Authors: Yifu Luo, Yongzhe Chang, Xueqian Wang
Abstract: Diffusion probability models have shown significant promise in offline reinforcement learning by directly modeling trajectory sequences. However, existing approaches primarily focus on time-domain features while overlooking frequency-domain features, leading to frequency shift and degraded performance according to our observation. In this paper, we investigate the RL problem from a new perspective of the frequency domain. We first observe that time-domain-only approaches inadvertently introduce shifts in the low-frequency components of the frequency domain, which results in trajectory instability and degraded performance. To address this issue, we propose Wavelet Fourier Diffuser (WFDiffuser), a novel diffusion-based RL framework that integrates Discrete Wavelet Transform to decompose trajectories into low- and high-frequency components. To further enhance diffusion modeling for each component, WFDiffuser employs Short-Time Fourier Transform and cross attention mechanisms to extract frequency-domain features and facilitate cross-frequency interaction. Extensive experiment results on the D4RL benchmark demonstrate that WFDiffuser effectively mitigates frequency shift, leading to smoother, more stable trajectories and improved decision-making performance over existing methods.
Authors: Mashkhal Abdalwahid Sidiq, Yimamu Kirubel Wondaferew
Abstract: In this paper, we focused on using deep learning methods for detecting money laundering in financial transaction networks, in order to demonstrate that it can be used as a complement or instead of the more commonly used rule-based systems and conventional Anti-Money Laundering (AML) systems. The paper explores the pivotal role played by Anti-Money Laundering (AML) activities in the global financial industry. It underscores the drawbacks of conventional AML systems, which exhibit high rates of false positives and lack the sophistication to uncover intricate money laundering schemes. To tackle these challenges, the paper proposes an advanced AML system that capitalizes on link analysis using deep learning techniques. At the heart of this system lies the utilization of centrality algorithms like Degree Centrality, Closeness Centrality, Betweenness Centrality, and PageRank. These algorithms enhance the system's capability to identify suspicious activities by examining the influence and interconnections within networks of financial transactions. The significance of Anti-Money Laundering (AML) efforts within the global financial sector is discussed in this paper. It highlights the limitations of traditional AML systems. The results showed the practicality and superiority of the new implementation of the GCN model, which is a preferable method for connectively structured data, meaning that a transaction or account is analyzed in the context of its financial environment. In addition, the paper delves into the prospects of Anti-Money Laundering (AML) efforts, proposing the integration of emerging technologies such as deep learning and centrality algorithms. This integration holds promise for enhancing the effectiveness of AML systems by refining their capabilities.
Authors: Benedikt W. Hosp
Abstract: Feature attribution is essential for interpreting deep learning models, particularly in time-series domains such as healthcare, biometrics, and human-AI interaction. However, standard attribution methods, such as Integrated Gradients or SHAP, are computationally intensive and not well-suited for real-time applications. We present DeepACTIF, a lightweight and architecture-aware feature attribution method that leverages internal activations of sequence models to estimate feature importance efficiently. Focusing on LSTM-based networks, we introduce an inverse-weighted aggregation scheme that emphasises stability and magnitude of activations across time steps. Our evaluation across three biometric gaze datasets shows that DeepACTIF not only preserves predictive performance under severe feature reduction (top 10% of features) but also significantly outperforms established methods, including SHAP, IG, and DeepLIFT, in terms of both accuracy and statistical robustness. Using Wilcoxon signed-rank tests and effect size analysis, we demonstrate that DeepACTIF yields more informative feature rankings with significantly lower error across all top-k conditions (10 - 40%). Our experiments demonstrate that DeepACTIF not only reduces computation time and memory usage by orders of magnitude but also preserves model accuracy when using only top-ranked features. That makes DeepACTIF a viable solution for real-time interpretability on edge devices such as mobile XR headsets or embedded health monitors.
Authors: Zhuqi Wang, Qinghe Zhang, Zhuopei Cheng
Abstract: Credit card fraud is assuming growing proportions as a major threat to the financial position of American household, leading to unpredictable changes in household economic behavior. To solve this problem, in this paper, a new hybrid analysis method is presented by using the Enhanced ANFIS. The model proposes several advances of the conventional ANFIS framework and employs a multi-resolution wavelet decomposition module and a temporal attention mechanism. The model performs discrete wavelet transformations on historical transaction data and macroeconomic indicators to generate localized economic shock signals. The transformed features are then fed into a deep fuzzy rule library which is based on Takagi-Sugeno fuzzy rules with adaptive Gaussian membership functions. The model proposes a temporal attention encoder that adaptively assigns weights to multi-scale economic behavior patterns, increasing the effectiveness of relevance assessment in the fuzzy inference stage and enhancing the capture of long-term temporal dependencies and anomalies caused by fraudulent activities. The proposed method differs from classical ANFIS which has fixed input-output relations since it integrates fuzzy rule activation with the wavelet basis selection and the temporal correlation weights via a modular training procedure. Experimental results show that the RMSE was reduced by 17.8% compared with local neuro-fuzzy models and conventional LSTM models.
Authors: Buhe Li, Berkay Kaplan, Maksym Lazirko, Aleksandr Kogan
Abstract: This study investigates the effectiveness of unsupervised outlier detection methods in audit analytics, utilizing USA spending data from the U.S. Department of Health and Human Services (DHHS) as a case example. We employ and compare multiple outlier detection algorithms, including Histogram-based Outlier Score (HBOS), Robust Principal Component Analysis (PCA), Minimum Covariance Determinant (MCD), and K-Nearest Neighbors (KNN) to identify anomalies in federal spending patterns. The research addresses the growing need for efficient and accurate anomaly detection in large-scale governmental datasets, where traditional auditing methods may fall short. Our methodology involves data preparation, algorithm implementation, and performance evaluation using precision, recall, and F1 scores. Results indicate that a hybrid approach, combining multiple detection strategies, enhances the robustness and accuracy of outlier identification in complex financial data. This study contributes to the field of audit analytics by providing insights into the comparative effectiveness of various outlier detection models and demonstrating the potential of unsupervised learning techniques in improving audit quality and efficiency. The findings have implications for auditors, policymakers, and researchers seeking to leverage advanced analytics in governmental financial oversight and risk management.
Authors: Zuzanna Dubanowska, Maciej \.Zelaszczyk, Micha{\l} Brzozowski, Paolo Mandica, Micha{\l} Karpowicz
Abstract: We critically assess the efficacy of the current SOTA in hallucination detection and find that its performance on the RAGTruth dataset is largely driven by a spurious correlation with data. Controlling for this effect, state-of-the-art performs no better than supervised linear probes, while requiring extensive hyperparameter tuning across datasets. Out-of-distribution generalization is currently out of reach, with all of the analyzed methods performing close to random. We propose a set of guidelines for hallucination detection and its evaluation.
Authors: Mridul Sharma (Faculty of Dental Medicine and Oral Health Sciences, McGill University, Montreal, Canada), Adeetya Patel (Faculty of Dental Medicine and Oral Health Sciences, McGill University, Montreal, Canada), Zaneta D' Souza (Faculty of Dental Medicine and Oral Health Sciences, McGill University, Montreal, Canada), Samira Abbasgholizadeh Rahimi (Faculty of Dental Medicine and Oral Health Sciences, McGill University, Montreal, Canada, Mila-Quebec Artificial Intelligence Institute, Montreal, Canada), Siva Reddy (School of Computer Science, McGill University, Montreal, Canada, Mila-Quebec Artificial Intelligence Institute, Montreal, Canada), Sreenath Madathil (Faculty of Dental Medicine and Oral Health Sciences, McGill University, Montreal, Canada)
Abstract: Despite their widespread applications, Large Language Models (LLMs) often struggle to express uncertainty, posing a challenge for reliable deployment in high stakes and safety critical domains like clinical diagnostics. Existing standard baseline methods such as model logits and elicited probabilities produce overconfident and poorly calibrated estimates. In this work, we propose Approximate Bayesian Computation (ABC), a likelihood-free Bayesian inference, based approach that treats LLMs as a stochastic simulator to infer posterior distributions over predictive probabilities. We evaluate our ABC approach on two clinically relevant benchmarks: a synthetic oral lesion diagnosis dataset and the publicly available GretelAI symptom-to-diagnosis dataset. Compared to standard baselines, our approach improves accuracy by up to 46.9\%, reduces Brier scores by 74.4\%, and enhances calibration as measured by Expected Calibration Error (ECE) and predictive entropy.
Authors: Matthew Grofsky
Abstract: We address temporal failures in RAG systems using two methods on cybersecurity data. A simple recency prior achieved an accuracy of 1.00 on freshness tasks. In contrast, a clustering heuristic for topic evolution failed (0.08 F1-score), showing trend detection requires methods beyond simple heuristics.
Authors: Returaj Burnwal, Hriday Mehta, Nirav Pravinbhai Bhatt, Balaraman Ravindran
Abstract: Imitation Learning (IL) algorithms offer an efficient way to train an agent by mimicking an expert's behavior without requiring a reward function. IL algorithms often necessitate access to state and action information from expert demonstrations. Although expert actions can provide detailed guidance, requiring such action information may prove impractical for real-world applications where expert actions are difficult to obtain. To address this limitation, the concept of learning from observation (LfO) or state-only imitation learning (SOIL) has recently gained attention, wherein the imitator only has access to expert state visitation information. In this paper, we present a framework for LfO and use it to survey and classify existing LfO methods in terms of their trajectory construction, assumptions and algorithm's design choices. This survey also draws connections between several related fields like offline RL, model-based RL and hierarchical RL. Finally, we use our framework to identify open problems and suggest future research directions.
Authors: Axel Marmoret, Reda Bensaid, Jonathan Lys, Vincent Gripon, Fran\c{c}ois Leduc-Primeau
Abstract: Low-Rank Adaptation (LoRA) is widely used to efficiently adapt Transformers by adding trainable low-rank matrices to attention projections. While effective, these matrices are considered independent for each attention projection (Query, Key, and Value) and each layer. Recent extensions have considered joint, tensor-based adaptations, but only in limited forms and without a systematic framework. We introduce TensLoRA, a unified framework that aggregates LoRA updates into higher-order tensors and models a broad family of tensor-based low-rank adaptations. Our formulation generalizes existing tensor-based methods and enables mode-specific compression rates, allowing parameter budgets to be tailored according to the modality and task. Experiments on vision and language benchmarks reveal that the tensor construction directly impacts performance, sometimes better than standard LoRA under similar parameter counts.
Authors: Sahil Tyagi, Andrei Cozma, Olivera Kotevska, Feiyi Wang
Abstract: Federated Learning (FL) is critical for edge and High Performance Computing (HPC) where data is not centralized and privacy is crucial. We present OmniFed, a modular framework designed around decoupling and clear separation of concerns for configuration, orchestration, communication, and training logic. Its architecture supports configuration-driven prototyping and code-level override-what-you-need customization. We also support different topologies, mixed communication protocols within a single deployment, and popular training algorithms. It also offers optional privacy mechanisms including Differential Privacy (DP), Homomorphic Encryption (HE), and Secure Aggregation (SA), as well as compression strategies. These capabilities are exposed through well-defined extension points, allowing users to customize topology and orchestration, learning logic, and privacy/compression plugins, all while preserving the integrity of the core system. We evaluate multiple models and algorithms to measure various performance metrics. By unifying topology configuration, mixed-protocol communication, and pluggable modules in one stack, OmniFed streamlines FL deployment across heterogeneous environments. Github repository is available at https://github.com/at-aaims/OmniFed.
