiff, a diffusion-based surrogate model with a hybrid-backbone denoising network. This hybrid design combines the power of convolutional feature extraction and transformer-based global attention to generate more adaptable and accurate representations of flow structures. FoilDiff takes advantage of Denoising Diffusion Implicit Model (DDIM) sampling to optimise the efficiency of the sampling process at no additional cost to model generalisation. We used encoded representations of Reynolds number, angle of attack, and airfoil geometry to define the input space for generalisation across a wide range of aerodynamic conditions. When evaluated against state-of-the-art models, FoilDiff shows significant performance improvements, with mean prediction errors reducing by up to 85\% on the same datasets. The results have demonstrated that FoilDiff can provide both more accurate predictions and better-calibrated predictive uncertainty than existing diffusion-based models.
Authors: Lino Gerlach, Liv V{\aa}ge, Thore Gerlach, Elliott Kauffman
Abstract: Fast and efficient machine learning is of growing interest to the scientific community and has spurred significant research into novel model architectures and hardware-aware design. Recent hard? and software co-design approaches have demonstrated impressive results with entirely multiplication-free models. Differentiable Logic Gate Networks (DLGNs), for instance, provide a gradient-based framework for learning optimal combinations of low-level logic gates, setting state-of-the-art trade-offs between accuracy, resource usage, and latency. However, these models suffer from high computational cost during training and do not generalize well to logic blocks with more inputs. In this work, we introduce Walsh-Assisted Relaxation for Probabilistic Look-Up Tables (WARP-LUTs) - a novel gradient-based method that efficiently learns combinations of logic gates with substantially fewer trainable parameters. We demonstrate that WARP-LUTs achieve significantly faster convergence on CIFAR-10 compared to DLGNs, while maintaining comparable accuracy. Furthermore, our approach suggests potential for extension to higher-input logic blocks, motivating future research on extremely efficient deployment on modern FPGAs and its real-time science applications.
Authors: Yanna Ding, Songtao Lu, Yingdong Lu, Tomasz Nowicki, Jianxi Gao
Abstract: Transformer architectures can solve unseen tasks based on input-output pairs in a given prompt due to in-context learning (ICL). Existing theoretical studies on ICL have mainly focused on linear regression tasks, often with i.i.d. inputs. To understand how transformers express ICL when modeling dynamics-driven functions, we investigate Markovian function learning through a structured ICL setup, where we characterize the loss landscape to reveal underlying optimization behaviors. Specifically, we (1) provide the closed-form expression of the global minimizer (in an enlarged parameter space) for a single-layer linear self-attention (LSA) model; (2) prove that recovering transformer parameters that realize the optimal solution is NP-hard in general, revealing a fundamental limitation of one-layer LSA in representing structured dynamical functions; and (3) supply a novel interpretation of a multilayer LSA as performing preconditioned gradient descent to optimize multiple objectives beyond the square loss. These theoretical results are numerically validated using simplified transformers.
Authors: Zhiyang Xun, Shivam Gupta, Eric Price
Abstract: Given a noisy linear measurement $y = Ax + \xi$ of a distribution $p(x)$, and a good approximation to the prior $p(x)$, when can we sample from the posterior $p(x \mid y)$? Posterior sampling provides an accurate and fair framework for tasks such as inpainting, deblurring, and MRI reconstruction, and several heuristics attempt to approximate it. Unfortunately, approximate posterior sampling is computationally intractable in general. To sidestep this hardness, we focus on (local or global) log-concave distributions $p(x)$. In this regime, Langevin dynamics yields posterior samples when the exact scores of $p(x)$ are available, but it is brittle to score--estimation error, requiring an MGF bound (sub-exponential error). By contrast, in the unconditional setting, diffusion models succeed with only an $L^2$ bound on the score error. We prove that combining diffusion models with an annealed variant of Langevin dynamics achieves conditional sampling in polynomial time using merely an $L^4$ bound on the score error.
Authors: Zaishuo Xia, Yukuan Lu, Xinyi Li, Yifan Xu, Yubei Chen
Abstract: A world model is an internal model that simulates how the world evolves. Given past observations and actions, it predicts the future physical state of both the embodied agent and its environment. Accurate world models are essential for enabling agents to think, plan, and reason effectively in complex, dynamic settings. However, existing world models often focus on random generation of open worlds, but neglect the need for high-fidelity modeling of deterministic scenarios (such as fixed-map mazes and static space robot navigation). In this work, we take a step toward building a truly accurate world model by addressing a fundamental yet open problem: constructing a model that can fully clone a deterministic 3D world. 1) Through diagnostic experiment, we quantitatively demonstrate that high-fidelity cloning is feasible and the primary bottleneck for long-horizon fidelity is the geometric structure of the latent representation, not the dynamics model itself. 2) Building on this insight, we show that applying temporal contrastive learning principle as a geometric regularization can effectively curate a latent space that better reflects the underlying physical state manifold, demonstrating that contrastive constraints can serve as a powerful inductive bias for stable world modeling; we call this approach Geometrically-Regularized World Models (GRWM). At its core is a lightweight geometric regularization module that can be seamlessly integrated into standard autoencoders, reshaping their latent space to provide a stable foundation for effective dynamics modeling. By focusing on representation quality, GRWM offers a simple yet powerful pipeline for improving world model fidelity.
Authors: Kshitij Nikhal, Lucas Ackerknecht, Benjamin S. Riggan, Phillip Stahlfeld
Abstract: The decarbonization of transportation relies on the widespread adoption of electric vehicles (EVs), which requires an accurate understanding of charging behavior to ensure cost-effective, grid-resilient infrastructure. Existing work is constrained by small-scale datasets, simple proximity-based modeling of temporal dependencies, and weak generalization to sites with limited operational history. To overcome these limitations, this work proposes a framework that integrates clustering with few-shot forecasting to uncover site archetypes using a novel large-scale dataset of charging demand. The results demonstrate that archetype-specific expert models outperform global baselines in forecasting demand at unseen sites. By establishing forecast performance as a basis for infrastructure segmentation, we generate actionable insights that enable operators to lower costs, optimize energy and pricing strategies, and support grid resilience critical to climate goals.
Authors: Tyrus Whitman, Andrew Particka, Christopher Diers, Ian Griffin, Charuka Wickramasinghe, Pradeep Ranaweera
Abstract: In this study, we present and validate the predictive capability of the Physics-Informed Neural Networks (PINNs) methodology for solving a variety of engineering and biological dynamical systems governed by ordinary differential equations (ODEs). While traditional numerical methods a re effective for many ODEs, they often struggle to achieve convergence in problems involving high stiffness, shocks, irregular domains, singular perturbations, high dimensions, or boundary discontinuities. Alternatively, PINNs offer a powerful approach for handling challenging numerical scenarios. In this study, classical ODE problems are employed as controlled testbeds to systematically evaluate the accuracy, training efficiency, and generalization capability under controlled conditions of the PINNs framework. Although not a universal solution, PINNs can achieve superior results by embedding physical laws directly into the learning process. We first analyze the existence and uniqueness properties of several benchmark problems and subsequently validate the PINNs methodology on these model systems. Our results demonstrate that for complex problems to converge to correct solutions, the loss function components data loss, initial condition loss, and residual loss must be appropriately balanced through careful weighting. We further establish that systematic tuning of hyperparameters, including network depth, layer width, activation functions, learning rate, optimization algorithms, w eight initialization schemes, and collocation point sampling, plays a crucial role in achieving accurate solutions. Additionally, embedding prior knowledge and imposing hard constraints on the network architecture, without loss the generality of the ODE system, significantly enhances the predictive capability of PINNs.
Authors: Kohei Tsuchiyama, Andre Roehm, Takatomo Mihana, Ryoichi Horisaki
Abstract: Physical Neural Networks (PNN) are promising platforms for next-generation computing systems. However, recent advances in digital neural network performance are largely driven by the rapid growth in the number of trainable parameters and, so far, demonstrated PNNs are lagging behind by several orders of magnitude in terms of scale. This mirrors size and performance constraints found in early digital neural networks. In that period, efficient reuse of parameters contributed to the development of parameter-efficient architectures such as convolutional neural networks. In this work, we numerically investigate hardware-friendly weight-tying for PNNs. Crucially, with many PNN systems, there is a time-scale separation between the fast dynamic active elements of the forward pass and the only slowly trainable elements implementing weights and biases. With this in mind,we propose the Reuse of Layers for eXpanding a Neural Network (ReLaX-Net) architecture, which employs a simple layer-by-layer time-multiplexing scheme to increase the effective network depth and efficiently use the number of parameters. We only require the addition of fast switches for existing PNNs. We validate ReLaX-Nets via numerical experiments on image classification and natural language processing tasks. Our results show that ReLaX-Net improves computational performance with only minor modifications to a conventional PNN. We observe a favorable scaling, where ReLaX-Nets exceed the performance of equivalent traditional RNNs or DNNs with the same number of parameters.
Authors: Yong-Woon Kim, Chulung Kang, Yung-Cheol Byun
Abstract: Green hydrogen production via polymer electrolyte membrane (PEM) water electrolysis is pivotal for energy transition, yet hydrogen crossover through membranes threatens safety and economic viability-approaching explosive limits (4 mol% H$_2$ in O$_2$) while reducing Faradaic efficiency by 2.5%. Current physics-based models require extensive calibration and computational resources that preclude real-time implementation, while purely data-driven approaches fail to extrapolate beyond training conditions-critical for dynamic electrolyzer operation. Here we present the first application of physics-informed neural networks (PINNs) for hydrogen crossover prediction, integrating mass conservation, Fick's diffusion law, and Henry's solubility law within a compact architecture (17,793 parameters). Validated across six membranes under industrially relevant conditions (0.05-5.0 A/cm$^2$, 1-200 bar, 25-85{\deg}C), our PINN achieves exceptional accuracy (R$^{2}$ = 99.84% $\pm$ 0.15\%, RMSE = 0.0932% $\pm$ 0.0438%) based on five-fold cross-validation, with sub-millisecond inference times suitable for real-time control. Remarkably, the model maintains R$^2$ > 86% when predicting crossover at pressures 2.5x beyond training range-substantially outperforming pure neural networks (R$^2$ = 43.4%). The hardware-agnostic deployment, from desktop CPUs to edge devices (Raspberry Pi 4), enables distributed safety monitoring essential for gigawatt-scale installations. By bridging physical rigor and computational efficiency, this work establishes a new paradigm for real-time electrolyzer monitoring, accelerating deployment of safe, efficient green hydrogen infrastructure crucial for net-zero emissions targets.
Authors: Lejun Ai, Yulong Li, Haodong Yi, Jixuan Xie, Yue Wang, Jia Liu, Min Chen, Rui Wang
Abstract: Automatic sleep staging plays a vital role in assessing sleep quality and diagnosing sleep disorders. Most existing methods rely heavily on long and continuous EEG recordings, which poses significant challenges for data acquisition in resource-constrained systems, such as wearable or home-based monitoring systems. In this paper, we propose the task of resource-efficient sleep staging, which aims to reduce the amount of signal collected per sleep epoch while maintaining reliable classification performance. To solve this task, we adopt the masking and prompt learning strategy and propose a novel framework called Mask-Aware Sleep Staging (MASS). Specifically, we design a multi-level masking strategy to promote effective feature modeling under partial and irregular observations. To mitigate the loss of contextual information introduced by masking, we further propose a hierarchical prompt learning mechanism that aggregates unmasked data into a global prompt, serving as a semantic anchor for guiding both patch-level and epoch-level feature modeling. MASS is evaluated on four datasets, demonstrating state-of-the-art performance, especially when the amount of data is very limited. This result highlights its potential for efficient and scalable deployment in real-world low-resource sleep monitoring environments.
Authors: Lionel Levine, Haniyeh Ehsani Oskouie, Sajjad Ghiasvand, Majid Sarrafzadeh
Abstract: We introduce MI-to-Mid Distilled Compression (M2M-DC), a two-scale, shape-safe compression framework that interleaves information-guided block pruning with progressive inner slicing and staged knowledge distillation (KD). First, M2M-DC ranks residual (or inverted-residual) blocks by a label-aware mutual information (MI) signal and removes the least informative units (structured prune-after-training). It then alternates short KD phases with stage-coherent, residual-safe channel slicing: (i) stage "planes" (co-slicing conv2 out-channels with the downsample path and next-stage inputs), and (ii) an optional mid-channel trim (conv1 out / bn1 / conv2 in). This targets complementary redundancy, whole computational motifs and within-stage width while preserving residual shape invariants. On CIFAR-100, M2M-DC yields a clean accuracy-compute frontier. For ResNet-18, we obtain 85.46% Top-1 with 3.09M parameters and 0.0139 GMacs (72% params, 63% GMacs vs. teacher; mean final 85.29% over three seeds). For ResNet-34, we reach 85.02% Top-1 with 5.46M params and 0.0195 GMacs (74% / 74% vs. teacher; mean final 84.62%). Extending to inverted-residuals, MobileNetV2 achieves a mean final 68.54% Top-1 at 1.71M params (27%) and 0.0186 conv GMacs (24%), improving over the teacher's 66.03% by +2.5 points across three seeds. Because M2M-DC exposes only a thin, architecture-aware interface (blocks, stages, and down sample/skip wiring), it generalizes across residual CNNs and extends to inverted-residual families with minor legalization rules. The result is a compact, practical recipe for deployment-ready models that match or surpass teacher accuracy at a fraction of the compute.
Authors: Weixuan Ou, Yanzhao Zheng, Shuoshuo Sun, Wei Zhang, Baohua Dong, Hangcheng Zhu, Ruohui Huang, Gang Yu, Pengwei Yan, Yifan Qiao
Abstract: Reinforcement Learning (RL) has been shown to improve the capabilities of large language models (LLMs). However, applying RL to open-domain tasks faces two key challenges: (1) the inherent subjectivity of these tasks prevents the verifiable rewards as required by Reinforcement Learning with Verifiable Rewards (RLVR); (2) Reinforcement Learning from Human Feedback (RLHF) relies on external reward mechanisms. To overcome these limitations, we propose Self-Examining Reinforcement Learning (SERL), a novel self-improving framework where the LLM serves as both Actor and Judge. SERL introduces two synergistic reward mechanisms without any external signals. On the one hand, to improve the Actor's capability, we derive rewards from Copeland-style pairwise comparison judgments across a group of generated responses. On the other hand, a self-consistency reward that encourages coherent judgments is proposed to improve the Judge's reliability. This process refines the Judge's capability, which in turn provides a more robust reward for Actor. Experiments show that our method outperforms existing self-improvement training methods. SERL improves the LC win rate of Qwen3-8B on AlpacaEval 2 from 52.37% to 59.90%. To the best of our knowledge, our method achieves state-of-the-art performance among self-improving approaches. Furthermore, it achieves a performance comparable to significantly larger models like Qwen3-32B, demonstrating superior effectiveness and robustness on open-domain tasks.
Authors: Yiming Tang, Abhijeet Sinha, Dianbo Liu
Abstract: Although recent generative models are remarkably capable of producing instruction-following and realistic outputs, they remain prone to notable physical plausibility failures. Though critical in applications, these physical plausibility errors often escape detection by existing evaluation methods. Furthermore, no framework exists for automatically identifying and interpreting specific physical error patterns in natural language, preventing targeted model improvements. We introduce Matryoshka Transcoders, a novel framework for the automatic discovery and interpretation of physical plausibility features in generative models. Our approach extends the Matryoshka representation learning paradigm to transcoder architectures, enabling hierarchical sparse feature learning at multiple granularity levels. By training on intermediate representations from a physical plausibility classifier and leveraging large multimodal models for interpretation, our method identifies diverse physics-related failure modes without manual feature engineering, achieving superior feature relevance and feature accuracy compared to existing approaches. We utilize the discovered visual patterns to establish a benchmark for evaluating physical plausibility in generative models. Our analysis of eight state-of-the-art generative models provides valuable insights into how these models fail to follow physical constraints, paving the way for further model improvements.
Authors: Bingyu Hui, Lebin Yu, Quanming Yao, Yunpeng Qu, Xudong Zhang, Jian Wang
Abstract: Zero-shot coordination(ZSC), a key challenge in multi-agent game theory, has become a hot topic in reinforcement learning (RL) research recently, especially in complex evolving games. It focuses on the generalization ability of agents, requiring them to coordinate well with collaborators from a diverse, potentially evolving, pool of partners that are not seen before without any fine-tuning. Population-based training, which approximates such an evolving partner pool, has been proven to provide good zero-shot coordination performance; nevertheless, existing methods are limited by computational resources, mainly focusing on optimizing diversity in small populations while neglecting the potential performance gains from scaling population size. To address this issue, this paper proposes the Scalable Population Training (ScaPT), an efficient RL training framework comprising two key components: a meta-agent that efficiently realizes a population by selectively sharing parameters across agents, and a mutual information regularizer that guarantees population diversity. To empirically validate the effectiveness of ScaPT, this paper evaluates it along with representational frameworks in Hanabi cooperative game and confirms its superiority.
Authors: Rajit Rajpal, Benedict Leimkuhler, Yuanhao Jiang
Abstract: Bayesian neural networks (BNNs) require scalable sampling algorithms to approximate posterior distributions over parameters. Existing stochastic gradient Markov Chain Monte Carlo (SGMCMC) methods are highly sensitive to the choice of stepsize and adaptive variants such as pSGLD typically fail to sample the correct invariant measure without addition of a costly divergence correction term. In this work, we build on the recently proposed `SamAdams' framework for timestep adaptation (Leimkuhler, Lohmann, and Whalley 2025), introducing an adaptive scheme: SA-SGLD, which employs time rescaling to modulate the stepsize according to a monitored quantity (typically the local gradient norm). SA-SGLD can automatically shrink stepsizes in regions of high curvature and expand them in flatter regions, improving both stability and mixing without introducing bias. We show that our method can achieve more accurate posterior sampling than SGLD on high-curvature 2D toy examples and in image classification with BNNs using sharp priors.
Authors: Shuvom Sadhuka, Sophia Lin, Emma Pierson, Bonnie Berger
Abstract: Many sequential decision settings in healthcare feature funnel structures characterized by a series of stages, such as screenings or evaluations, where the number of patients who advance to each stage progressively decreases and decisions become increasingly costly. For example, an oncologist may first conduct a breast exam, followed by a mammogram for patients with concerning exams, followed by a biopsy for patients with concerning mammograms. A key challenge is that the ground truth outcome, such as the biopsy result, is only revealed at the end of this funnel. The selective censoring of the ground truth can introduce statistical biases in risk estimation, especially in underserved patient groups, whose outcomes are more frequently censored. We develop a Bayesian model for funnel decision structures, drawing from prior work on selective labels and censoring. We first show in synthetic settings that our model is able to recover the true parameters and predict outcomes for censored patients more accurately than baselines. We then apply our model to a dataset of emergency department visits, where in-hospital mortality is observed only for those who are admitted to either the hospital or ICU. We find that there are gender-based differences in hospital and ICU admissions. In particular, our model estimates that the mortality risk threshold to admit women to the ICU is higher for women (5.1%) than for men (4.5%).
Authors: Qing Yao, Lijian Gao, Qirong Mao, Dong Ming
Abstract: Diffusion models serve as a powerful generative framework for solving inverse problems. However, they still face two key challenges: 1) the distortion-perception tradeoff, where improving perceptual quality often degrades reconstruction fidelity, and 2) the exposure bias problem, where the training-inference input mismatch leads to prediction error accumulation and reduced reconstruction quality. In this work, we propose the Regularized Schr\"odinger Bridge (RSB), an adaptation of Schr\"odinger Bridge tailored for inverse problems that addresses the above limitations. RSB employs a novel regularized training strategy that perturbs both the input states and targets, effectively mitigating exposure bias by exposing the model to simulated prediction errors and also alleviating distortion by well-designed interpolation via the posterior mean. Extensive experiments on two typical inverse problems for speech enhancement demonstrate that RSB outperforms state-of-the-art methods, significantly improving distortion metrics and effectively reducing exposure bias.
Authors: Julia Peters, Karin Mora, Miguel D. Mahecha, Chaonan Ji, David Montero, Clemens Mosig, Guido Kraemer
Abstract: Earth observation (EO) foundation models have emerged as an effective approach to derive latent representations of the Earth system from various remote sensing sensors. These models produce embeddings that can be used as analysis-ready datasets, enabling the modelling of ecosystem dynamics without extensive sensor-specific preprocessing. However, existing models typically operate at fixed spatial or temporal scales, limiting their use for ecological analyses that require both fine spatial detail and high temporal fidelity. To overcome these limitations, we propose a representation learning framework that integrates different EO modalities into a unified feature space at high spatio-temporal resolution. We introduce the framework using Sentinel-1 and Sentinel-2 data as representative modalities. Our approach produces a latent space at native 10 m resolution and the temporal frequency of cloud-free Sentinel-2 acquisitions. Each sensor is first modeled independently to capture its sensor-specific characteristics. Their representations are then combined into a shared model. This two-stage design enables modality-specific optimisation and easy extension to new sensors, retaining pretrained encoders while retraining only fusion layers. This enables the model to capture complementary remote sensing data and to preserve coherence across space and time. Qualitative analyses reveal that the learned embeddings exhibit high spatial and semantic consistency across heterogeneous landscapes. Quantitative evaluation in modelling Gross Primary Production reveals that they encode ecologically meaningful patterns and retain sufficient temporal fidelity to support fine-scale analyses. Overall, the proposed framework provides a flexible, analysis-ready representation learning approach for environmental applications requiring diverse spatial and temporal resolutions.