Authors: Kuiye Ding, Fanda Fan, Chunyi Hou, Zheya Wang, Lei Wang, Zhengxin Yang, Jianfeng Zhan
Abstract: Multivariate time series forecasting is essential in domains such as finance, transportation, climate, and energy. However, existing patch-based methods typically adopt fixed-length segmentation, overlooking the heterogeneity of local temporal dynamics and the decoding heterogeneity of forecasting. Such designs lose details in information-dense regions, introduce redundancy in stable segments, and fail to capture the distinct complexities of short-term and long-term horizons. We propose TimeMosaic, a forecasting framework that aims to address temporal heterogeneity. TimeMosaic employs adaptive patch embedding to dynamically adjust granularity according to local information density, balancing motif reuse with structural clarity while preserving temporal continuity. In addition, it introduces segment-wise decoding that treats each prediction horizon as a related subtask and adapts to horizon-specific difficulty and information requirements, rather than applying a single uniform decoder. Extensive evaluations on benchmark datasets demonstrate that TimeMosaic delivers consistent improvements over existing methods, and our model trained on the large-scale corpus with 321 billion observations achieves performance competitive with state-of-the-art TSFMs.
Authors: Obu-Amoah Ampomah, Edmund Agyemang, Kofi Acheampong, Louis Agyekum
Abstract: This study examines credit default prediction by comparing three techniques, namely SMOTE, SMOTE-Tomek, and ADASYN, that are commonly used to address the class imbalance problem in credit default situations. Recognizing that credit default datasets are typically skewed, with defaulters comprising a much smaller proportion than non-defaulters, we began our analysis by evaluating machine learning (ML) models on the imbalanced data without any resampling to establish baseline performance. These baseline results provide a reference point for understanding the impact of subsequent balancing methods. In addition to traditional classifiers such as Naive Bayes and K-Nearest Neighbors (KNN), our study also explores the suitability of advanced ensemble boosting algorithms, including Extreme Gradient Boosting (XGBoost), AdaBoost, Gradient Boosting Machines (GBM), and Light GBM for credit default prediction using Boruta feature selection and DBSCAN-based outlier detection, both before and after resampling. A real-world credit default data set sourced from the University of Cleveland ML Repository was used to build ML classifiers, and their performances were tested. The criteria chosen to measure model performance are the area under the receiver operating characteristic curve (ROC-AUC), area under the precision-recall curve (PR-AUC), G-mean, and F1-scores. The results from this empirical study indicate that the Boruta+DBSCAN+SMOTE-Tomek+GBM classifier outperformed the other ML models (F1-score: 82.56%, G-mean: 82.98%, ROC-AUC: 90.90%, PR-AUC: 91.85%) in a credit default context. The findings establish a foundation for future progress in creating more resilient and adaptive credit default systems, which will be essential as credit-based transactions continue to rise worldwide.
Authors: Andreas Lebedev, Abhinav Das, Sven Pappert, Stephan Schl\"uter
Abstract: Precise probabilistic forecasts are fundamental for energy risk management, and there is a wide range of both statistical and machine learning models for this purpose. Inherent to these probabilistic models is some form of uncertainty quantification. However, most models do not capture the full extent of uncertainty, which arises not only from the data itself but also from model and distributional choices. In this study, we examine uncertainty quantification in state-of-the-art statistical and deep learning probabilistic forecasting models for electricity price forecasting in the German market. In particular, we consider deep distributional neural networks (DDNNs) and augment them with an ensemble approach, Monte Carlo (MC) dropout, and conformal prediction to account for model uncertainty. Additionally, we consider the LASSO-estimated autoregressive (LEAR) approach combined with quantile regression averaging (QRA), generalized autoregressive conditional heteroskedasticity (GARCH), and conformal prediction. Across a range of performance metrics, we find that the LEAR-based models perform well in terms of probabilistic forecasting, irrespective of the uncertainty quantification method. Furthermore, we find that DDNNs benefit from incorporating both data and model uncertainty, improving both point and probabilistic forecasting. Uncertainty itself appears to be best captured by the models using conformal prediction. Overall, our extensive study shows that all models under consideration perform competitively. However, their relative performance depends on the choice of metrics for point and probabilistic forecasting.
Authors: Birk Torpmann-Hagen, P{\aa}l Halvorsen, Michael A. Riegler, Dag Johansen
Abstract: Despite achieving excellent performance on benchmarks, deep neural networks often underperform in real-world deployment due to sensitivity to minor, often imperceptible shifts in input data, known as distributional shifts. These shifts are common in practical scenarios but are rarely accounted for during evaluation, leading to inflated performance metrics. To address this gap, we propose a novel methodology for the verification, evaluation, and risk assessment of deep learning systems. Our approach explicitly models the incidence of distributional shifts at runtime by estimating their probability from outputs of out-of-distribution detectors. We combine these estimates with conditional probabilities of network correctness, structuring them in a binary tree. By traversing this tree, we can compute credible and precise estimates of network accuracy. We assess our approach on five different datasets, with which we simulate deployment conditions characterized by differing frequencies of distributional shift. Our approach consistently outperforms conventional evaluation, with accuracy estimation errors typically ranging between 0.01 and 0.1. We further showcase the potential of our approach on a medical segmentation benchmark, wherein we apply our methods towards risk assessment by associating costs with tree nodes, informing cost-benefit analyses and value-judgments. Ultimately, our approach offers a robust framework for improving the reliability and trustworthiness of deep learning systems, particularly in safety-critical applications, by providing more accurate performance estimates and actionable risk assessments.
Authors: Kin G. Olivares, Malcolm Wolff, Tatiana Konstantinova, Shankar Ramasubramanian, Andrew Gordon Wilson, Andres Potapczynski, Willa Potosnak, Mengfei Cao, Boris Oreshkin, 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: Javier Castro, Benjamin Gess
Abstract: Physics-Informed Neural Networks (PINNs) are a class of deep learning models aiming to approximate solutions of PDEs by training neural networks to minimize the residual of the equation. Focusing on non-equilibrium fluctuating systems, we propose a physically informed choice of penalization that is consistent with the underlying fluctuation structure, as characterized by a large deviations principle. This approach yields a novel formulation of PINNs in which the penalty term is chosen to penalize improbable deviations, rather than being selected heuristically. The resulting thermodynamically consistent extension of PINNs, termed THINNs, is subsequently analyzed by establishing analytical a posteriori estimates, and providing empirical comparisons to established penalization strategies.
Authors: Hunjae Lee, Corey Clark
Abstract: Variable count is among the main scalability bottlenecks for transformer modeling in multivariate time series (MTS) data. On top of this, a growing consensus in the field points to indiscriminate inter-variable mixing as a potential source of noise-accumulation and performance degradation. This is likely exacerbated by sparsity of informative signals characteristic of many MTS systems coupled with representational misalignment stemming from indiscriminate information mixing between (heterogeneous) variables. While scalability and performance are often seen as competing interests in transformer design, we show that both can be improved simultaneously in MTS by strategically constraining the representational capacity of inter-variable mixing. Our proposed method, transformer with Delegate Token Attention (DELTAformer), constrains inter-variable modeling through what we call delegate tokens which are then used to perform full, unconstrained, inter-temporal modeling. Delegate tokens act as an implicit regularizer that forces the model to be highly selective about what inter-variable information is allowed to propagate through the network. Our results show that DELTAformer scales linearly with variable-count while actually outperforming standard transformers, achieving state-of-the-art performance across benchmarks and baselines. In addition, DELTAformer can focus on relevant signals better than standard transformers in noisy MTS environments and overall exhibit superior noise-resilience. Overall, results across various experiments confirm that by aligning our model design to leverage domain-specific challenges in MTS to our advantage, DELTAformer can simultaneously achieve linear scaling while actually improving its performance against standard, quadratic transformers.
Authors: Trung Nguyen Thanh, Huyen Giang Thi Thu, Tai Le Quy, Ha-Bang Ban
Abstract: Counterfactual explanation (CE) is a widely used post-hoc method that provides individuals with actionable changes to alter an unfavorable prediction from a machine learning model. Plausible CE methods improve realism by considering data distribution characteristics, but their optimization models introduce a large number of constraints, leading to high computational cost. In this work, we revisit the DACE framework and propose a refined Mixed-Integer Linear Programming (MILP) formulation that significantly reduces the number of constraints in the local outlier factor (LOF) objective component. We also apply the method to a linear SVM classifier with standard scaler. The experimental results show that our approach achieves faster solving times while maintaining explanation quality. These results demonstrate the promise of more efficient LOF modeling in counterfactual explanation and data science applications.
Authors: Mohan Guo (Faculty of Science, University of Amsterdam), Cong Liu (AMLab, University of Amsterdam), Patrick Forr\'e (AMLab, University of Amsterdam)
Abstract: Recent methods for molecular generation face a trade-off: they either enforce strict equivariance with costly architectures or relax it to gain scalability and flexibility. We propose a frame-based diffusion paradigm that achieves deterministic E(3)-equivariance while decoupling symmetry handling from the backbone. Building on this paradigm, we investigate three variants: Global Frame Diffusion (GFD), which assigns a shared molecular frame; Local Frame Diffusion (LFD), which constructs node-specific frames and benefits from additional alignment constraints; and Invariant Frame Diffusion (IFD), which relies on pre-canonicalized invariant representations. To enhance expressivity, we further utilize EdgeDiT, a Diffusion Transformer with edge-aware attention. On the QM9 dataset, GFD with EdgeDiT achieves state-of-the-art performance, with a test NLL of -137.97 at standard scale and -141.85 at double scale, alongside atom stability of 98.98%, and molecular stability of 90.51%. These results surpass all equivariant baselines while maintaining high validity and uniqueness and nearly 2x faster sampling compared to EDM. Altogether, our study establishes frame-based diffusion as a scalable, flexible, and physically grounded paradigm for molecular generation, highlighting the critical role of global structure preservation.
Authors: Ali Baheri, Lars Lindemann
Abstract: Metriplectic conditional flow matching (MCFM) learns dissipative dynamics without violating first principles. Neural surrogates often inject energy and destabilize long-horizon rollouts; MCFM instead builds the conservative-dissipative split into both the vector field and a structure preserving sampler. MCFM trains via conditional flow matching on short transitions, avoiding long rollout adjoints. In inference, a Strang-prox scheme alternates a symplectic update with a proximal metric step, ensuring discrete energy decay; an optional projection enforces strict decay when a trusted energy is available. We provide continuous and discrete time guarantees linking this parameterization and sampler to conservation, monotonic dissipation, and stable rollouts. On a controlled mechanical benchmark, MCFM yields phase portraits closer to ground truth and markedly fewer energy-increase and positive energy rate events than an equally expressive unconstrained neural flow, while matching terminal distributional fit.
Authors: Zongyue Li, Xiao Han, Yusong Li, Niklas Strauss, Matthias Schubert
Abstract: Diffusion-based world models have demonstrated strong capabilities in synthesizing realistic long-horizon trajectories for offline reinforcement learning (RL). However, many existing methods do not directly generate actions alongside states and rewards, limiting their compatibility with standard value-based offline RL algorithms that rely on one-step temporal difference (TD) learning. While prior work has explored joint modeling of states, rewards, and actions to address this issue, such formulations often lead to increased training complexity and reduced performance in practice. We propose \textbf{DAWM}, a diffusion-based world model that generates future state-reward trajectories conditioned on the current state, action, and return-to-go, paired with an inverse dynamics model (IDM) for efficient action inference. This modular design produces complete synthetic transitions suitable for one-step TD-based offline RL, enabling effective and computationally efficient training. Empirically, we show that conservative offline RL algorithms such as TD3BC and IQL benefit significantly from training on these augmented trajectories, consistently outperforming prior diffusion-based baselines across multiple tasks in the D4RL benchmark.
Authors: Yi Ren
Abstract: This thesis explores how deep learning models learn over time, using ideas inspired by force analysis. Specifically, we zoom in on the model's training procedure to see how one training example affects another during learning, like analyzing how forces move objects. We break this influence into two parts: how similar the two examples are, and how strong the updating force is. This framework helps us understand a wide range of the model's behaviors in different real systems. For example, it explains why certain examples have non-trivial learning paths, why (and why not) some LLM finetuning methods work, and why simpler, more structured patterns tend to be learned more easily. We apply this approach to various learning tasks and uncover new strategies for improving model training. While the method is still developing, it offers a new way to interpret models' behaviors systematically.
Authors: Alexander Ho, Sukyeong Lee, Francis T. F. Tsai
Abstract: We introduce FragAtlas-62M, a specialized foundation model trained on the largest fragment dataset to date. Built on the complete ZINC-22 fragment subset comprising over 62 million molecules, it achieves unprecedented coverage of fragment chemical space. Our GPT-2 based model (42.7M parameters) generates 99.90% chemically valid fragments. Validation across 12 descriptors and three fingerprint methods shows generated fragments closely match the training distribution (all effect sizes < 0.4). The model retains 53.6% of known ZINC fragments while producing 22% novel structures with practical relevance. We release FragAtlas-62M with training code, preprocessed data, documentation, and model weights to accelerate adoption.
Authors: Jichi Wang, Eduardo D. Sontag, Domitilla Del Vecchio
Abstract: In several applications, including in synthetic biology, one often has input/output data on a system composed of many modules, and although the modules' input/output functions and signals may be unknown, knowledge of the composition architecture can significantly reduce the amount of training data required to learn the system's input/output mapping. Learning the modules' input/output functions is also necessary for designing new systems from different composition architectures. Here, we propose a modular learning framework, which incorporates prior knowledge of the system's compositional structure to (a) identify the composing modules' input/output functions from the system's input/output data and (b) achieve this by using a reduced amount of data compared to what would be required without knowledge of the compositional structure. To achieve this, we introduce the notion of modular identifiability, which allows recovery of modules' input/output functions from a subset of the system's input/output data, and provide theoretical guarantees on a class of systems motivated by genetic circuits. We demonstrate the theory on computational studies showing that a neural network (NNET) that accounts for the compositional structure can learn the composing modules' input/output functions and predict the system's output on inputs outside of the training set distribution. By contrast, a neural network that is agnostic of the structure is unable to predict on inputs that fall outside of the training set distribution. By reducing the need for experimental data and allowing module identification, this framework offers the potential to ease the design of synthetic biological circuits and of multi-module systems more generally.
Authors: Jiayi Xin, Aniruddh Raghu, Nick Bhattacharya, Adam Carr, Melanie Montgomery, Hunter Elliott
Abstract: Modern therapeutic antibody design often involves composing multi-part assemblages of individual functional domains, each of which may be derived from a different source or engineered independently. While these complex formats can expand disease applicability and improve safety, they present a significant engineering challenge: the function and stability of individual domains are not guaranteed in the novel format, and the entire molecule may no longer be synthesizable. To address these challenges, we develop a machine learning framework to predict "reformatting success" -- whether converting an antibody from one format to another will succeed or not. Our framework incorporates both antibody sequence and structural context, incorporating an evaluation protocol that reflects realistic deployment scenarios. In experiments on a real-world antibody reformatting dataset, we find the surprising result that large pretrained protein language models (PLMs) fail to outperform simple, domain-tailored, multimodal representations. This is particularly evident in the most difficult evaluation setting, where we test model generalization to a new starting antibody. In this challenging "new antibody, no data" scenario, our best multimodal model achieves high predictive accuracy, enabling prioritization of promising candidates and reducing wasted experimental effort.
Authors: Juan Manuel Perez, Kevin Garcia, Brooklyn Berry, Dongjin Song, Yifeng Gao
Abstract: Indexing time series by creating compact binary representations is a fundamental task in time series data mining. Recently, deep learning-based hashing methods have proven effective for indexing time series based on semantic meaning rather than just raw similarity. The purpose of deep hashing is to map samples with the same semantic meaning to identical binary hash codes, enabling more efficient search and retrieval. Unlike other supervised representation learning methods, supervised deep hashing requires a discretization step to convert real-valued representations into binary codes, but this can induce significant information loss. In this paper, we propose a von Mises-Fisher (vMF) hashing loss. The proposed deep hashing model maps data to an M-dimensional hyperspherical space to effectively reduce information loss and models each data class as points following distinct vMF distributions. The designed loss aims to maximize the separation between each modeled vMF distribution to provide a better way to maximize the margin between each semantically different data sample. Experimental results show that our method outperforms existing baselines. The implementation is publicly available at https://github.com/jmpq97/vmf-hashing
Authors: Peng Lu, Jerry Huang, Qiuhao Zeng, Xinyu Wang, Boxing Wang, Philippe Langlais, Yufei Cui
Abstract: The quadratic complexity of the attention mechanism in Transformer models has motivated the development of alternative architectures with sub-quadratic scaling, such as state-space models. Among these, Mamba has emerged as a leading architecture, achieving state-of-the-art results across a range of language modeling tasks. However, Mamba's performance significantly deteriorates when applied to contexts longer than those seen during pre-training, revealing a sharp sensitivity to context length extension. Through detailed analysis, we attribute this limitation to the out-of-distribution behaviour of its state-space dynamics, particularly within the parameterization of the state transition matrix $\mathbf{A}$. Unlike recent works which attribute this sensitivity to the vanished accumulation of discretization time steps, $\exp(-\sum_{t=1}^N\Delta_t)$, we establish a connection between state convergence behavior as the input length approaches infinity and the spectrum of the transition matrix $\mathbf{A}$, offering a well-founded explanation of its role in length extension. Next, to overcome this challenge, we propose an approach that applies spectrum scaling to pre-trained Mamba models to enable robust long-context generalization by selectively modulating the spectrum of $\mathbf{A}$ matrices in each layer. We show that this can significantly improve performance in settings where simply modulating $\Delta_t$ fails, validating our insights and providing avenues for better length generalization of state-space models with structured transition matrices.
Authors: MohammadReza EskandariNasab, Shah Muhammad Hamdi, Soukaina Filali Boubrahimi
Abstract: Generating high-quality synthetic time series is a fundamental yet challenging task across domains such as forecasting and anomaly detection, where real data can be scarce, noisy, or costly to collect. Unlike static data generation, synthesizing time series requires modeling both the marginal distribution of observations and the conditional temporal dependencies that govern sequential dynamics. We propose TIMED, a unified generative framework that integrates a denoising diffusion probabilistic model (DDPM) to capture global structure via a forward-reverse diffusion process, a supervisor network trained with teacher forcing to learn autoregressive dependencies through next-step prediction, and a Wasserstein critic that provides adversarial feedback to ensure temporal smoothness and fidelity. To further align the real and synthetic distributions in feature space, TIMED incorporates a Maximum Mean Discrepancy (MMD) loss, promoting both diversity and sample quality. All components are built using masked attention architectures optimized for sequence modeling and are trained jointly to effectively capture both unconditional and conditional aspects of time series data. Experimental results across diverse multivariate time series benchmarks demonstrate that TIMED generates more realistic and temporally coherent sequences than state-of-the-art generative models.
Authors: Hongyi Chen, Xiucheng Li, Xinyang Chen, Yun Cheng, Jing Li, Kehai Chen, Liqiang Nie
Abstract: Global Station Weather Forecasting (GSWF) is a key meteorological research area, critical to energy, aviation, and agriculture. Existing time series forecasting methods often ignore or unidirectionally model spatial correlation when conducting large-scale global station forecasting. This contradicts the intrinsic nature underlying observations of the global weather system, limiting forecast performance. To address this, we propose a novel Spatial Structured Attention Block in this paper. It partitions the spatial graph into a set of subgraphs and instantiates Intra-subgraph Attention to learn local spatial correlation within each subgraph, and aggregates nodes into subgraph representations for message passing among the subgraphs via Inter-subgraph Attention -- considering both spatial proximity and global correlation. Building on this block, we develop a multiscale spatiotemporal forecasting model by progressively expanding subgraph scales. The resulting model is both scalable and able to produce structured spatial correlation, and meanwhile, it is easy to implement. The experimental results show that it can achieve performance improvements up to 16.8% over time series forecasting baselines at low running costs.
Authors: Kevin Garcia, Cassandra Garza, Brooklyn Berry, Yifeng Gao
Abstract: The surge in the significance of time series in digital health domains necessitates advanced methodologies for extracting meaningful patterns and representations. Self-supervised contrastive learning has emerged as a promising approach for learning directly from raw data. However, time series data in digital health is known to be highly noisy, inherently involves concept drifting, and poses a challenge for training a generalizable deep learning model. In this paper, we specifically focus on data distribution shift caused by different human behaviors and propose a self-supervised learning framework that is aware of the bag-of-symbol representation. The bag-of-symbol representation is known for its insensitivity to data warping, location shifts, and noise existed in time series data, making it potentially pivotal in guiding deep learning to acquire a representation resistant to such data shifting. We demonstrate that the proposed method can achieve significantly better performance where significant data shifting exists.
Authors: Puning Zhao, Zhikun Zhang, Bo Sun, Li Shen, Liang Zhang, Shaowei Wang, Zhe Liu
Abstract: Distribution estimation under local differential privacy (LDP) is a fundamental and challenging task. Significant progresses have been made on categorical data. However, due to different evaluation metrics, these methods do not work well when transferred to numerical data. In particular, we need to prevent the probability mass from being misplaced far away. In this paper, we propose a new approach that express the sample distribution using wavelet expansions. The coefficients of wavelet series are estimated under LDP. Our method prioritizes the estimation of low-order coefficients, in order to ensure accurate estimation at macroscopic level. Therefore, the probability mass is prevented from being misplaced too far away from its ground truth. We establish theoretical guarantees for our methods. Experiments show that our wavelet expansion method significantly outperforms existing solutions under Wasserstein and KS distances.