Authors: Amisha Priyadarshini, Sergio Gago-Masague
Abstract: Despite recent advances in fairness-aware machine learning, predictive models often exhibit discriminatory behavior towards marginalized groups. Such unfairness might arise from biased training data, model design, or representational disparities across groups, posing significant challenges in high-stakes decision-making domains such as college admissions. While existing fair learning models aim to mitigate bias, achieving an optimal trade-off between fairness and accuracy remains a challenge. Moreover, the reliance on black-box models hinders interpretability, limiting their applicability in socially sensitive domains. To circumvent these issues, we propose integrating Kolmogorov-Arnold Networks (KANs) within a fair adversarial learning framework. Leveraging the adversarial robustness and interpretability of KANs, our approach facilitates stable adversarial learning. We derive theoretical insights into the spline-based KAN architecture that ensure stability during adversarial optimization. Additionally, an adaptive fairness penalty update mechanism is proposed to strike a balance between fairness and accuracy. We back these findings with empirical evidence on two real-world admissions datasets, demonstrating the proposed framework's efficiency in achieving fairness across sensitive attributes while preserving predictive performance.
Authors: Wanlong Fang, Tianle Zhang, Alvin Chan
Abstract: Multimodal learning often relies on aligning representations across modalities to enable effective information integration, an approach traditionally assumed to be universally beneficial. However, prior research has primarily taken an observational approach, examining naturally occurring alignment in multimodal data and exploring its correlation with model performance, without systematically studying the direct effects of explicitly enforced alignment between representations of different modalities. In this work, we investigate how explicit alignment influences both model performance and representation alignment under different modality-specific information structures. Specifically, we introduce a controllable contrastive learning module that enables precise manipulation of alignment strength during training, allowing us to explore when explicit alignment improves or hinders performance. Our results on synthetic and real datasets under different data characteristics show that the impact of explicit alignment on the performance of unimodal models is related to the characteristics of the data: the optimal level of alignment depends on the amount of redundancy between the different modalities. We identify an optimal alignment strength that balances modality-specific signals and shared redundancy in the mixed information distributions. This work provides practical guidance on when and how explicit alignment should be applied to achieve optimal unimodal encoder performance.
Authors: Runhao Jiang, Chengzhi Jiang, Rui Yan, Huajin Tang
Abstract: The surrogate gradient (SG) method has shown significant promise in enhancing the performance of deep spiking neural networks (SNNs), but it also introduces vulnerabilities to adversarial attacks. Although spike coding strategies and neural dynamics parameters have been extensively studied for their impact on robustness, the critical role of gradient magnitude, which reflects the model's sensitivity to input perturbations, remains underexplored. In SNNs, the gradient magnitude is primarily determined by the interaction between the membrane potential distribution (MPD) and the SG function. In this study, we investigate the relationship between the MPD and SG and their implications for improving the robustness of SNNs. Our theoretical analysis reveals that reducing the proportion of membrane potentials lying within the gradient-available range of the SG function effectively mitigates the sensitivity of SNNs to input perturbations. Building upon this insight, we propose a novel MPD-driven surrogate gradient regularization (MPD-SGR) method, which enhances robustness by explicitly regularizing the MPD based on its interaction with the SG function. Extensive experiments across multiple image classification benchmarks and diverse network architectures confirm that the MPD-SGR method significantly enhances the resilience of SNNs to adversarial perturbations and exhibits strong generalizability across diverse network configurations, SG functions, and spike encoding schemes.
Authors: Yanxin Peng, Qingping Li, Baodong Wu, Shigang Li, Guohao Dai, Shengen Yan, Yu Wang
Abstract: As large language models (LLMs) continue to grow in size and complexity, efficient checkpoint saving\&loading has become crucial for managing storage, memory usage, and fault tolerance in LLM training. The current works do not comprehensively take into account the optimization of these several aspects. This paper proposes a novel checkpoint sparsification and quantization method that adapts dynamically to different training stages and model architectures. We present a comprehensive analysis of existing lossy and lossless compression techniques, identify current limitations, and introduce our adaptive approach that balances compression ratio, speed, and precision impact throughout the training process. Experiments on different sizes of LLMs demonstrate that our bitmask-based sparsification method achieves 16x compression ratio without compromising model accuracy. Additionally, the cluster-based quantization method achieves 2x compression ratio with little precision loss.
Authors: Hao Wei, Aleksandra Franz, Bjoern List, Nils Thuerey
Abstract: When simulating partial differential equations, hybrid solvers combine coarse numerical solvers with learned correctors. They promise accelerated simulations while adhering to physical constraints. However, as shown in our theoretical framework, directly applying learned corrections to solver outputs leads to significant autoregressive errors, which originate from amplified perturbations that accumulate during long-term rollouts, especially in chaotic regimes. To overcome this, we propose the Indirect Neural Corrector ($\mathrm{INC}$), which integrates learned corrections into the governing equations rather than applying direct state updates. Our key insight is that $\mathrm{INC}$ reduces the error amplification on the order of $\Delta t^{-1} + L$, where $\Delta t$ is the timestep and $L$ the Lipschitz constant. At the same time, our framework poses no architectural requirements and integrates seamlessly with arbitrary neural networks and solvers. We test $\mathrm{INC}$ in extensive benchmarks, covering numerous differentiable solvers, neural backbones, and test cases ranging from a 1D chaotic system to 3D turbulence. $\mathrm{INC}$ improves the long-term trajectory performance ($R^2$) by up to 158.7%, stabilizes blowups under aggressive coarsening, and for complex 3D turbulence cases yields speed-ups of several orders of magnitude. $\mathrm{INC}$ thus enables stable, efficient PDE emulation with formal error reduction, paving the way for faster scientific and engineering simulations with reliable physics guarantees. Our source code is available at https://github.com/tum-pbs/INC
Authors: Dahao Tang, Nan Yang, Yanli Li, Zhiyu Zhu, Zhibo Jin, Dong Yuan
Abstract: Selecting an appropriate look-back horizon remains a fundamental challenge in time series forecasting (TSF), particularly in the federated learning scenarios where data is decentralized, heterogeneous, and often non-independent. While recent work has explored horizon selection by preserving forecasting-relevant information in an intrinsic space, these approaches are primarily restricted to centralized and independently distributed settings. This paper presents a principled framework for adaptive horizon selection in federated time series forecasting through an intrinsic space formulation. We introduce a synthetic data generator (SDG) that captures essential temporal structures in client data, including autoregressive dependencies, seasonality, and trend, while incorporating client-specific heterogeneity. Building on this model, we define a transformation that maps time series windows into an intrinsic representation space with well-defined geometric and statistical properties. We then derive a decomposition of the forecasting loss into a Bayesian term, which reflects irreducible uncertainty, and an approximation term, which accounts for finite-sample effects and limited model capacity. Our analysis shows that while increasing the look-back horizon improves the identifiability of deterministic patterns, it also increases approximation error due to higher model complexity and reduced sample efficiency. We prove that the total forecasting loss is minimized at the smallest horizon where the irreducible loss starts to saturate, while the approximation loss continues to rise. This work provides a rigorous theoretical foundation for adaptive horizon selection for time series forecasting in federated learning.
Authors: Akira Tamamori
Abstract: Kernel-based learning methods can dramatically increase the storage capacity of Hopfield networks, yet the dynamical mechanism behind this enhancement remains poorly understood. We address this gap by conducting a geometric analysis of the network's energy landscape. We introduce a novel metric, "Pinnacle Sharpness," to quantify the local stability of attractors. By systematically varying the kernel width and storage load, we uncover a rich phase diagram of attractor shapes. Our central finding is the emergence of a "ridge of optimization," where the network maximizes attractor stability under challenging high-load and global-kernel conditions. Through a theoretical decomposition of the landscape gradient into a direct "driving" force and an indirect "feedback" force, we reveal the origin of this phenomenon. The optimization ridge corresponds to a regime of strong anti-correlation between the two forces, where the direct force, amplified by the high storage load, dominates the opposing collective feedback force. This demonstrates a sophisticated self-organization mechanism: the network adaptively harnesses inter-pattern interactions as a cooperative feedback control system to sculpt a robust energy landscape. Our findings provide a new physical picture for the stability of high-capacity associative memories and offer principles for their design.
Authors: Zichong Wang, Zhipeng Yin, Liping Yang, Jun Zhuang, Rui Yu, Qingzhao Kong, Wenbin Zhang
Abstract: Ensuring fairness in Graph Neural Networks is fundamental to promoting trustworthy and socially responsible machine learning systems. In response, numerous fair graph learning methods have been proposed in recent years. However, most of them assume full access to demographic information, a requirement rarely met in practice due to privacy, legal, or regulatory restrictions. To this end, this paper introduces a novel fair graph learning framework that mitigates bias in graph learning under limited demographic information. Specifically, we propose a mechanism guided by partial demographic data to generate proxies for demographic information and design a strategy that enforces consistent node embeddings across demographic groups. In addition, we develop an adaptive confidence strategy that dynamically adjusts each node's contribution to fairness and utility based on prediction confidence. We further provide theoretical analysis demonstrating that our framework, FairGLite, achieves provable upper bounds on group fairness metrics, offering formal guarantees for bias mitigation. Through extensive experiments on multiple datasets and fair graph learning frameworks, we demonstrate the framework's effectiveness in both mitigating bias and maintaining model utility.
Authors: Kaichi Irie, Shuhei Watanabe, Masaki Onishi
Abstract: Bayesian optimization (BO) efficiently finds high-performing parameters by maximizing an acquisition function, which models the promise of parameters. A major computational bottleneck arises in acquisition function optimization, where multi-start optimization (MSO) with quasi-Newton (QN) methods is required due to the non-convexity of the acquisition function. BoTorch, a widely used BO library, currently optimizes the summed acquisition function over multiple points, leading to the speedup of MSO owing to PyTorch batching. Nevertheless, this paper empirically demonstrates the suboptimality of this approach in terms of off-diagonal approximation errors in the inverse Hessian of a QN method, slowing down its convergence. To address this problem, we propose to decouple QN updates using a coroutine while batching the acquisition function calls. Our approach not only yields the theoretically identical convergence to the sequential MSO but also drastically reduces the wall-clock time compared to the previous approaches. Our approach is available in GPSampler in Optuna, effectively reducing its computational overhead.