Authors: Andrew Wang, Jiashuo Zhang, Michael Oberst
Abstract: Public healthcare datasets of Chest X-Rays (CXRs) have long been a popular benchmark for developing computer vision models in healthcare. However, strong average-case performance of machine learning (ML) models on these datasets is insufficient to certify their clinical utility. In this paper, we use clinical context, as captured by prior discharge summaries, to provide a more holistic evaluation of current ``state-of-the-art'' models for the task of CXR diagnosis. Using discharge summaries recorded prior to each CXR, we derive a ``prior'' or ``pre-test'' probability of each CXR label, as a proxy for existing contextual knowledge available to clinicians when interpreting CXRs. Using this measure, we demonstrate two key findings: First, for several diagnostic labels, CXR models tend to perform best on cases where the pre-test probability is very low, and substantially worse on cases where the pre-test probability is higher. Second, we use pre-test probability to assess whether strong average-case performance reflects true diagnostic signal, rather than an ability to infer the pre-test probability as a shortcut. We find that performance drops sharply on a balanced test set where this shortcut does not exist, which may indicate that much of the apparent diagnostic power derives from inferring this clinical context. We argue that this style of analysis, using context derived from clinical notes, is a promising direction for more rigorous and fine-grained evaluation of clinical vision models.
Authors: Kunlun Xu, Yibo Feng, Jiangmeng Li, Yongsheng Qi, Jiahuan Zhou
Abstract: Federated continual learning (FCL) tackles scenarios of learning from continuously emerging task data across distributed clients, where the key challenge lies in addressing both temporal forgetting over time and spatial forgetting simultaneously. Recently, prompt-based FCL methods have shown advanced performance through task-wise prompt communication.In this study, we underscore that the existing prompt-based FCL methods are prone to class-wise knowledge coherence between prompts across clients. The class-wise knowledge coherence includes two aspects: (1) intra-class distribution gap across clients, which degrades the learned semantics across prompts, (2) inter-prompt class-wise relevance, which highlights cross-class knowledge confusion. During prompt communication, insufficient class-wise coherence exacerbates knowledge conflicts among new prompts and induces interference with old prompts, intensifying both spatial and temporal forgetting. To address these issues, we propose a novel Class-aware Client Knowledge Interaction (C${}^2$Prompt) method that explicitly enhances class-wise knowledge coherence during prompt communication. Specifically, a local class distribution compensation mechanism (LCDC) is introduced to reduce intra-class distribution disparities across clients, thereby reinforcing intra-class knowledge consistency. Additionally, a class-aware prompt aggregation scheme (CPA) is designed to alleviate inter-class knowledge confusion by selectively strengthening class-relevant knowledge aggregation. Extensive experiments on multiple FCL benchmarks demonstrate that C${}^2$Prompt achieves state-of-the-art performance. Our source code is available at https://github.com/zhoujiahuan1991/NeurIPS2025-C2Prompt
URLs: https://github.com/zhoujiahuan1991/NeurIPS2025-C2Prompt
Authors: Gunbir Singh Baveja, Mark Schmidt
Abstract: Loss of trainability (LoT) in continual learning occurs when gradient steps no longer yield improvement as tasks evolve, so accuracy stalls or degrades despite adequate capacity and supervision. We analyze LoT incurred with Adam through an optimization lens and find that single indicators such as Hessian rank, sharpness level, weight or gradient norms, gradient-to-parameter ratios, and unit-sign entropy are not reliable predictors. Instead we introduce two complementary criteria: a batch-size-aware gradient-noise bound and a curvature volatility-controlled bound that combine into a per-layer predictive threshold that anticipates trainability behavior. Using this threshold, we build a simple per-layer scheduler that keeps each layers effective step below a safe limit, stabilizing training and improving accuracy across concatenated ReLU (CReLU), Wasserstein regularization, and L2 weight decay, with learned learning-rate trajectories that mirror canonical decay.
Authors: Patrick Lutz, Aditya Gangrade, Hadi Daneshmand, Venkatesh Saligrama
Abstract: We train a linear attention transformer on millions of masked-block matrix completion tasks: each prompt is masked low-rank matrix whose missing block may be (i) a scalar prediction target or (ii) an unseen kernel slice of Nystr\"om extrapolation. The model sees only input-output pairs and a mean-squared loss; it is given no normal equations, no handcrafted iterations, and no hint that the tasks are related. Surprisingly, after training, algebraic unrolling reveals the same parameter-free update rule across three distinct computational regimes (full visibility, rank-limited updates, and distributed computation). We prove that this rule achieves second-order convergence on full-batch problems, cuts distributed iteration complexity, and remains accurate with rank-limited attention. Thus, a transformer trained solely to patch missing blocks implicitly discovers a unified, resource-adaptive iterative solver spanning prediction, estimation, and Nystr\"om extrapolation, highlighting a powerful capability of in-context learning.
Authors: J. Ben Tamo, Nishant S. Chouhan, Micky C. Nnamdi, Yining Yuan, Shreya S. Chivilkar, Wenqi Shi, Steven W. Hwang, B. Randall Brenn, May D. Wang
Abstract: Surgical decision-making is complex and requires understanding causal relationships between patient characteristics, interventions, and outcomes. In high-stakes settings like spinal fusion or scoliosis correction, accurate estimation of individualized treatment effects (ITEs) remains limited due to the reliance on traditional statistical methods that struggle with complex, heterogeneous data. In this study, we develop a multi-task meta-learning framework, X-MultiTask, for ITE estimation that models each surgical decision (e.g., anterior vs. posterior approach, surgery vs. no surgery) as a distinct task while learning shared representations across tasks. To strengthen causal validity, we incorporate the inverse probability weighting (IPW) into the training objective. We evaluate our approach on two datasets: (1) a public spinal fusion dataset (1,017 patients) to assess the effect of anterior vs. posterior approaches on complication severity; and (2) a private AIS dataset (368 patients) to analyze the impact of posterior spinal fusion (PSF) vs. non-surgical management on patient-reported outcomes (PROs). Our model achieves the highest average AUC (0.84) in the anterior group and maintains competitive performance in the posterior group (0.77). It outperforms baselines in treatment effect estimation with the lowest overall $\epsilon_{\text{NN-PEHE}}$ (0.2778) and $\epsilon_{\text{ATE}}$ (0.0763). Similarly, when predicting PROs in AIS, X-MultiTask consistently shows superior performance across all domains, with $\epsilon_{\text{NN-PEHE}}$ = 0.2551 and $\epsilon_{\text{ATE}}$ = 0.0902. By providing robust, patient-specific causal estimates, X-MultiTask offers a powerful tool to advance personalized surgical care and improve patient outcomes. The code is available at https://github.com/Wizaaard/X-MultiTask.
Authors: Kainat
Abstract: This research presents a novel method for noninvasive arterial blood pressure ABP prediction using speech signals employing a BERT based regression model Arterial blood pressure is a vital indicator of cardiovascular health and accurate monitoring is essential in preventing hypertension related complications Traditional cuff based methods often yield inconsistent results due to factors like whitecoat and masked hypertension Our approach leverages the acoustic characteristics of speech capturing voice features to establish correlations with blood pressure levels Utilizing advanced deep learning techniques we analyze speech signals to extract relevant patterns enabling real time monitoring without the discomfort of conventional methods In our study we employed a dataset comprising recordings from 95 participants ensuring diverse representation The BERT model was fine tuned on extracted features from speech leading to impressive performance metrics achieving a mean absolute error MAE of 136 mmHg for systolic blood pressure SBP and 124 mmHg for diastolic blood pressure DBP with R scores of 099 and 094 respectively These results indicate the models robustness in accurately predicting blood pressure levels Furthermore the training and validation loss analysis demonstrates effective learning and minimal overfitting Our findings suggest that integrating deep learning with speech analysis presents a viable alternative for blood pressure monitoring paving the way for improved applications in telemedicine and remote health monitoring By providing a user friendly and accurate method for blood pressure assessment this research has significant implications for enhancing patient care and proactive management of cardiovascular health
Authors: Hyunwoo Kim, Hyo Kyung Lee
Abstract: We draw an analogy between static friction in classical mechanics and extrapolation error in off-policy RL, and use it to formulate a constraint that prevents the policy from drifting toward unsupported actions. In this study, we present Frictional Q-learning, a deep reinforcement learning algorithm for continuous control, which extends batch-constrained reinforcement learning. Our algorithm constrains the agent's action space to encourage behavior similar to that in the replay buffer, while maintaining a distance from the manifold of the orthonormal action space. The constraint preserves the simplicity of batch-constrained, and provides an intuitive physical interpretation of extrapolation error. Empirically, we further demonstrate that our algorithm is robustly trained and achieves competitive performance across standard continuous control benchmarks.
Authors: Jong Kwon Oh, Hanbaek Lyu, Hwijae Son
Abstract: Sobolev training, which integrates target derivatives into the loss functions, has been shown to accelerate convergence and improve generalization compared to conventional $L^2$ training. However, the underlying mechanisms of this training method remain only partially understood. In this work, we present the first rigorous theoretical framework proving that Sobolev training accelerates the convergence of Rectified Linear Unit (ReLU) networks. Under a student-teacher framework with Gaussian inputs and shallow architectures, we derive exact formulas for population gradients and Hessians, and quantify the improvements in conditioning of the loss landscape and gradient-flow convergence rates. Extensive numerical experiments validate our theoretical findings and show that the benefits of Sobolev training extend to modern deep learning tasks.
Authors: Xiaocheng Fang, Jiarui Jin, Haoyu Wang, Che Liu, Jieyi Cai, Guangkun Nie, Jun Li, Hongyan Li, Shenda Hong
Abstract: In clinical practice, electrocardiography (ECG) remains the gold standard for cardiac monitoring, providing crucial insights for diagnosing a wide range of cardiovascular diseases (CVDs). However, its reliance on specialized equipment and trained personnel limits feasibility for continuous routine monitoring. Photoplethysmography (PPG) offers accessible, continuous monitoring but lacks definitive electrophysiological information, preventing conclusive diagnosis. Generative models present a promising approach to translate PPG into clinically valuable ECG signals, yet current methods face substantial challenges, including the misalignment of physiological semantics in generative models and the complexity of modeling in high-dimensional signals. To this end, we propose PPGFlowECG, a two-stage framework that aligns PPG and ECG in a shared latent space via the CardioAlign Encoder and employs latent rectified flow to generate ECGs with high fidelity and interpretability. To the best of our knowledge, this is the first study to experiment on MCMED, a newly released clinical-grade dataset comprising over 10 million paired PPG-ECG samples from more than 118,000 emergency department visits with expert-labeled cardiovascular disease annotations. Results demonstrate the effectiveness of our method for PPG-to-ECG translation and cardiovascular disease detection. Moreover, cardiologist-led evaluations confirm that the synthesized ECGs achieve high fidelity and improve diagnostic reliability, underscoring our method's potential for real-world cardiovascular screening.
Authors: Ziyi Han, Xutong Liu, Ruiting Zhou, Xiangxiang Dai, John C. S. Lui
Abstract: Sparse Mixture of Experts (SMoE) has become a preferred architecture for scaling Transformer capacity without increasing computational cost, as it activates only a small subset of experts for each input. However, deploying such an approach for \textit{online inference} remains challenging due to the large size of a full SMoE model and the complexity of expert routing, especially in resource-constrained edge networks. Moreover, during the online inference, task information is often unavailable, making the task-level routing error-prone. In this work, we propose a novel tree-structured adaptive neural bandit router, \texttt{Tanbr}, to enable efficient and reliable online MoE inference. Instead of relying on explicit task tags, \texttt{Tanbr} estimates the task distribution over time from historical data and uses it to guide task-aware expert merging within a given pre-trained MoE. To handle the large continuous space of merging weights, \texttt{Tanbr} employs a binary tree to progressively partition the space and generate finer candidate weights. It then applies a neural bandit to learn the non-linear mapping from merging weight to model performance and decides optimal expert merging. We prove that \texttt{Tanbr} achieves a sublinear regret bound of {\small $\mathcal{O}(\sqrt{T} \log(T))$} over {\small $T$} rounds, despite operating over a continuous decision space, matching regret bounds compared to existing methods. Extensive experiments show that \texttt{Tanbr} reduces inference latency by at least {\small $45\%$} and memory usage by up to {\small $25\%$}, while maintaining a high accuracy compared to many state-of-the-art methods.
Authors: Carlo Bosio, Greg Woelki, Noureldin Hendy, Nicholas Roy, Byungsoo Kim
Abstract: Human drivers focus only on a handful of agents at any one time. On the other hand, autonomous driving systems process complex scenes with numerous agents, regardless of whether they are pedestrians on a crosswalk or vehicles parked on the side of the road. While attention mechanisms offer an implicit way to reduce the input to the elements that affect decisions, existing attention mechanisms for capturing agent interactions are quadratic, and generally computationally expensive. We propose RDAR, a strategy to learn per-agent relevance -- how much each agent influences the behavior of the controlled vehicle -- by identifying which agents can be excluded from the input to a pre-trained behavior model. We formulate the masking procedure as a Markov Decision Process where the action consists of a binary mask indicating agent selection. We evaluate RDAR on a large-scale driving dataset, and demonstrate its ability to learn an accurate numerical measure of relevance by achieving comparable driving performance, in terms of overall progress, safety and performance, while processing significantly fewer agents compared to a state of the art behavior model.
Authors: Guochao Jiang, Wenfeng Feng, Guofeng Quan, Chuzhan Hao, Yuewei Zhang, Guohua Liu, Hao Wang
Abstract: Policy-based reinforcement learning currently plays an important role in improving LLMs on mathematical reasoning tasks. However, existing rollout-based reinforcement learning methods (GRPO, DAPO, GSPO, etc.) fail to explicitly consider LLMs' learning ability for samples of different difficulty levels, which is contrary to the human cognitive process of mathematical reasoning tasks from easy to difficult. Intuitively, we find that the variance of the rollout group's reward in RLVR partly reflects the difficulty of the current sample for LLMs. Samples that are too easy or too difficult have a lower variance, while samples with moderate difficulty have a higher variance. Based on this, we propose VCRL, a curriculum reinforcement learning framework that dynamically controls the difficulty of training samples based on the variance of group rewards. Experiments on five mathematical benchmarks and two models reveal the advantages of VCRL over the current LLM RL baselines.
Authors: Zhijun Zeng, Weiye Gan, Junqing Chen, Zuoqiang Shi
Abstract: In many engineering and applied science domains, high-dimensional nonlinear filtering is still a challenging problem. Recent advances in score-based diffusion models offer a promising alternative for posterior sampling but require repeated retraining to track evolving priors, which is impractical in high dimensions. In this work, we propose the Conditional Score-based Filter (CSF), a novel algorithm that leverages a set-transformer encoder and a conditional diffusion model to achieve efficient and accurate posterior sampling without retraining. By decoupling prior modeling and posterior sampling into offline and online stages, CSF enables scalable score-based filtering across diverse nonlinear systems. Extensive experiments on benchmark problems show that CSF achieves superior accuracy, robustness, and efficiency across diverse nonlinear filtering scenarios.
Authors: Wei Liu, Eleni Chatzi, Zhilu Lai
Abstract: Kolmogorov-Arnold Networks (KANs) offer a structured and interpretable framework for multivariate function approximation by composing univariate transformations through additive or multiplicative aggregation. This paper establishes theoretical convergence guarantees for KANs when the univariate components are represented by B-splines. We prove that both additive and hybrid additive-multiplicative KANs attain the minimax-optimal convergence rate $O(n^{-2r/(2r+1)})$ for functions in Sobolev spaces of smoothness $r$. We further derive guidelines for selecting the optimal number of knots in the B-splines. The theory is supported by simulation studies that confirm the predicted convergence rates. These results provide a theoretical foundation for using KANs in nonparametric regression and highlight their potential as a structured alternative to existing methods.
Authors: Kevin Bradley Dsouza, Enoch Ofosu, Daniel Chukwuemeka Amaogu, J\'er\^ome Pigeon, Richard Boudreault, Pooneh Maghoul, Juan Moreno-Cruz, Yuri Leonenko
Abstract: Boreal forests store 30-40% of terrestrial carbon, much in climate-vulnerable permafrost soils, making their management critical for climate mitigation. However, optimizing forest management for both carbon sequestration and permafrost preservation presents complex trade-offs that current tools cannot adequately address. We introduce $\textbf{BoreaRL}$, the first multi-objective reinforcement learning environment for climate-adaptive boreal forest management, featuring a physically-grounded simulator of coupled energy, carbon, and water fluxes. BoreaRL supports two training paradigms: site-specific mode for controlled studies and generalist mode for learning robust policies under environmental stochasticity. Through evaluation of multi-objective RL algorithms, we reveal a fundamental asymmetry in learning difficulty: carbon objectives are significantly easier to optimize than thaw (permafrost preservation) objectives, with thaw-focused policies showing minimal learning progress across both paradigms. In generalist settings, standard preference-conditioned approaches fail entirely, while a naive curriculum learning approach achieves superior performance by strategically selecting training episodes. Analysis of learned strategies reveals distinct management philosophies, where carbon-focused policies favor aggressive high-density coniferous stands, while effective multi-objective policies balance species composition and density to protect permafrost while maintaining carbon gains. Our results demonstrate that robust climate-adaptive forest management remains challenging for current MORL methods, establishing BoreaRL as a valuable benchmark for developing more effective approaches. We open-source BoreaRL to accelerate research in multi-objective RL for climate applications.
Authors: Henrik Voigt, Paul Kahlmeyer, Kai Lawonn, Michael Habeck, Joachim Giesen
Abstract: Symbolic regression algorithms search a space of mathematical expressions for formulas that explain given data. Transformer-based models have emerged as a promising, scalable approach shifting the expensive combinatorial search to a large-scale pre-training phase. However, the success of these models is critically dependent on their pre-training data. Their ability to generalize to problems outside of this pre-training distribution remains largely unexplored. In this work, we conduct a systematic empirical study to evaluate the generalization capabilities of pre-trained, transformer-based symbolic regression. We rigorously test performance both within the pre-training distribution and on a series of out-of-distribution challenges for several state of the art approaches. Our findings reveal a significant dichotomy: while pre-trained models perform well in-distribution, the performance consistently degrades in out-of-distribution scenarios. We conclude that this generalization gap is a critical barrier for practitioners, as it severely limits the practical use of pre-trained approaches for real-world applications.
Authors: Samir Brahim Belhaouari, Yunis Carreon Kahalan, Humaira Shaffique, Ismael Belhaouari, Ashhadul Islam
Abstract: The effectiveness of machine learning models, particularly in unbalanced classification tasks, is often hindered by the failure to differentiate between critical instances near the decision boundary and redundant samples concentrated in the core of the data distribution. In this paper, we propose a method to systematically identify and differentiate between these two types of data. Through extensive experiments on multiple benchmark datasets, we show that the boundary data oversampling method improves the F1 score by up to 10\% on 96\% of the datasets, whereas our core-aware reduction method compresses datasets up to 90\% while preserving their accuracy, making it 10 times more powerful than the original dataset. Beyond imbalanced classification, our method has broader implications for efficient model training, particularly in computationally expensive domains such as Large Language Model (LLM) training. By prioritizing high-quality, decision-relevant data, our approach can be extended to text, multimodal, and self-supervised learning scenarios, offering a pathway to faster convergence, improved generalization, and significant computational savings. This work paves the way for future research in data-efficient learning, where intelligent sampling replaces brute-force expansion, driving the next generation of AI advancements. Our code is available as a Python package at https://pypi.org/project/adaptive-resampling/ .
Authors: Shi Yin, Zujian Dai, Xinyang Pan, Lixin He
Abstract: Deep learning methods for electronic-structure Hamiltonian prediction has offered significant computational efficiency advantages over traditional DFT methods, yet the diversity of atomic types, structural patterns, and the high-dimensional complexity of Hamiltonians pose substantial challenges to the generalization performance. In this work, we contribute on both the methodology and dataset sides to advance universal deep learning paradigm for Hamiltonian prediction. On the method side, we propose NextHAM, a neural E(3)-symmetry and expressive correction method for efficient and generalizable materials electronic-structure Hamiltonian prediction. First, we introduce the zeroth-step Hamiltonians, which can be efficiently constructed by the initial charge density of DFT, as informative descriptors of neural regression model in the input level and initial estimates of the target Hamiltonian in the output level, so that the regression model directly predicts the correction terms to the target ground truths, thereby significantly simplifying the input-output mapping for learning. Second, we present a neural Transformer architecture with strict E(3)-Symmetry and high non-linear expressiveness for Hamiltonian prediction. Third, we propose a novel training objective to ensure the accuracy performance of Hamiltonians in both real space and reciprocal space, preventing error amplification and the occurrence of "ghost states" caused by the large condition number of the overlap matrix. On the dataset side, we curate a high-quality broad-coverage large benchmark, namely Materials-HAM-SOC, comprising 17,000 material structures spanning 68 elements from six rows of the periodic table and explicitly incorporating SOC effects. Experimental results on Materials-HAM-SOC demonstrate that NextHAM achieves excellent accuracy and efficiency in predicting Hamiltonians and band structures.
Authors: Lorenzo Perini, Daniel Haimovich, Fridolin Linder, Niek Tax, Dima Karamshuk, Milan Vojnovic, Nastaran Okati, Pavlos Athanasios Apostolopoulos
Abstract: We propose MCGrad, a novel and scalable multicalibration algorithm. Multicalibration - calibration in sub-groups of the data - is an important property for the performance of machine learning-based systems. Existing multicalibration methods have thus far received limited traction in industry. We argue that this is because existing methods (1) require such subgroups to be manually specified, which ML practitioners often struggle with, (2) are not scalable, or (3) may harm other notions of model performance such as log loss and Area Under the Precision-Recall Curve (PRAUC). MCGrad does not require explicit specification of protected groups, is scalable, and often improves other ML evaluation metrics instead of harming them. MCGrad has been in production at Meta, and is now part of hundreds of production models. We present results from these deployments as well as results on public datasets.
Authors: Katja Naasunnguaq Jagd, Rachael DeVries, Ole Winther
Abstract: Domain-specific foundation models for healthcare have expanded rapidly in recent years, yet foundation models for critical care time series remain relatively underexplored due to the limited size and availability of datasets. In this work, we introduce an early-stage pre-trained foundation model for critical care time-series based on the Bi-Axial Transformer (BAT), trained on pooled electronic health record datasets. We demonstrate effective transfer learning by fine-tuning the model on a dataset distinct from the training sources for mortality prediction, where it outperforms supervised baselines, particularly for small datasets ($<5,000$). These contributions highlight the potential of self-supervised foundation models for critical care times series to support generalizable and robust clinical applications in resource-limited settings.