Authors: Matthew T. C. Li, Tiangang Cui, Fengyi Li, Youssef Marzouk, Olivier Zahm
Abstract: Identifying low-dimensional structure in high-dimensional probability measures is an essential pre-processing step for efficient sampling. We introduce a method for identifying and approximating a target measure $\pi$ as a perturbation of a given reference measure $\mu$ along a few significant directions of $\mathbb{R}^{d}$. The reference measure can be a Gaussian or a nonlinear transformation of a Gaussian, as commonly arising in generative modeling. Our method extends prior work on minimizing majorizations of the Kullback--Leibler divergence to identify optimal approximations within this class of measures. Our main contribution unveils a connection between the \emph{dimensional} logarithmic Sobolev inequality (LSI) and approximations with this ansatz. Specifically, when the target and reference are both Gaussian, we show that minimizing the dimensional LSI is equivalent to minimizing the KL divergence restricted to this ansatz. For general non-Gaussian measures, the dimensional LSI produces majorants that uniformly improve on previous majorants for gradient-based dimension reduction. We further demonstrate the applicability of this analysis to the squared Hellinger distance, where analogous reasoning shows that the dimensional Poincar\'e inequality offers improved bounds.
Authors: Xinnan Zhang, Jialin Wu, Junyi Xie, Tianlong Chen, Kaixiong Zhou
Abstract: The prediction modeling of drug-target interactions is crucial to drug discovery and design, which has seen rapid advancements owing to deep learning technologies. Recently developed methods, such as those based on graph neural networks (GNNs) and Transformers, demonstrate exceptional performance across various datasets by effectively extracting structural information. However, the benchmarking of these novel methods often varies significantly in terms of hyperparameter settings and datasets, which limits algorithmic progress. In view of these, we conducted a comprehensive survey and benchmark for drug-target interaction modeling from a structural perspective via integrating tens of explicit (i.e., GNN-based) and implicit (i.e., Transformer-based) structure learning algorithms. We conducted a macroscopical comparison between these two classes of encoding strategies as well as the different featurization techniques that inform molecules' chemical and physical properties. We then carry out the microscopical comparison between all the integrated models across the six datasets via comprehensively benchmarking their effectiveness and efficiency. To ensure fairness, we investigate model performance under individually optimized configuration. Remarkably, the summarized insights from the benchmark studies lead to the design of model combos. We demonstrate that our combos can achieve new state-of-the-art performance on various datasets associated with cost-effective memory and computation.
Authors: Shih-Kai Chou, Jernej Hribar, Vid Han\v{z}el, Mihael Mohor\v{c}i\v{c}, Carolina Fortuna
Abstract: Artificial Intelligence (AI) is being incorporated in several optimization, scheduling, orchestration as well as in native communication network functions. This paradigm shift results in increased energy consumption, however, quantifying the end-to-end energy consumption of adding intelligence to communication systems remains an open challenge since conventional energy consumption metrics focus on either communication, computation infrastructure, or model development. To address this, we propose a new metric, the Energy Cost of AI Lifecycle (eCAL) of an AI model in a system. eCAL captures the energy consumption throughout the development, deployment and utilization of an AI-model providing intelligence in a communication network by (i) analyzing the complexity of data collection and manipulation in individual components and (ii) deriving overall and per-bit energy consumption. We show that as a trained AI model is used more frequently for inference, its energy cost per inference decreases, since the fixed training energy is amortized over a growing number of inferences. For a simple case study we show that eCAL for 100 inferences is 2.73 times higher than for 1000 inferences. Additionally, we have developed a modular and extendable open-source simulation tool to enable researchers, practitioners, and engineers to calculate the end-to-end energy cost with various configurations and across various systems, ensuring adaptability to diverse use cases.
Authors: Florentina Voboril, Vaidyanathan Peruvemba Ramaswamy, Stefan Szeider
Abstract: Streamlining constraints (or streamliners, for short) narrow the search space, enhancing the speed and feasibility of solving complex constraint satisfaction problems. Traditionally, streamliners were crafted manually or generated through systematically combined atomic constraints with high-effort offline testing. Our approach utilizes the creativity of Large Language Models (LLMs) to propose effective streamliners for problems specified in the MiniZinc constraint programming language and integrates feedback to the LLM with quick empirical tests for validation. Evaluated across seven diverse constraint satisfaction problems, our method achieves substantial runtime reductions. We compare the results to obfuscated and disguised variants of the problem to see whether the results depend on LLM memorization. We also analyze whether longer off-line runs improve the quality of streamliners and whether the LLM can propose good combinations of streamliners.
Authors: Ji Ma
Abstract: As Large Language Model (LLM)-based agents increasingly engage with human society, how well do we understand their prosocial behaviors? We (1) investigate how LLM agents' prosocial behaviors can be induced by different personas and benchmarked against human behaviors; and (2) introduce a social science approach to evaluate LLM agents' decision-making. We explored how different personas and experimental framings affect these AI agents' altruistic behavior in dictator games and compared their behaviors within the same LLM family, across various families, and with human behaviors. The findings reveal that merely assigning a human-like identity to LLMs does not produce human-like behaviors. These findings suggest that LLM agents' reasoning does not consistently exhibit textual markers of human decision-making in dictator games and that their alignment with human behavior varies substantially across model architectures and prompt formulations; even worse, such dependence does not follow a clear pattern. As society increasingly integrates machine intelligence, "Prosocial AI" emerges as a promising and urgent research direction in philanthropic studies.
Authors: Jingkun Ma, Runzhe Zhan, Yang Li, Di Sun, Hou Pong Chan, Lidia S. Chao, Derek F. Wong
Abstract: A hallmark of advanced artificial intelligence is the capacity to progress from passive visual perception to the strategic modification of visual information to facilitate complex reasoning. This advanced capability, however, remains critically underdeveloped in current Large Multi-modal Models (LMMs). The deficiency is often masked by evaluation metrics that prioritize final-answer accuracy, creating an illusion of competence where genuine reasoning is absent. Using the domain of geometric problem-solving as a precise instrument, we probe this issue through tasks that require constructing visual aids. To this end, we introduce \textbf{VisAidMath}, a challenging benchmark, and our novel Three-Layered Funnel Evaluation Framework. This framework moves beyond simple accuracy (ACCU) to scrutinize the generation of valid visual aids (PVA) and the soundness of subsequent reasoning steps (SPRS). Our extensive experiments on state-of-the-art models, including Doubao-Seed-1.6 and o4, reveal a profound ``Reasoning Illusion''. We observe that high surface-level accuracy conceals a catastrophic failure in the models' ability to produce valid visual aids or to reason from them. Our findings expose a fundamental schism between visual perception and logical deduction in modern LMMs. We host an evaluation platform at CodaBench for testing publicly. Homepage: https://nlp2ct.github.io/VisAidMathHomepage/ Evaluation: https://www.codabench.org/competitions/7634/
URLs: https://nlp2ct.github.io/VisAidMathHomepage/, https://www.codabench.org/competitions/7634/
Authors: Ziyao Zeng, Jingcheng Ni, Daniel Wang, Patrick Rim, Younjoon Chung, Fengyu Yang, Byung-Woo Hong, Alex Wong
Abstract: Traditional monocular depth estimation suffers from inherent ambiguity and visual nuisances. We demonstrate that language can enhance monocular depth estimation by providing an additional condition (rather than images alone) aligned with plausible 3D scenes, thereby reducing the solution space for depth estimation. This conditional distribution is learned during the text-to-image pre-training of diffusion models. To generate images under various viewpoints and layouts that precisely reflect textual descriptions, the model implicitly models object sizes, shapes, and scales, their spatial relationships, and the overall scene structure. In this paper, Iris, we investigate the benefits of our strategy to integrate text descriptions into training and inference of diffusion-based depth estimation models. We experiment with three different diffusion-based monocular depth estimators (Marigold, Lotus, and E2E-FT) and their variants. By training on HyperSim and Virtual KITTI, and evaluating on NYUv2, KITTI, ETH3D, ScanNet, and DIODE, we find that our strategy improves the overall monocular depth estimation accuracy, especially in small areas. It also improves the model's depth perception of specific regions described in the text. We find that by providing more details in the text, the depth prediction can be iteratively refined. Simultaneously, we find that language can act as a constraint to accelerate the convergence of both training and the inference diffusion trajectory. Code and generated text data will be released upon acceptance.
Authors: Florin Condrea, Saikiran Rapaka, Marius Leordeanu
Abstract: Pulmonary Embolism (PE) are a leading cause of cardiovascular death. Computed tomographic pulmonary angiography (CTPA) is the gold standard for PE diagnosis, with growing interest in AI-based diagnostic assistance. However, these algorithms are limited by scarce fine-grained annotations of thromboembolic burden. We address this challenge with iExplain, a weakly supervised learning algorithm that transforms coarse image-level annotations into detailed pixel-level PE masks through iterative model explainability. Our approach generates soft segmentation maps used to mask detected regions, enabling the process to repeat and discover additional embolisms that would be missed in a single pass. This iterative refinement effectively captures complete PE regions and detects multiple distinct embolisms. Models trained on these automatically generated annotations achieve excellent PE detection performance, with significant improvements at each iteration. We demonstrate iExplain's effectiveness on the RSPECT augmented dataset, achieving results comparable to strongly supervised methods while outperforming existing weakly supervised methods.
Authors: Tim-Lukas Habich, Aran Mohammad, Simon F. G. Ehlers, Martin Bensch, Thomas Seel, Moritz Schappler
Abstract: Soft robots can revolutionize several applications with high demands on dexterity and safety. When operating these systems, real-time estimation and control require fast and accurate models. However, prediction with first-principles (FP) models is slow, and learned black-box models have poor generalizability. Physics-informed machine learning offers excellent advantages here, but it is currently limited to simple, often simulated systems without considering changes after training. We propose physics-informed neural networks (PINNs) for articulated soft robots (ASRs) with a focus on data efficiency. The amount of expensive real-world training data is reduced to a minimum -- one dataset in one system domain. Two hours of data in different domains are used for a comparison against two gold-standard approaches: In contrast to a recurrent neural network, the PINN provides a high generalizability. The prediction speed of an accurate FP model is exceeded with the PINN by up to a factor of 467 at slightly reduced accuracy. This enables nonlinear model predictive control (MPC) of a pneumatic ASR. Accurate position tracking with the MPC running at 47 Hz is achieved in six dynamic experiments.
Authors: Jusen Du, Weigao Sun, Disen Lan, Jiaxi Hu, Yu Cheng
Abstract: Linear sequence modeling methods, such as linear attention, state space modeling, and linear RNNs, offer significant efficiency improvements by reducing the complexity of training and inference. However, these methods typically compress the entire input sequence into a single fixed-size memory state, which leads to suboptimal performance on recall-intensive tasks. To address this limitation, we introduce a novel architecture called Mixture-of-Memories (MoM). MoM utilizes multiple independent memory states, with a router network directing input tokens to specific memory states. This approach greatly enhances the overall memory capacity while minimizing memory interference. MoM serves as a general framework that can be seamlessly combined with diverse memory update mechanisms across linear models. As a result, MoM performs exceptionally well on recall-intensive tasks, surpassing existing linear sequence modeling techniques. Despite incorporating multiple memory states, the computation of each memory state remains linear in complexity, allowing MoM to retain the linear-complexity advantage during training, while constant-complexity during inference. Our experimental results show that MoM outperforms current linear sequence models on downstream language tasks, particularly recall-intensive tasks, and even achieves performance comparable to Transformer models. The code is released at https://github.com/OpenSparseLLMs/MoM and is also released as a part of https://github.com/OpenSparseLLMs/Linear-MoE.
URLs: https://github.com/OpenSparseLLMs/MoM, https://github.com/OpenSparseLLMs/Linear-MoE.
Authors: Gabriel Patron, Zhiwei Xu, Ishan Kapnadak, Felipe Maia Polo
Abstract: Personalization is central to human-AI interaction, yet current diffusion-based image generation systems remain largely insensitive to user diversity. Existing attempts to address this often rely on costly paired preference data or introduce latency through Large Language Models. In this work, we introduce REBECA (REcommendations BEyond CAtalogs), a lightweight and scalable framework for personalized image generation that learns directly from implicit feedback signals such as likes, ratings, and clicks. Instead of fine-tuning the underlying diffusion model, REBECA employs a two-stage process: training a conditional diffusion model to sample user- and rating-specific image embeddings, which are subsequently decoded into images using a pretrained diffusion backbone. This approach enables efficient, fine-tuning-free personalization across large user bases. We rigorously evaluate REBECA on real-world datasets, proposing a novel statistical personalization verifier and a permutation-based hypothesis test to assess preference alignment. Our results demonstrate that REBECA consistently produces high-fidelity images tailored to individual tastes, outperforming baselines while maintaining computational efficiency.
Authors: Fethi Harkat (EDP, DT), Guillaume Gey (DT), Val\'erie Perrier (EDP), K\'evin Polisano (SVH), Tiphaine Deuberet (DT)
Abstract: Traditional feature extraction and projection techniques, such as Principal Component Analysis, struggle to adequately represent X-Ray Transmission (XRT) Multi-Energy (ME) images, limiting the performance of neural networks in decision-making processes. To address this issue, we propose a method that approximates the dataset topology by constructing adjacency graphs using the Uniform Manifold Approximation and Projection. This approach captures nonlinear correlations within the data, significantly improving the performance of machine learning algorithms, particularly in processing Hyperspectral Images (HSI) from X-ray transmission spectroscopy. This technique not only preserves the global structure of the data but also enhances feature separability, leading to more accurate and robust classification results.
Authors: Yingyu Lin, Erchi Wang, Yi-An Ma, Yu-Xiang Wang
Abstract: We propose a framework to convert $(\varepsilon, \delta)$-approximate Differential Privacy (DP) mechanisms into $(\varepsilon', 0)$-pure DP mechanisms under certain conditions, a process we call ``purification.'' This algorithmic technique leverages randomized post-processing with calibrated noise to eliminate the $\delta$ parameter while achieving near-optimal privacy-utility tradeoff for pure DP. It enables a new design strategy for pure DP algorithms: first run an approximate DP algorithm with certain conditions, and then purify. This approach allows one to leverage techniques such as strong composition and propose-test-release that require $\delta>0$ in designing pure-DP methods with $\delta=0$. We apply this framework in various settings, including Differentially Private Empirical Risk Minimization (DP-ERM), stability-based release, and query release tasks. To the best of our knowledge, this is the first work with a statistically and computationally efficient reduction from approximate DP to pure DP. Finally, we illustrate the use of this reduction for proving lower bounds under approximate DP constraints with explicit dependence in $\delta$, avoiding the sophisticated fingerprinting code construction.
Authors: Guanyuan Pan, Tiansheng Zhou, Jianxiang Zhao, Zhi Li, Yugui Lin, Bingtao Ma, Yaqi Wang, Pietro Li\`o, Shuai Wang
Abstract: Circuit link prediction, which identifies missing component connections from incomplete netlists, is crucial in analog circuit design automation. However, existing methods face three main challenges: 1) Insufficient use of topological patterns in circuit graphs reduces prediction accuracy; 2) Data scarcity due to the complexity of annotations hinders model generalization; 3) Limited adaptability to various netlist formats restricts model flexibility. We propose Graph Neural Networks Based Analog Circuit Link Prediction (GNN-ACLP), a graph neural networks (GNNs) based method featuring three innovations to tackle these challenges. First, we introduce the SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction) framework and achieve port-level accuracy in circuit link prediction. Second, we propose Netlist Babel Fish, a netlist format conversion tool that leverages retrieval-augmented generation (RAG) with a large language model (LLM) to enhance the compatibility of netlist formats. Finally, we build a comprehensive dataset, SpiceNetlist, comprising 775 annotated circuits of 7 different types across 10 component classes. Experiments demonstrate accuracy improvements of 16.08% on SpiceNetlist, 11.38% on Image2Net, and 16.01% on Masala-CHAI compared to the baseline in intra-dataset evaluation, while maintaining accuracy from 92.05% to 99.07% in cross-dataset evaluation, demonstrating robust feature transfer capabilities. However, its linear computational complexity makes processing large-scale netlists challenging and requires future addressing.
Authors: Olivia Holguin, Rachel Donati, Seyed bagher Hashemi Natanzi, Bo Tang
Abstract: Mobile jammers pose a critical threat to 5G networks, particularly in military communications. We propose an intelligent anti-jamming framework that integrates Multiple Signal Classification (MUSIC) for high-resolution Direction-of-Arrival (DoA) estimation, Minimum Variance Distortionless Response (MVDR) beamforming for adaptive interference suppression, and machine learning (ML) to enhance DoA prediction for mobile jammers. Extensive simulations in a realistic highway scenario demonstrate that our hybrid approach achieves an average Signal-to-Noise Ratio (SNR) improvement of 9.58 dB (maximum 11.08 dB) and up to 99.8% DoA estimation accuracy. The framework's computational efficiency and adaptability to dynamic jammer mobility patterns outperform conventional anti-jamming techniques, making it a robust solution for securing 5G communications in contested environments.
Authors: Mingxuan Zhang, Oubo Ma, Kang Wei, Songze Li, Shouling Ji
Abstract: Deep reinforcement learning (DRL) has achieved remarkable success in a wide range of sequential decision-making applications, including robotics, healthcare, smart grids, and finance. Recent studies reveal that adversaries can implant backdoors into DRL agents during the training phase. These backdoors can later be activated by specific triggers during deployment, compelling the agent to execute targeted actions and potentially leading to severe consequences, such as drone crashes or vehicle collisions. However, existing backdoor attacks utilize simplistic and heuristic trigger configurations, overlooking the critical impact of trigger design on attack effectiveness. To address this gap, we introduce TooBadRL, the first framework to systematically optimize DRL backdoor triggers across three critical aspects: injection timing, trigger dimension, and manipulation magnitude. Specifically, we first introduce a performance-aware adaptive freezing mechanism to determine the injection timing during training. Then, we formulate trigger selection as an influence attribution problem and apply Shapley value analysis to identify the most influential trigger dimension for injection. Furthermore, we propose an adversarial input synthesis method to optimize the manipulation magnitude under environmental constraints. Extensive evaluations on three DRL algorithms and nine benchmark tasks demonstrate that TooBadRL outperforms five baseline methods in terms of attack success rate while only slightly affecting normal task performance. We further evaluate potential defense strategies from detection and mitigation perspectives. We open-source our code to facilitate reproducibility and further research.
Authors: Simon Ghyselincks, Valeriia Okhmak, Stefano Zampini, George Turkiyyah, David Keyes, Eldad Haber
Abstract: Reconstructing the structural geology and mineral composition of the first few kilometers of the Earth's subsurface from sparse or indirect surface observations remains a long-standing challenge with critical applications in mineral exploration, geohazard assessment, and geotechnical engineering. This inherently ill-posed problem is often addressed by classical geophysical inversion methods, which typically yield a single maximum-likelihood model that fails to capture the full range of plausible geology. The adoption of modern deep learning methods has been limited by the lack of large 3D training datasets. We address this gap with \textit{StructuralGeo}, a geological simulation engine that mimics eons of tectonic, magmatic, and sedimentary processes to generate a virtually limitless supply of realistic synthetic 3D lithological models. Using this dataset, we train both unconditional and conditional generative flow-matching models with a 3D attention U-net architecture. The resulting foundation model can reconstruct multiple plausible 3D scenarios from surface topography and sparse borehole data, depicting structures such as layers, faults, folds, and dikes. By sampling many reconstructions from the same observations, we introduce a probabilistic framework for estimating the size and extent of subsurface features. While the realism of the output is bounded by the fidelity of the training data to true geology, this combination of simulation and generative AI functions offers a flexible prior for probabilistic modeling, regional fine-tuning, and use as an AI-based regularizer in traditional geophysical inversion workflows.
Authors: Zhenting Qi, Fan Nie, Alexandre Alahi, James Zou, Himabindu Lakkaraju, Yilun Du, Eric Xing, Sham Kakade, Hanlin Zhang
Abstract: Modern language model (LM) training has been divided into multiple stages, making it difficult for downstream developers to evaluate the impact of design choices made at each stage. We present EvoLM, a model suite that enables systematic and transparent analysis of LMs' training dynamics across pre-training, continued pre-training, supervised fine-tuning, and reinforcement learning. We train over 100 LMs with 1B and 4B parameters from scratch, and evaluate both upstream (language modeling) and downstream (problem-solving) capabilities, including considerations of both in-domain and out-of-domain generalization. Key insights highlight the diminishing returns from excessive pre-training and post-training, the importance and practices of mitigating forgetting during domain-specific continued pre-training, the crucial role of continued pre-training in bridging pre-training and post-training phases, and various intricate trade-offs when configuring supervised fine-tuning and reinforcement learning. To facilitate open research and reproducibility, we release all pre-trained and post-trained models, training datasets for all stages, and our entire training and evaluation pipeline.
Authors: Charlesquin Kemajou Mbakam, Jonathan Spence, Marcelo Pereyra
Abstract: Diffusion models (DMs) have emerged as powerful image priors in Bayesian computational imaging. Two primary strategies have been proposed for leveraging DMs in this context: Plug-and-Play methods, which are zero-shot and highly flexible but rely on approximations; and specialized conditional DMs, which achieve higher accuracy and faster inference for specific tasks through supervised training. In this work, we introduce a novel framework that integrates deep unfolding and model distillation to transform a DM image prior into a few-step conditional model for posterior sampling. A central innovation of our approach is the unfolding of a Markov chain Monte Carlo (MCMC) algorithm - specifically, the recently proposed LATINO Langevin sampler (Spagnoletti et al., 2025) - representing the first known instance of deep unfolding applied to a Monte Carlo sampling scheme. We demonstrate our proposed unfolded and distilled samplers through extensive experiments and comparisons with the state of the art, where they achieve excellent accuracy and computational efficiency, while retaining the flexibility to adapt to variations in the forward model at inference time.