Authors: Xueliang Zhao, Wei Wu, Jian Guan, Zhuocheng Gong, Lingpeng Kong
Abstract: Large language models (LLMs) are evolving from conversational systems into strong reasoners for tasks such as Olympiad mathematics and competitive programming. While scaling parameters and test-time computation has driven progress, a key bottleneck is the lack of high-quality training problems: human-curated datasets are costly and limited, while existing synthetic corpora are often too easy or narrow. PromptCoT 1.0 showed that injecting rationales into prompt synthesis increases problem difficulty. Building on this, we present PromptCoT 2.0, a scalable framework that replaces hand-crafted heuristics with an expectation-maximization (EM) loop, where rationales are iteratively refined to guide prompt construction. This produces problems that are both harder and more diverse than prior corpora. The synthetic prompts support two post-training regimes: (1) Self-Play, where strong models improve autonomously via verifiable feedback without stronger teachers; and (2) Supervised Fine-Tuning (SFT), where weaker models learn from teacher-distilled traces. Extensive experiments demonstrate the effectiveness of this approach. In self-play, applying PromptCoT 2.0 to Qwen3-30B-A3B-Thinking-2507 sets new state-of-the-art results at the 30B scale, with +4.4, +4.8, and +5.3 on AIME 24/25 and HMMT 25, +6.1 and +5.0 on LiveCodeBench v5/v6, and +35 Elo on Codeforces. In SFT, training Qwen2.5-7B-Instruct solely on synthetic prompts boosts accuracy to 73.1 (AIME 24), 65.6 (AIME 25), and 53.4 (LiveCodeBench v5), surpassing models trained on human or hybrid data. Analyses further confirm that PromptCoT 2.0 yields fundamentally harder and distributionally distinct problems. These results establish prompt synthesis as a new axis for scaling reasoning and position PromptCoT 2.0 as a scalable foundation for future open-source models. The implementation is available at https://github.com/inclusionAI/PromptCoT.
Authors: Xinyu Liu, Chao Qin, Wei You
Abstract: We study pure exploration in structured stochastic multi-armed bandits, aiming to efficiently identify the correct hypothesis from a finite set of alternatives. For a broad class of tasks, asymptotic analyses reduce to a maximin optimization that admits a two-player zero-sum game interpretation between an experimenter and a skeptic: the experimenter allocates measurements to rule out alternatives while the skeptic proposes alternatives. We reformulate the game by allowing the skeptic to adopt a mixed strategy, yielding a concave-convex saddle-point problem. This viewpoint leads to Frank-Wolfe Self-Play (FWSP): a projection-free, regularization-free, tuning-free method whose one-hot updates on both sides match the bandit sampling paradigm. However, structural constraints introduce sharp pathologies that complicate algorithm design and analysis: our linear-bandit case study exhibits nonunique optima, optimal designs with zero mass on the best arm, bilinear objectives, and nonsmoothness at the boundary. We address these challenges via a differential-inclusion argument, proving convergence of the game value for best-arm identification in linear bandits. Our analysis proceeds through a continuous-time limit: a differential inclusion with a Lyapunov function that decays exponentially, implying a vanishing duality gap and convergence to the optimal value. Although Lyapunov analysis requires differentiability of the objective, which is not guaranteed on the boundary, we show that along continuous trajectories the algorithm steers away from pathological nonsmooth points and achieves uniform global convergence to the optimal game value. We then embed the discrete-time updates into a perturbed flow and show that the discrete game value also converges. Building on FWSP, we further propose a learning algorithm based on posterior sampling. Numerical experiments demonstrate a vanishing duality gap.
Authors: Songtao Li, Zhenyu Liao, Tianqi Hou, Ting Gao
Abstract: Few-shot generation, the synthesis of high-quality and diverse samples from limited training data, remains a significant challenge in generative modeling. Existing methods trained from scratch often fail to overcome overfitting and mode collapse, and fine-tuning large models can inherit biases while neglecting the crucial geometric structure of the latent space. To address these limitations, we introduce Latent Iterative Refinement Flow (LIRF), a novel approach that reframes few-shot generation as the progressive densification of geometrically structured manifold. LIRF establishes a stable latent space using an autoencoder trained with our novel \textbf{manifold-preservation loss} $L_{\text{manifold}}$. This loss ensures that the latent space maintains the geometric and semantic correspondence of the input data. Building on this, we propose an iterative generate-correct-augment cycle. Within this cycle, candidate samples are refined by a geometric \textbf{correction operator}, a provably contractive mapping that pulls samples toward the data manifold while preserving diversity. We also provide the \textbf{Convergence Theorem} demonstrating a predictable decrease in Hausdorff distance between generated and true data manifold. We also demonstrate the framework's scalability by generating coherent, high-resolution images on AFHQ-Cat. Ablation studies confirm that both the manifold-preserving latent space and the contractive correction mechanism are critical components of this success. Ultimately, LIRF provides a solution for data-scarce generative modeling that is not only theoretically grounded but also highly effective in practice.
Authors: Balazs Pejo, Marcell Frank, Krisztian Varga, Peter Veliczky
Abstract: This paper investigates the fragility of contribution evaluation in federated learning, a critical mechanism for ensuring fairness and incentivizing participation. We argue that contribution scores are susceptible to significant distortions from two fundamental perspectives: architectural sensitivity and intentional manipulation. First, we explore how different model aggregation methods impact these scores. While most research assumes a basic averaging approach, we demonstrate that advanced techniques, including those designed to handle unreliable or diverse clients, can unintentionally yet significantly alter the final scores. Second, we explore vulnerabilities posed by poisoning attacks, where malicious participants strategically manipulate their model updates to inflate their own contribution scores or reduce the importance of other participants. Through extensive experiments across diverse datasets and model architectures, implemented within the Flower framework, we rigorously show that both the choice of aggregation method and the presence of attackers are potent vectors for distorting contribution scores, highlighting a critical need for more robust evaluation schemes.
Authors: Remo Sasso, Michelangelo Conserva, Dominik Jeurissen, Paulo Rauber
Abstract: Exploration in reinforcement learning (RL) remains challenging, particularly in sparse-reward settings. While foundation models possess strong semantic priors, their capabilities as zero-shot exploration agents in classic RL benchmarks are not well understood. We benchmark LLMs and VLMs on multi-armed bandits, Gridworlds, and sparse-reward Atari to test zero-shot exploration. Our investigation reveals a key limitation: while VLMs can infer high-level objectives from visual input, they consistently fail at precise low-level control: the "knowing-doing gap". To analyze a potential bridge for this gap, we investigate a simple on-policy hybrid framework in a controlled, best-case scenario. Our results in this idealized setting show that VLM guidance can significantly improve early-stage sample efficiency, providing a clear analysis of the potential and constraints of using foundation models to guide exploration rather than for end-to-end control.
Authors: Jana Sweidan, Mounim A. El-Yacoubi, Nasredine Semmar
Abstract: Prompting large language models is a training-free method for detecting Alzheimer's disease from speech transcripts. Using the ADReSS dataset, we revisit zero-shot prompting and study few-shot prompting with a class-balanced protocol using nested interleave and a strict schema, sweeping up to 20 examples per class. We evaluate two variants achieving state-of-the-art prompting results. (i) MMSE-Proxy Prompting: each few-shot example carries a probability anchored to Mini-Mental State Examination bands via a deterministic mapping, enabling AUC computing; this reaches 0.82 accuracy and 0.86 AUC (ii) Reasoning-augmented Prompting: few-shot examples pool is generated with a multimodal LLM (GPT-5) that takes as input the Cookie Theft image, transcript, and MMSE to output a reasoning and MMSE-aligned probability; evaluation remains transcript-only and reaches 0.82 accuracy and 0.83 AUC. To our knowledge, this is the first ADReSS study to anchor elicited probabilities to MMSE and to use multimodal construction to improve interpretability.
Authors: Emmanouil Panagiotou, Beno\^it Ronval, Arjun Roy, Ludwig Bothmann, Bernd Bischl, Siegfried Nijssen, Eirini Ntoutsi
Abstract: Ensuring fairness in machine learning remains a significant challenge, as models often inherit biases from their training data. Generative models have recently emerged as a promising approach to mitigate bias at the data level while preserving utility. However, many rely on deep architectures, despite evidence that simpler models can be highly effective for tabular data. In this work, we introduce TABFAIRGDT, a novel method for generating fair synthetic tabular data using autoregressive decision trees. To enforce fairness, we propose a soft leaf resampling technique that adjusts decision tree outputs to reduce bias while preserving predictive performance. Our approach is non-parametric, effectively capturing complex relationships between mixed feature types, without relying on assumptions about the underlying data distributions. We evaluate TABFAIRGDT on benchmark fairness datasets and demonstrate that it outperforms state-of-the-art (SOTA) deep generative models, achieving better fairness-utility trade-off for downstream tasks, as well as higher synthetic data quality. Moreover, our method is lightweight, highly efficient, and CPU-compatible, requiring no data pre-processing. Remarkably, TABFAIRGDT achieves a 72% average speedup over the fastest SOTA baseline across various dataset sizes, and can generate fair synthetic data for medium-sized datasets (10 features, 10K samples) in just one second on a standard CPU, making it an ideal solution for real-world fairness-sensitive applications.
Authors: Mohammad Tabish, Benedict Leimkuhler, Stefan Klus
Abstract: We propose a randomized neural network approach called RaNNDy for learning transfer operators and their spectral decompositions from data. The weights of the hidden layers of the neural network are randomly selected and only the output layer is trained. The main advantage is that without a noticeable reduction in accuracy, this approach significantly reduces the training time and resources while avoiding common problems associated with deep learning such as sensitivity to hyperparameters and slow convergence. Additionally, the proposed framework allows us to compute a closed-form solution for the output layer which directly represents the eigenfunctions of the operator. Moreover, it is possible to estimate uncertainties associated with the computed spectral properties via ensemble learning. We present results for different dynamical operators, including Koopman and Perron-Frobenius operators, which have important applications in analyzing the behavior of complex dynamical systems, and the Schr\"odinger operator. The numerical examples, which highlight the strengths but also weaknesses of the proposed framework, include several stochastic dynamical systems, protein folding processes, and the quantum harmonic oscillator.
Authors: Feiyang Fu, Tongxian Guo, Zhaoqiang Liu
Abstract: Discrete diffusion models (DDMs) have shown powerful generation ability for discrete data modalities like text and molecules. However, their practical application is hindered by inefficient sampling, requiring a large number of sampling steps. Accelerating DDMs by using larger step sizes typically introduces significant problems in generation quality, as it amplifies the impact of both the compounding decoding error due to factorized predictions and discretization error from numerical approximations, leading to a significant decrease in sampling quality. To address these challenges, we propose learnable sampler distillation (LSD), a novel approach to train fast and high-fidelity samplers for DDMs. LSD employs a distillation approach where a student sampler with a few steps learns to align its intermediate score trajectory with that of a high-quality teacher sampler with numerous steps. This alignment is achieved by optimizing learnable sampler coefficients that adaptively adjust sampling dynamics. Additionally, we further propose LSD+, which also learns time schedules that allocate steps non-uniformly. Experiments across text generation, image generation, and synthetic tasks demonstrate that our proposed approaches outperform existing samplers for DDMs, achieving substantially higher sampling quality with significantly fewer sampling steps. Our code is available at \href{https://github.com/feiyangfu/LSD}{https://github.com/feiyangfu/LSD}.