Authors: Jiaqi Zhang, Juntuo Wang, Zhixin Sun, John Zou, Randall Balestriero
Abstract: Large-scale vision foundation models such as DINOv2 boast impressive performances by leveraging massive architectures and training datasets. But numerous scenarios require practitioners to reproduce those pre-training solutions, such as on private data, new modalities, or simply for scientific questioning--which is currently extremely demanding computation-wise. We thus propose a novel pre-training strategy for DINOv2 that simultaneously accelerates convergence--and strengthens robustness to common corruptions as a by-product. Our approach involves a frequency filtering curriculum--low-frequency being seen first--and the Gaussian noise patching augmentation. Applied to a ViT-B/16 backbone trained on ImageNet-1K, while pre-training time and FLOPs are reduced by 1.6x and 2.25x, our method still achieves matching robustness in corruption benchmarks (ImageNet-C) and maintains competitive linear probing performance compared with baseline. This dual benefit of efficiency and robustness makes large-scale self-supervised foundation modeling more attainable, while opening the door to novel exploration around data curriculum and augmentation as means to improve self-supervised learning models robustness. The code is available at https://github.com/KevinZ0217/fast_dinov2
Authors: Naeem Paeedeh, Mahardhika Pratama, Imam Mustafa Kamal, Wolfgang Mayer, Jimmy Cao, Ryszard Kowlczyk
Abstract: Despite the progress in cross-domain few-shot learning, a model pre-trained with DINO combined with a prototypical classifier outperforms the latest SOTA methods. A crucial limitation that needs to be overcome is that updating too many parameters of the transformers leads to overfitting due to the scarcity of labeled samples. To address this challenge, we propose a new concept, coalescent projection, as an effective successor to soft prompts. Additionally, we propose a novel pseudo-class generation method, combined with self-supervised transformations, that relies solely on the base domain to prepare the network to encounter unseen samples from different domains. The proposed method exhibits its effectiveness in comprehensive experiments on the extreme domain-shift problem of the BSCD-FSL benchmark. Our code is published at \href{https://github.com/Naeem-Paeedeh/CPLSR}{https://github.com/Naeem-Paeedeh/CPLSR}.
URLs: https://github.com/Naeem-Paeedeh/CPLSR, https://github.com/Naeem-Paeedeh/CPLSR
Authors: Saba Ahmadi, Rabiul Awal, Ankur Sikarwar, Amirhossein Kazemnejad, Ge Ya Luo, Juan A. Rodriguez, Sai Rajeswar, Siva Reddy, Christopher Pal, Benno Krojer, Aishwarya Agrawal
Abstract: While image generation techniques are now capable of producing high-quality images that respect prompts which span multiple sentences, the task of text-guided image editing remains a challenge. Even edit requests that consist of only a few words often fail to be executed correctly. We explore three strategies to enhance performance on a wide range of image editing tasks: supervised fine-tuning (SFT), reinforcement learning (RL), and Chain-of-Thought (CoT) reasoning. In order to study all these components in one consistent framework, we adopt an autoregressive multimodal model that processes textual and visual tokens in a unified manner. We find RL combined with a large multi-modal LLM verifier to be the most effective of these strategies. As a result, we release EARL: Editing with Autoregression and RL, a strong RL-based image editing model that performs competitively on a diverse range of edits compared to strong baselines, despite using much less training data. Thus, EARL pushes the frontier of autoregressive multimodal models on image editing. We release our code, training data, and trained models at https://github.com/mair-lab/EARL.
Authors: Peiran Wang, Yang Liu, Yunfei Lu, Yifeng Cai, Hongbo Chen, Qingyou Yang, Jie Zhang, Jue Hong, Ye Wu
Abstract: Large Language Model (LLM) agents offer a powerful new paradigm for solving various problems by combining natural language reasoning with the execution of external tools. However, their dynamic and non-transparent behavior introduces critical security risks, particularly in the presence of prompt injection attacks. In this work, we propose a novel insight that treats the agent runtime traces as structured programs with analyzable semantics. Thus, we present AgentArmor, a program analysis framework that converts agent traces into graph intermediate representation-based structured program dependency representations (e.g., CFG, DFG, and PDG) and enforces security policies via a type system. AgentArmor consists of three key components: (1) a graph constructor that reconstructs the agent's runtime traces as graph-based intermediate representations with control and data flow described within; (2) a property registry that attaches security-relevant metadata of interacted tools \& data, and (3) a type system that performs static inference and checking over the intermediate representation. By representing agent behavior as structured programs, AgentArmor enables program analysis for sensitive data flow, trust boundaries, and policy violations. We evaluate AgentArmor on the AgentDojo benchmark, the results show that AgentArmor can reduce the ASR to 3\%, with the utility drop only 1\%.
Authors: Soumik Dey, Benjamin Braun, Naveen Ravipati, Hansi Wu, Binbin Li
Abstract: E-commerce sellers are advised to bid on keyphrases to boost their advertising campaigns. These keyphrases must be relevant to prevent irrelevant items from cluttering search systems and to maintain positive seller perception. It is vital that keyphrase suggestions align with seller, search and buyer judgments. Given the challenges in collecting negative feedback in these systems, LLMs have been used as a scalable proxy to human judgments. This paper presents an empirical study on a major ecommerce platform of a distillation framework involving an LLM teacher, a cross-encoder assistant and a bi-encoder Embedding Based Retrieval (EBR) student model, aimed at mitigating click-induced biases in keyphrase recommendations.
Authors: Hubert Baniecki, Maximilian Muschalik, Fabian Fumagalli, Barbara Hammer, Eyke H\"ullermeier, Przemyslaw Biecek
Abstract: Language-image pre-training (LIP) enables the development of vision-language models capable of zero-shot classification, localization, multimodal retrieval, and semantic understanding. Various explanation methods have been proposed to visualize the importance of input image-text pairs on the model's similarity outputs. However, popular saliency maps are limited by capturing only first-order attributions, overlooking the complex cross-modal interactions intrinsic to such encoders. We introduce faithful interaction explanations of LIP models (FIxLIP) as a unified approach to decomposing the similarity in vision-language encoders. FIxLIP is rooted in game theory, where we analyze how using the weighted Banzhaf interaction index offers greater flexibility and improves computational efficiency over the Shapley interaction quantification framework. From a practical perspective, we propose how to naturally extend explanation evaluation metrics, such as the pointing game and area between the insertion/deletion curves, to second-order interaction explanations. Experiments on the MS COCO and ImageNet-1k benchmarks validate that second-order methods, such as FIxLIP, outperform first-order attribution methods. Beyond delivering high-quality explanations, we demonstrate the utility of FIxLIP in comparing different models, e.g. CLIP vs. SigLIP-2.
Authors: Daoze Zhang, Zhanheng Nie, Jianyu Liu, Chenghan Fu, Wanxian Guan, Yuan Gao, Jun Song, Pengjie Wang, Jian Xu, Bo Zheng
Abstract: With the rapid advancement of e-commerce, exploring general representations rather than task-specific ones has attracted increasing research attention. For product understanding, although existing discriminative dual-flow architectures drive progress in this field, they inherently struggle to model the many-to-one alignment between multiple images and texts of products. Therefore, we argue that generative Multimodal Large Language Models (MLLMs) hold significant potential for improving product representation learning. Nevertheless, achieving this goal still remains non-trivial due to several key challenges: the lack of multimodal and aspect-aware modeling modules in typical LLMs; the common presence of background noise in product images; and the absence of a standard benchmark for evaluation. To address these issues, we propose the first generative MLLM-based model named MOON for product representation learning. Our method (1) employs a guided Mixture-of-Experts (MoE) module for targeted modeling of multimodal and aspect-specific product content; (2) effectively detects core semantic regions in product images to mitigate the distraction and interference caused by background noise; and (3) introduces the specialized negative sampling strategy to increase the difficulty and diversity of negative samples. In addition, we release a large-scale multimodal benchmark MBE for various product understanding tasks. Experimentally, our model demonstrates competitive zero-shot performance on both our benchmark and the public dataset, showcasing strong generalization across various downstream tasks, including cross-modal retrieval, product classification, and attribute prediction. Furthermore, the case study and visualization illustrate the effectiveness of MOON for product understanding.
Authors: Promise Osaine Ekpo, Brian La, Thomas Wiener, Saesha Agarwal, Arshia Agrawal, Gonzalo Gonzalez-Pumariega, Lekan P. Molu, Angelique Taylor
Abstract: Fairness in multi-agent reinforcement learning (MARL) is often framed as a workload balance problem, overlooking agent expertise and the structured coordination required in real-world domains. In healthcare, equitable task allocation requires workload balance or expertise alignment to prevent burnout and overuse of highly skilled agents. Workload balance refers to distributing an approximately equal number of subtasks or equalised effort across healthcare workers, regardless of their expertise. We make two contributions to address this problem. First, we propose FairSkillMARL, a framework that defines fairness as the dual objective of workload balance and skill-task alignment. Second, we introduce MARLHospital, a customizable healthcare-inspired environment for modeling team compositions and energy-constrained scheduling impacts on fairness, as no existing simulators are well-suited for this problem. We conducted experiments to compare FairSkillMARL in conjunction with four standard MARL methods, and against two state-of-the-art fairness metrics. Our results suggest that fairness based solely on equal workload might lead to task-skill mismatches and highlight the need for more robust metrics that capture skill-task misalignment. Our work provides tools and a foundation for studying fairness in heterogeneous multi-agent systems where aligning effort with expertise is critical.
Authors: Chanju Park (Swansea University), Biagio Lucini (Queen Mary University of London), Gert Aarts (Swansea University)
Abstract: Hyperparameter tuning is one of the essential steps to guarantee the convergence of machine learning models. We argue that intuition about the optimal choice of hyperparameters for stochastic gradient descent can be obtained by studying a neural network's phase diagram, in which each phase is characterised by distinctive dynamics of the singular values of weight matrices. Taking inspiration from disordered systems, we start from the observation that the loss landscape of a multilayer neural network with mean squared error can be interpreted as a disordered system in feature space, where the learnt features are mapped to soft spin degrees of freedom, the initial variance of the weight matrices is interpreted as the strength of the disorder, and temperature is given by the ratio of the learning rate and the batch size. As the model is trained, three phases can be identified, in which the dynamics of weight matrices is qualitatively different. Employing a Langevin equation for stochastic gradient descent, previously derived using Dyson Brownian motion, we demonstrate that the three dynamical regimes can be classified effectively, providing practical guidance for the choice of hyperparameters of the optimiser.