URLs: https://github.com/feiyangfu/LSD, https://github.com/feiyangfu/LSD
Authors: Xilin Dai, Zhijian Xu, Wanxu Cai, Qiang Xu
Abstract: Most state-of-the-art probabilistic time series forecasting models rely on sampling to represent future uncertainty. However, this paradigm suffers from inherent limitations, such as lacking explicit probabilities, inadequate coverage, and high computational costs. In this work, we introduce \textbf{Probabilistic Scenarios}, an alternative paradigm designed to address the limitations of sampling. It operates by directly producing a finite set of \{Scenario, Probability\} pairs, thus avoiding Monte Carlo-like approximation. To validate this paradigm, we propose \textbf{TimePrism}, a simple model composed of only three parallel linear layers. Surprisingly, TimePrism achieves 9 out of 10 state-of-the-art results across five benchmark datasets on two metrics. The effectiveness of our paradigm comes from a fundamental reframing of the learning objective. Instead of modeling an entire continuous probability space, the model learns to represent a set of plausible scenarios and corresponding probabilities. Our work demonstrates the potential of the Probabilistic Scenarios paradigm, opening a promising research direction in forecasting beyond sampling.
Authors: Abdulla Jasem Almansoori, Maria Ivanova, Andrey Veprikov, Aleksandr Beznosikov, Samuel Horv\'ath, Martin Tak\'a\v{c}
Abstract: Low-Rank Adaptation (LoRA) fine-tunes large models by learning low-rank updates on top of frozen weights, dramatically reducing trainable parameters and memory. However, there is still a gap between full training with low-rank projections (SVDLoRA) and LoRA fine-tuning, indicating that LoRA steps can be further improved. In this study, we propose OPLoRA, a memory-efficient optimizer that closes this gap by casting LoRA optimization as an interpretable sub-problem and solving it efficiently with alternating least squares updates, where 1-2 alternating steps are empirically found to be sufficient to closely match truncated SVD without ever forming the full matrix. We also retrieve the recently proposed preconditioning methods for LoRA as a special case. OPLoRA supports momentum by maintaining a low-rank estimate using the same subroutine (LoRSum) for computing the step, with a memory budget of 3 times the number of LoRA parameters (i.e., same as Adam). We also propose an experimental scaled variant that uses the K-FAC metric, which could be of interest. Across a linear task, MNIST, CIFAR-100, and RoBERTa-base (MNLI), OPLoRA consistently approaches SVDLoRA's performance using significantly less memory.
Authors: Haolin Li, Tianjie Dai, Zhe Chen, Siyuan Du, Jiangchao Yao, Ya Zhang, Yanfeng Wang
Abstract: Clinical diagnosis is a highly specialized discipline requiring both domain expertise and strict adherence to rigorous guidelines. While current AI-driven medical research predominantly focuses on knowledge graphs or natural text pretraining paradigms to incorporate medical knowledge, these approaches primarily rely on implicitly encoded knowledge within model parameters, neglecting task-specific knowledge required by diverse downstream tasks. To address this limitation, we propose Retrieval-Augmented Diagnosis (RAD), a novel framework that explicitly injects external knowledge into multimodal models directly on downstream tasks. Specifically, RAD operates through three key mechanisms: retrieval and refinement of disease-centered knowledge from multiple medical sources, a guideline-enhanced contrastive loss that constrains the latent distance between multi-modal features and guideline knowledge, and the dual transformer decoder that employs guidelines as queries to steer cross-modal fusion, aligning the models with clinical diagnostic workflows from guideline acquisition to feature extraction and decision-making. Moreover, recognizing the lack of quantitative evaluation of interpretability for multimodal diagnostic models, we introduce a set of criteria to assess the interpretability from both image and text perspectives. Extensive evaluations across four datasets with different anatomies demonstrate RAD's generalizability, achieving state-of-the-art performance. Furthermore, RAD enables the model to concentrate more precisely on abnormal regions and critical indicators, ensuring evidence-based, trustworthy diagnosis. Our code is available at https://github.com/tdlhl/RAD.
Authors: Sepehr Maleki, Negar Pourmoazemi
Abstract: Anomalies in multivariate time series often arise from temporal context and cross-channel coordination rather than isolated outliers. We present Pi-Transformer, a physics-informed transformer with two attention pathways: a data-driven series attention and a smoothly evolving prior attention that encodes temporal invariants such as scale-related self-similarity and phase synchrony. The prior acts as a stable reference that calibrates reconstruction error. During training, we pair a reconstruction objective with a divergence term that encourages agreement between the two attentions while keeping them meaningfully distinct; the prior is regularised to evolve smoothly and is lightly distilled towards dataset-level statistics. At inference, the model combines an alignment-weighted reconstruction signal (Energy) with a mismatch signal that highlights timing and phase disruptions, and fuses them into a single score for detection. Across five benchmarks (SMD, MSL, SMAP, SWaT, and PSM), Pi-Transformer achieves state-of-the-art or highly competitive F1, with particular strength on timing and phase-breaking anomalies. Case analyses show complementary behaviour of the two streams and interpretable detections around regime changes. Embedding physics-informed priors into attention yields a calibrated and robust approach to anomaly detection in complex multivariate systems. Code is publicly available at this GitHub repository\footnote{https://github.com/sepehr-m/Pi-Transformer}.
Authors: Raphael Simon, Pieter Libin, Wim Mees
Abstract: Penetration testing, the simulation of cyberattacks to identify security vulnerabilities, presents a sequential decision-making problem well-suited for reinforcement learning (RL) automation. Like many applications of RL to real-world problems, partial observability presents a major challenge, as it invalidates the Markov property present in Markov Decision Processes (MDPs). Partially Observable MDPs require history aggregation or belief state estimation to learn successful policies. We investigate stochastic, partially observable penetration testing scenarios over host networks of varying size, aiming to better reflect real-world complexity through more challenging and representative benchmarks. This approach leads to the development of more robust and transferable policies, which are crucial for ensuring reliable performance across diverse and unpredictable real-world environments. Using vanilla Proximal Policy Optimization (PPO) as a baseline, we compare a selection of PPO variants designed to mitigate partial observability, including frame-stacking, augmenting observations with historical information, and employing recurrent or transformer-based architectures. We conduct a systematic empirical analysis of these algorithms across different host network sizes. We find that this task greatly benefits from history aggregation. Converging three times faster than other approaches. Manual inspection of the learned policies by the algorithms reveals clear distinctions and provides insights that go beyond quantitative results.
Authors: Rami Zewail
Abstract: Learning robust representations for biosignals is often hampered by the challenge of designing effective data augmentations.Traditional methods can fail to capture the complex variations inherent in physiological data. Within this context, we propose a novel hybrid framework, Diffusion-Augmented Contrastive Learning (DACL), that fuses concepts from diffusion models and supervised contrastive learning. The DACL framework operates on a latent space created by a lightweight Variational Autoencoder (VAE) trained on our novel Scattering Transformer (ST) features [12]. It utilizes the diffusion forward process as a principled data augmentation technique to generate multiple noisy views of these latent embeddings. A U-Net style encoder is then trained with a supervised contrastive objective to learn a representation that balances class discrimination with robustness to noise across various diffusion time steps. We evaluated this proof-of-concept method on the PhysioNet 2017 ECG dataset, achieving a competitive AUROC of 0.7815. This work establishes a new paradigm for representation learning by using the diffusion process itself to drive the contrastive objective, creating noise-invariant embeddings that demonstrate a strong foundation for class separability.
Authors: Shiqi Liu, Wenhan Cao, Chang Liu, Zeyu He, Tianyi Zhang, Shengbo Eben Li
Abstract: Estimating hidden states in dynamical systems, also known as optimal filtering, is a long-standing problem in various fields of science and engineering. In this paper, we introduce a general filtering framework, \textbf{LLM-Filter}, which leverages large language models (LLMs) for state estimation by embedding noisy observations with text prototypes. In various experiments for classical dynamical systems, we find that first, state estimation can significantly benefit from the reasoning knowledge embedded in pre-trained LLMs. By achieving proper modality alignment with the frozen LLM, LLM-Filter outperforms the state-of-the-art learning-based approaches. Second, we carefully design the prompt structure, System-as-Prompt (SaP), incorporating task instructions that enable the LLM to understand the estimation tasks. Guided by these prompts, LLM-Filter exhibits exceptional generalization, capable of performing filtering tasks accurately in changed or even unseen environments. We further observe a scaling-law behavior in LLM-Filter, where accuracy improves with larger model sizes and longer training times. These findings make LLM-Filter a promising foundation model of filtering.
Authors: Chen-Yu Liu, Leonardo Placidi, Kuan-Cheng Chen, Samuel Yen-Chi Chen, Gabriel Matos
Abstract: Quantum machine learning (QML) models conventionally rely on repeated measurements (shots) of observables to obtain reliable predictions. This dependence on large shot budgets leads to high inference cost and time overhead, which is particularly problematic as quantum hardware access is typically priced proportionally to the number of shots. In this work we propose You Only Measure Once (Yomo), a simple yet effective design that achieves accurate inference with dramatically fewer measurements, down to the single-shot regime. Yomo replaces Pauli expectation-value outputs with a probability aggregation mechanism and introduces loss functions that encourage sharp predictions. Our theoretical analysis shows that Yomo avoids the shot-scaling limitations inherent to expectation-based models, and our experiments on MNIST and CIFAR-10 confirm that Yomo consistently outperforms baselines across different shot budgets and under simulations with depolarizing channels. By enabling accurate single-shot inference, Yomo substantially reduces the financial and computational costs of deploying QML, thereby lowering the barrier to practical adoption of QML.
Authors: Zihan Zhou, Chenguang Wang, Hongyi Ye, Yongtao Guan, Tianshu Yu
Abstract: Learning physical dynamics from data is a fundamental challenge in machine learning and scientific modeling. Real-world observational data are inherently incomplete and irregularly sampled, posing significant challenges for existing data-driven approaches. In this work, we propose a principled diffusion-based framework for learning physical systems from incomplete training samples. To this end, our method strategically partitions each such sample into observed context and unobserved query components through a carefully designed splitting strategy, then trains a conditional diffusion model to reconstruct the missing query portions given available contexts. This formulation enables accurate imputation across arbitrary observation patterns without requiring complete data supervision. Specifically, we provide theoretical analysis demonstrating that our diffusion training paradigm on incomplete data achieves asymptotic convergence to the true complete generative process under mild regularity conditions. Empirically, we show that our method significantly outperforms existing baselines on synthetic and real-world physical dynamics benchmarks, including fluid flows and weather systems, with particularly strong performance in limited and irregular observation regimes. These results demonstrate the effectiveness of our theoretically principled approach for learning and imputing partially observed dynamics.
Authors: Erkan Karabulut, Daniel Daza, Paul Groth, Victoria Degeler
Abstract: Association Rule Mining (ARM) aims to discover patterns between features in datasets in the form of propositional rules, supporting both knowledge discovery and interpretable machine learning in high-stakes decision-making. However, in high-dimensional settings, rule explosion and computational overhead render popular algorithmic approaches impractical without effective search space reduction, challenges that propagate to downstream tasks. Neurosymbolic methods, such as Aerial+, have recently been proposed to address the rule explosion in ARM. While they tackle the high dimensionality of the data, they also inherit limitations of neural networks, particularly reduced performance in low-data regimes. This paper makes three key contributions to association rule discovery in high-dimensional tabular data. First, we empirically show that Aerial+ scales one to two orders of magnitude better than state-of-the-art algorithmic and neurosymbolic baselines across five real-world datasets. Second, we introduce the novel problem of ARM in high-dimensional, low-data settings, such as gene expression data from the biomedicine domain with around 18k features and 50 samples. Third, we propose two fine-tuning approaches to Aerial+ using tabular foundation models. Our proposed approaches are shown to significantly improve rule quality on five real-world datasets, demonstrating their effectiveness in low-data, high-dimensional scenarios.