Authors: Arnau Quera-Bofarull, Nicholas Bishop, Joel Dyer, Daniel Jarne Ornia, Anisoara Calinescu, Doyne Farmer, Michael Wooldridge
Abstract: Agent-based models (ABMs) simulate complex systems by capturing the bottom-up interactions of individual agents comprising the system. Many complex systems of interest, such as epidemics or financial markets, involve thousands or even millions of agents. Consequently, ABMs often become computationally demanding and rely on the calibration of numerous free parameters, which has significantly hindered their widespread adoption. In this paper, we demonstrate that automatic differentiation (AD) techniques can effectively alleviate these computational burdens. By applying AD to ABMs, the gradients of the simulator become readily available, greatly facilitating essential tasks such as calibration and sensitivity analysis. Specifically, we show how AD enables variational inference (VI) techniques for efficient parameter calibration. Our experiments demonstrate substantial performance improvements and computational savings using VI on three prominent ABMs: Axtell's model of firms; Sugarscape; and the SIR epidemiological model. Our approach thus significantly enhances the practicality and scalability of ABMs for studying complex systems.
Authors: Katsuaki Nakano, Reza Fayyazi, Shanchieh Jay Yang, Michael Zuzak
Abstract: Recent advances in Large Language Models (LLMs) have driven interest in automating cybersecurity penetration testing workflows, offering the promise of faster and more consistent vulnerability assessment for enterprise systems. Existing LLM agents for penetration testing primarily rely on self-guided reasoning, which can produce inaccurate or hallucinated procedural steps. As a result, the LLM agent may undertake unproductive actions, such as exploiting unused software libraries or generating cyclical responses that repeat prior tactics. In this work, we propose a guided reasoning pipeline for penetration testing LLM agents that incorporates a deterministic task tree built from the MITRE ATT&CK Matrix, a proven penetration testing kll chain, to constrain the LLM's reaoning process to explicitly defined tactics, techniques, and procedures. This anchors reasoning in proven penetration testing methodologies and filters out ineffective actions by guiding the agent towards more productive attack procedures. To evaluate our approach, we built an automated penetration testing LLM agent using three LLMs (Llama-3-8B, Gemini-1.5, and GPT-4) and applied it to navigate 10 HackTheBox cybersecurity exercises with 103 discrete subtasks representing real-world cyberattack scenarios. Our proposed reasoning pipeline guided the LLM agent through 71.8\%, 72.8\%, and 78.6\% of subtasks using Llama-3-8B, Gemini-1.5, and GPT-4, respectively. Comparatively, the state-of-the-art LLM penetration testing tool using self-guided reasoning completed only 13.5\%, 16.5\%, and 75.7\% of subtasks and required 86.2\%, 118.7\%, and 205.9\% more model queries. This suggests that incorporating a deterministic task tree into LLM reasoning pipelines can enhance the accuracy and efficiency of automated cybersecurity assessments
Authors: Filippo Fabiani, Andrea Simonetto
Abstract: We study data-driven least squares (LS) problems with semidefinite (SD) constraints and derive finite-sample guarantees on the spectrum of their optimal solutions when these constraints are relaxed. In particular, we provide a high confidence bound allowing one to solve a simpler program in place of the full SDLS problem, while ensuring that the eigenvalues of the resulting solution are $\varepsilon$-close of those enforced by the SD constraints. The developed certificate, which consistently shrinks as the number of data increases, turns out to be easy-to-compute, distribution-free, and only requires independent and identically distributed samples. Moreover, when the SDLS is used to learn an unknown quadratic function, we establish bounds on the error between a gradient descent iterate minimizing the surrogate cost obtained with no SD constraints and the true minimizer.
Authors: Amirhossein Yousefiramandi, Ciaran Cooney
Abstract: Transformer-based language models such as BERT have become foundational in NLP, yet their performance degrades in specialized domains like patents, which contain long, technical, and legally structured text. Prior approaches to patent NLP have primarily relied on fine-tuning general-purpose models or domain-adapted variants pretrained with limited data. In this work, we pretrain 3 domain-specific masked language models for patents, using the ModernBERT architecture and a curated corpus of over 60 million patent records. Our approach incorporates architectural optimizations, including FlashAttention, rotary embeddings, and GLU feed-forward layers. We evaluate our models on four downstream patent classification tasks. Our model, ModernBERT-base-PT, consistently outperforms the general-purpose ModernBERT baseline on three out of four datasets and achieves competitive performance with a baseline PatentBERT. Additional experiments with ModernBERT-base-VX and Mosaic-BERT-large demonstrate that scaling the model size and customizing the tokenizer further enhance performance on selected tasks. Notably, all ModernBERT variants retain substantially faster inference over - 3x that of PatentBERT - underscoring their suitability for time-sensitive applications. These results underscore the benefits of domain-specific pretraining and architectural improvements for patent-focused NLP tasks.
Authors: Jan Philipp Schneider, Pratik Singh Bisht, Ilya Chugunov, Andreas Kolb, Michael Moeller, Felix Heide
Abstract: Learning editable high-resolution scene representations for dynamic scenes is an open problem with applications across the domains from autonomous driving to creative editing - the most successful approaches today make a trade-off between editability and supporting scene complexity: neural atlases represent dynamic scenes as two deforming image layers, foreground and background, which are editable in 2D, but break down when multiple objects occlude and interact. In contrast, scene graph models make use of annotated data such as masks and bounding boxes from autonomous-driving datasets to capture complex 3D spatial relationships, but their implicit volumetric node representations are challenging to edit view-consistently. We propose Neural Atlas Graphs (NAGs), a hybrid high-resolution scene representation, where every graph node is a view-dependent neural atlas, facilitating both 2D appearance editing and 3D ordering and positioning of scene elements. Fit at test-time, NAGs achieve state-of-the-art quantitative results on the Waymo Open Dataset - by 5 dB PSNR increase compared to existing methods - and make environmental editing possible in high resolution and visual quality - creating counterfactual driving scenarios with new backgrounds and edited vehicle appearance. We find that the method also generalizes beyond driving scenes and compares favorably - by more than 7 dB in PSNR - to recent matting and video editing baselines on the DAVIS video dataset with a diverse set of human and animal-centric scenes. Project Page: https://princeton-computational-imaging.github.io/nag/
URLs: https://princeton-computational-imaging.github.io/nag/
Authors: Hongkuan Zhou, Lavdim Halilaj, Sebastian Monka, Stefan Schmid, Yuqicheng Zhu, Jingcheng Wu, Nadeem Nazer, Steffen Staab
Abstract: Open-domain visual entity recognition aims to identify and link entities depicted in images to a vast and evolving set of real-world concepts, such as those found in Wikidata. Unlike conventional classification tasks with fixed label sets, it operates under open-set conditions, where most target entities are unseen during training and exhibit long-tail distributions. This makes the task inherently challenging due to limited supervision, high visual ambiguity, and the need for semantic disambiguation. We propose a Knowledge-guided Contrastive Learning (KnowCoL) framework that combines both images and text descriptions into a shared semantic space grounded by structured information from Wikidata. By abstracting visual and textual inputs to a conceptual level, the model leverages entity descriptions, type hierarchies, and relational context to support zero-shot entity recognition. We evaluate our approach on the OVEN benchmark, a large-scale open-domain visual recognition dataset with Wikidata IDs as the label space. Our experiments show that using visual, textual, and structured knowledge greatly improves accuracy, especially for rare and unseen entities. Our smallest model improves the accuracy on unseen entities by 10.5% compared to the state-of-the-art, despite being 35 times smaller.
Authors: Runxi Huang, Mingxuan Yu, Mingyu Tsoi, Xiaomin Ouyang
Abstract: Real-time multimodal inference on resource-constrained edge devices is essential for applications such as autonomous driving, human-computer interaction, and mobile health. However, prior work often overlooks the tight coupling between sensing dynamics and model execution, as well as the complex inter-modality dependencies. In this paper, we propose MMEdge, an new on-device multi-modal inference framework based on pipelined sensing and encoding. Instead of waiting for complete sensor inputs, MMEdge decomposes the entire inference process into a sequence of fine-grained sensing and encoding units, allowing computation to proceed incrementally as data arrive. MMEdge also introduces a lightweight but effective temporal aggregation module that captures rich temporal dynamics across different pipelined units to maintain accuracy performance. Such pipelined design also opens up opportunities for fine-grained cross-modal optimization and early decision-making during inference. To further enhance system performance under resource variability and input data complexity, MMEdge incorporates an adaptive multimodal configuration optimizer that dynamically selects optimal sensing and model configurations for each modality under latency constraints, and a cross-modal speculative skipping mechanism that bypasses future units of slower modalities when early predictions reach sufficient confidence. We evaluate MMEdge using two public multimodal datasets and deploy it on a real-world unmanned aerial vehicle (UAV)-based multimodal testbed. The results show that MMEdge significantly reduces end-to-end latency while maintaining high task accuracy across various system and data dynamics.
Authors: Ali Boudaghi, Hadi Zare
Abstract: Music editing has emerged as an important and practical area of artificial intelligence, with applications ranging from video game and film music production to personalizing existing tracks according to user preferences. However, existing models face significant limitations, such as being restricted to editing synthesized music generated by their own models, requiring highly precise prompts, or necessitating task-specific retraining, thus lacking true zero-shot capability. leveraging recent advances in rectified flow and diffusion transformers, we introduce MusRec, a zero-shot text-to-music editing model capable of performing diverse editing tasks on real-world music efficiently and effectively. Experimental results demonstrate that our approach outperforms existing methods in preserving musical content, structural consistency, and editing fidelity, establishing a strong foundation for controllable music editing in real-world scenarios.
Authors: Qinyi Zhang, Duanyu Feng, Ronghui Han, Yangshuai Wang, Hao Wang
Abstract: Simulating microstructure evolution (MicroEvo) is vital for materials design but demands high numerical accuracy, efficiency, and physical fidelity. Although recent studies on deep learning (DL) offer a promising alternative to traditional solvers, the field lacks standardized benchmarks. Existing studies are flawed due to a lack of comparing specialized MicroEvo DL models with state-of-the-art spatio-temporal architectures, an overemphasis on numerical accuracy over physical fidelity, and a failure to analyze error propagation over time. To address these gaps, we introduce MicroEvoEval, the first comprehensive benchmark for image-based microstructure evolution prediction. We evaluate 14 models, encompassing both domain-specific and general-purpose architectures, across four representative MicroEvo tasks with datasets specifically structured for both short- and long-term assessment. Our multi-faceted evaluation framework goes beyond numerical accuracy and computational cost, incorporating a curated set of structure-preserving metrics to assess physical fidelity. Our extensive evaluations yield several key insights. Notably, we find that modern architectures (e.g., VMamba), not only achieve superior long-term stability and physical fidelity but also operate with an order-of-magnitude greater computational efficiency. The results highlight the necessity of holistic evaluation and identify these modern architectures as a highly promising direction for developing efficient and reliable surrogate models in data-driven materials science.
Authors: Kangning Zhang, Wenxiang Jiao, Kounianhua Du, Yuan Lu, Weiwen Liu, Weinan Zhang, Yong Yu
Abstract: Augmenting Large Language Models (LLMs) with external tools enables them to execute complex, multi-step tasks. However, tool learning is hampered by the static synthetic data pipelines where data generation and model training are executed as two separate, non-interactive processes. This approach fails to adaptively focus on a model's specific weaknesses and allows noisy labels to persist, degrading training efficiency. We introduce LoopTool, a fully automated, model-aware data evolution framework that closes this loop by tightly integrating data synthesis and model training. LoopTool iteratively refines both the data and the model through three synergistic modules: (1) Greedy Capability Probing (GCP) diagnoses the model's mastered and failed capabilities; (2) Judgement-Guided Label Verification (JGLV) uses an open-source judge model to find and correct annotation errors, progressively purifying the dataset; and (3) Error-Driven Data Expansion (EDDE) generates new, challenging samples based on identified failures. This closed-loop process operates within a cost-effective, open-source ecosystem, eliminating dependence on expensive closed-source APIs. Experiments show that our 8B model trained with LoopTool significantly surpasses its 32B data generator and achieves new state-of-the-art results on the BFCL-v3 and ACEBench benchmarks for its scale. Our work demonstrates that closed-loop, self-refining data pipelines can dramatically enhance the tool-use capabilities of LLMs.
Authors: Jonghun Lee, YongKyung Oh, Sungil Kim, Dong-Young Lim
Abstract: Neural Differential Equations (NDEs) excel at modeling continuous-time dynamics, effectively handling challenges such as irregular observations, missing values, and noise. Despite their advantages, NDEs face a fundamental challenge in adopting dropout, a cornerstone of deep learning regularization, making them susceptible to overfitting. To address this research gap, we introduce Continuum Dropout, a universally applicable regularization technique for NDEs built upon the theory of alternating renewal processes. Continuum Dropout formulates the on-off mechanism of dropout as a stochastic process that alternates between active (evolution) and inactive (paused) states in continuous time. This provides a principled approach to prevent overfitting and enhance the generalization capabilities of NDEs. Moreover, Continuum Dropout offers a structured framework to quantify predictive uncertainty via Monte Carlo sampling at test time. Through extensive experiments, we demonstrate that Continuum Dropout outperforms existing regularization methods for NDEs, achieving superior performance on various time series and image classification tasks. It also yields better-calibrated and more trustworthy probability estimates, highlighting its effectiveness for uncertainty-aware modeling.
Authors: Chenghan Fu, Daoze Zhang, Yukang Lin, Zhanheng Nie, Xiang Zhang, Jianyu Liu, Yueran Liu, Wanxian Guan, Pengjie Wang, Jian Xu, Bo Zheng
Abstract: We introduce MOON, our comprehensive set of sustainable iterative practices for multimodal representation learning for e-commerce applications. MOON has already been fully deployed across all stages of Taobao search advertising system, including retrieval, relevance, ranking, and so on. The performance gains are particularly significant on click-through rate (CTR) prediction task, which achieves an overall +20.00% online CTR improvement. Over the past three years, this project has delivered the largest improvement on CTR prediction task and undergone five full-scale iterations. Throughout the exploration and iteration of our MOON, we have accumulated valuable insights and practical experience that we believe will benefit the research community. MOON contains a three-stage training paradigm of "Pretraining, Post-training, and Application", allowing effective integration of multimodal representations with downstream tasks. Notably, to bridge the misalignment between the objectives of multimodal representation learning and downstream training, we define the exchange rate to quantify how effectively improvements in an intermediate metric can translate into downstream gains. Through this analysis, we identify the image-based search recall as a critical intermediate metric guiding the optimization of multimodal models. Over three years and five iterations, MOON has evolved along four critical dimensions: data processing, training strategy, model architecture, and downstream application. The lessons and insights gained through the iterative improvements will also be shared. As part of our exploration into scaling effects in the e-commerce field, we further conduct a systematic study of the scaling laws governing multimodal representation learning, examining multiple factors such as the number of training tokens, negative samples, and the length of user behavior sequences.
Authors: Luthira Abeykoon, Ved Patel, Gawthaman Senthilvelan, Darshan Kasundra
Abstract: Chess has experienced a large increase in viewership since the pandemic, driven largely by the accessibility of online learning platforms. However, no equivalent assistance exists for physical chess games, creating a divide between analog and digital chess experiences. This paper presents CVChess, a deep learning framework for converting chessboard images to Forsyth-Edwards Notation (FEN), which is later input into online chess engines to provide you with the best next move. Our approach employs a convolutional neural network (CNN) with residual layers to perform piece recognition from smartphone camera images. The system processes RGB images of a physical chess board through a multistep process: image preprocessing using the Hough Line Transform for edge detection, projective transform to achieve a top-down board alignment, segmentation into 64 individual squares, and piece classification into 13 classes (6 unique white pieces, 6 unique black pieces and an empty square) using the residual CNN. Residual connections help retain low-level visual features while enabling deeper feature extraction, improving accuracy and stability during training. We train and evaluate our model using the Chess Recognition Dataset (ChessReD), containing 10,800 annotated smartphone images captured under diverse lighting conditions and angles. The resulting classifications are encoded as an FEN string, which can be fed into a chess engine to generate the most optimal move
Authors: Jialiang Shen, Jiyang Zheng, Yunqi Xue, Huajie Chen, Yu Yao, Hui Kang, Ruiqi Liu, Helin Gong, Yang Yang, Dadong Wang, Tongliang Liu
Abstract: With growing concerns over image authenticity and digital safety, the field of AI-generated image (AIGI) detection has progressed rapidly. Yet, most AIGI detectors still struggle under real-world degradations, particularly motion blur, which frequently occurs in handheld photography, fast motion, and compressed video. Such blur distorts fine textures and suppresses high-frequency artifacts, causing severe performance drops in real-world settings. We address this limitation with a blur-robust AIGI detection framework based on teacher-student knowledge distillation. A high-capacity teacher (DINOv3), trained on clean (i.e., sharp) images, provides stable and semantically rich representations that serve as a reference for learning. By freezing the teacher to maintain its generalization ability, we distill its feature and logit responses from sharp images to a student trained on blurred counterparts, enabling the student to produce consistent representations under motion degradation. Extensive experiments benchmarks show that our method achieves state-of-the-art performance under both motion-blurred and clean conditions, demonstrating improved generalization and real-world applicability. Source codes will be released at: https://github.com/JiaLiangShen/Dino-Detect-for-blur-robust-AIGC-Detection.
URLs: https://github.com/JiaLiangShen/Dino-Detect-for-blur-robust-AIGC-Detection.
Authors: Yehonatan Elisha, Seffi Cohen, Oren Barkan, Noam Koenigstein
Abstract: Saliency maps are widely used for visual explanations in deep learning, but a fundamental lack of consensus persists regarding their intended purpose and alignment with diverse user queries. This ambiguity hinders the effective evaluation and practical utility of explanation methods. We address this gap by introducing the Reference-Frame $\times$ Granularity (RFxG) taxonomy, a principled conceptual framework that organizes saliency explanations along two essential axes:Reference-Frame: Distinguishing between pointwise ("Why this prediction?") and contrastive ("Why this and not an alternative?") explanations. Granularity: Ranging from fine-grained class-level (e.g., "Why Husky?") to coarse-grained group-level (e.g., "Why Dog?") interpretations. Using the RFxG lens, we demonstrate critical limitations in existing evaluation metrics, which overwhelmingly prioritize pointwise faithfulness while neglecting contrastive reasoning and semantic granularity. To systematically assess explanation quality across both RFxG dimensions, we propose four novel faithfulness metrics. Our comprehensive evaluation framework applies these metrics to ten state-of-the-art saliency methods, four model architectures, and three datasets. By advocating a shift toward user-intent-driven evaluation, our work provides both the conceptual foundation and the practical tools necessary to develop visual explanations that are not only faithful to the underlying model behavior but are also meaningfully aligned with the complexity of human understanding and inquiry.
Authors: Zhenyu Chen, Yuguo Shao, Zhengwei Liu, Zhaohui Wei
Abstract: Quantum algorithms based on parameterized quantum circuits (PQCs) have enabled a wide range of applications on near-term quantum devices. However, existing PQC architectures face several challenges, among which the ``barren plateaus" phenomenon is particularly prominent. In such cases, the loss function concentrates exponentially with increasing system size, thereby hindering effective parameter optimization. To address this challenge, we propose a general and hardware-efficient method for eliminating barren plateaus in an arbitrary PQC. Specifically, our approach achieves this by inserting a layer of easily implementable quantum channels into the original PQC, each channel requiring only one ancilla qubit and four additional gates, yielding a modified PQC (MPQC) that is provably at least as expressive as the original PQC and, under mild assumptions, is guaranteed to be free from barren plateaus. Furthermore, by appropriately adjusting the structure of MPQCs, we rigorously prove that any parameter in the original PQC can be made trainable. Importantly, the absence of barren plateaus in MPQCs is robust against realistic noise, making our approach directly applicable to current noisy intermediate-scale quantum (NISQ) hardware. Numerically, we demonstrate the practicality of our method by modifying a commonly used PQC for thermal-state preparation. The results show that {barren plateaus are effectively eliminated} in this class of circuits with up to 100 qubits and 2400 layers, whereas the original ansatz suffers from severe gradient vanishing.
Authors: Davoud Shariat Panah, Alessandro Ragano, Dan Barry, Jan Skoglund, Andrew Hines
Abstract: This study presents a systematic evaluation of time-frequency feature design for binaural sound source localization (SSL), focusing on how feature selection influences model performance across diverse conditions. We investigate the performance of a convolutional neural network (CNN) model using various combinations of amplitude-based features (magnitude spectrogram, interaural level difference - ILD) and phase-based features (phase spectrogram, interaural phase difference - IPD). Evaluations on in-domain and out-of-domain data with mismatched head-related transfer functions (HRTFs) reveal that carefully chosen feature combinations often outperform increases in model complexity. While two-feature sets such as ILD + IPD are sufficient for in-domain SSL, generalization to diverse content requires richer inputs combining channel spectrograms with both ILD and IPD. Using the optimal feature sets, our low-complexity CNN model achieves competitive performance. Our findings underscore the importance of feature design in binaural SSL and provide practical guidance for both domain-specific and general-purpose localization.