Authors: Francesco Emanuele Stradi, Eleonora Fidelia Chiefari, Matteo Castiglioni, Alberto Marchesi, Nicola Gatti
Abstract: We study \emph{online episodic Constrained Markov Decision Processes} (CMDPs) under both stochastic and adversarial constraints. We provide a novel algorithm whose guarantees greatly improve those of the state-of-the-art best-of-both-worlds algorithm introduced by Stradi et al. (2025). In the stochastic regime, \emph{i.e.}, when the constraints are sampled from fixed but unknown distributions, our method achieves $\widetilde{\mathcal{O}}(\sqrt{T})$ regret and constraint violation without relying on Slater's condition, thereby handling settings where no strictly feasible solution exists. Moreover, we provide guarantees on the stronger notion of \emph{positive} constraint violation, which does not allow to recover from large violation in the early episodes by playing strictly safe policies. In the adversarial regime, \emph{i.e.}, when the constraints may change arbitrarily between episodes, our algorithm ensures sublinear constraint violation without Slater's condition, and achieves sublinear $\alpha$-regret with respect to the \emph{unconstrained} optimum, where $\alpha$ is a suitably defined multiplicative approximation factor. We further validate our results through synthetic experiments, showing the practical effectiveness of our algorithm.
Authors: Junjie Yao, Zhi-Qin John Xu
Abstract: The embedding space of language models is widely believed to capture the semantic relationships; for instance, embeddings of digits often exhibit an ordered structure that corresponds to their natural sequence. However, the mechanisms driving the formation of such structures remain poorly understood. In this work, we interpret the embedding structures via the data distribution. We propose a set of probability signatures that reflect the semantic relationships among tokens. Through experiments on the composite addition tasks using the linear model and feedforward network, combined with theoretical analysis of gradient flow dynamics, we reveal that these probability signatures significantly influence the embedding structures. We further generalize our analysis to large language models (LLMs) by training the Qwen2.5 architecture on the subsets of the Pile corpus. Our results show that the probability signatures are faithfully aligned with the embedding structures, particularly in capturing strong pairwise similarities among embeddings. Our work uncovers the mechanism of how data distribution guides the formation of embedding structures, establishing a novel understanding of the relationship between embedding organization and semantic patterns.
Authors: Xiong Peng, Bo Han, Fengfei Yu, Tongliang Liu, Feng Liu, Mingyuan Zhou
Abstract: Model inversion attacks (MIAs) aim to reconstruct class-representative samples from trained models. Recent generative MIAs utilize generative adversarial networks to learn image priors that guide the inversion process, yielding reconstructions with high visual quality and strong fidelity to the private training data. To explore the reason behind their effectiveness, we begin by examining the gradients of inversion loss with respect to synthetic inputs, and find that these gradients are surprisingly noisy. Further analysis reveals that generative inversion implicitly denoises these gradients by projecting them onto the tangent space of the generator manifold, filtering out off-manifold components while preserving informative directions aligned with the manifold. Our empirical measurements show that, in models trained with standard supervision, loss gradients often exhibit large angular deviations from the data manifold, indicating poor alignment with class-relevant directions. This observation motivates our central hypothesis: models become more vulnerable to MIAs when their loss gradients align more closely with the generator manifold. We validate this hypothesis by designing a novel training objective that explicitly promotes such alignment. Building on this insight, we further introduce a training-free approach to enhance gradient-manifold alignment during inversion, leading to consistent improvements over state-of-the-art generative MIAs.
Authors: Tiejun Wang, Rui Wang, Xudong Mou, Mengyuan Ma, Tianyu Wo, Renyu Yang, Xudong Liu
Abstract: Effective anomaly detection in time series is pivotal for modern industrial applications and financial systems. Due to the scarcity of anomaly labels and the high cost of manual labeling, reconstruction-based unsupervised approaches have garnered considerable attention. However, accurate anomaly detection remains an unsettled challenge, since the optimization objectives of reconstruction-based methods merely rely on point-by-point distance measures, ignoring the potential structural characteristics of time series and thus failing to tackle complex pattern-wise anomalies. In this paper, we propose StrAD, a novel structure-enhanced anomaly detection approach to enrich the optimization objective by incorporating structural information hidden in the time series and steering the data reconstruction procedure to better capture such structural features. StrAD accommodates the trend, seasonality, and shape in the optimization objective of the reconstruction model to learn latent structural characteristics and capture the intrinsic pattern variation of time series. The proposed structure-aware optimization objective mechanism can assure the alignment between the original data and the reconstructed data in terms of structural features, thereby keeping consistency in global fluctuation and local characteristics. The mechanism is pluggable and applicable to any reconstruction-based methods, enhancing the model sensitivity to both point-wise anomalies and pattern-wise anomalies. Experimental results show that StrAD improves the performance of state-of-the-art reconstruction-based models across five real-world anomaly detection datasets.
Authors: Fahmida Islam, Adnan Mahmood, Noorain Mukhtiar, Kasun Eranda Wijethilake, Quan Z. Sheng
Abstract: Federated Learning (FL) has been extensively employed for a number of applications in machine learning, i.e., primarily owing to its privacy preserving nature and efficiency in mitigating the communication overhead. Internet of Vehicles (IoV) is one of the promising applications, wherein FL can be utilized to train a model more efficiently. Since only a subset of the clients can participate in each FL training round, challenges arise pertinent to fairness in the client selection process. Over the years, a number of researchers from both academia and industry have proposed numerous FL frameworks. However, to the best of our knowledge, none of them have employed fairness for FL-based client selection in a dynamic and heterogeneous IoV environment. Accordingly, in this paper, we envisage a FairEquityFL framework to ensure an equitable opportunity for all the clients to participate in the FL training process. In particular, we have introduced a sampling equalizer module within the selector component for ensuring fairness in terms of fair collaboration opportunity for all the clients in the client selection process. The selector is additionally responsible for both monitoring and controlling the clients' participation in each FL training round. Moreover, an outlier detection mechanism is enforced for identifying malicious clients based on the model performance in terms of considerable fluctuation in either accuracy or loss minimization. The selector flags suspicious clients and temporarily suspend such clients from participating in the FL training process. We further evaluate the performance of FairEquityFL on a publicly available dataset, FEMNIST. Our simulation results depict that FairEquityFL outperforms baseline models to a considerable extent.
Authors: Albert Kj{\o}ller Jacobsen, Johanna Marie Gegenfurtner, Georgios Arvanitidis
Abstract: It has been shown that perturbing the input during training implicitly regularises the gradient of the learnt function, leading to smoother models and enhancing generalisation. However, previous research mostly considered the addition of ambient noise in the input space, without considering the underlying structure of the data. In this work, we propose several methods of adding geometry-aware input noise that accounts for the lower dimensional manifold the input space inhabits. We start by projecting ambient Gaussian noise onto the tangent space of the manifold. In a second step, the noise sample is mapped on the manifold via the associated geodesic curve. We also consider Brownian motion noise, which moves in random steps along the manifold. We show that geometry-aware noise leads to improved generalization and robustness to hyperparameter selection on highly curved manifolds, while performing at least as well as training without noise on simpler manifolds. Our proposed framework extends to learned data manifolds.
Authors: \'Alvaro Parafita, Tomas Garriga, Axel Brando, Francisco J. Cazorla
Abstract: Among explainability techniques, SHAP stands out as one of the most popular, but often overlooks the causal structure of the problem. In response, do-SHAP employs interventional queries, but its reliance on estimands hinders its practical application. To address this problem, we propose the use of estimand-agnostic approaches, which allow for the estimation of any identifiable query from a single model, making do-SHAP feasible on complex graphs. We also develop a novel algorithm to significantly accelerate its computation at a negligible cost, as well as a method to explain inaccessible Data Generating Processes. We demonstrate the estimation and computational performance of our approach, and validate it on two real-world datasets, highlighting its potential in obtaining reliable explanations.
Authors: Konrad Janik, Peter Benner
Abstract: Measurement data is often sampled irregularly, i.e., not on equidistant time grids. This is also true for Hamiltonian systems. However, existing machine learning methods, which learn symplectic integrators, such as SympNets [1] and H\'enonNets [2] still require training data generated by fixed step sizes. To learn time-adaptive symplectic integrators, an extension to SympNets called TSympNets is introduced in [3]. The aim of this work is to do a similar extension for H\'enonNets. We propose a novel neural network architecture called T-H\'enonNets, which is symplectic by design and can handle adaptive time steps. We also extend the T-H\'enonNet architecture to non-autonomous Hamiltonian systems. Additionally, we provide universal approximation theorems for both new architectures for separable Hamiltonian systems and discuss why it is difficult to handle non-separable Hamiltonian systems with the proposed methods. To investigate these theoretical approximation capabilities, we perform different numerical experiments.
Authors: Deokjae Lee, Hyun Oh Song
Abstract: We study weight-only post-training quantization (PTQ), which quantizes the weights of a large language model (LLM) without retraining, using little or no calibration data. Weight-only PTQ is crucial for reducing the memory footprint and latency of LLM inference, especially in memory-bound, small-batch inference scenarios, such as personalized inference on edge devices. Despite its importance, irregular weight distributions with heavy-tailed outliers in LLMs complicate quantization, recently motivating rotation-based methods that transform weights into near-Gaussian distributions, which are more regular with fewer outliers, thereby reducing quantization error. In this work, we first derive the information-theoretically optimal bit allocation for Gaussianized weights under given bit budgets, revealing that fine-grained fractional-bit quantizers approaching the Gaussian distortion-rate bound are essential to achieve near-optimal quantization performance. To bridge this theoretical insight and practical implementation, we introduce Q-Palette, a versatile collection of fractional-bit quantizers that range from trellis-coded quantizers offering near-optimal distortion to simpler vector and scalar quantizers optimized for faster inference, all efficiently implemented with optimized CUDA kernels across various bitwidths. Furthermore, leveraging Q-Palette as a foundational component, we propose a novel mixed-scheme quantization framework, jointly optimizing quantizer choices and layer fusion decisions given resource constraints. The code is available at https://github.com/snu-mllab/Q-Palette.
Authors: Wenhan Wu, Zheyuan Liu, Chongyang Gao, Ren Wang, Kaize Ding
Abstract: Current LLM unlearning methods face a critical security vulnerability that undermines their fundamental purpose: while they appear to successfully remove sensitive or harmful knowledge, this ``forgotten" information remains precariously recoverable through relearning attacks. We identify that the root cause is that conventional methods optimizing the forgetting loss at individual data points will drive model parameters toward sharp minima in the loss landscape. In these unstable regions, even minimal parameter perturbations can drastically alter the model's behaviors. Consequently, relearning attacks exploit this vulnerability by using just a few fine-tuning samples to navigate the steep gradients surrounding these unstable regions, thereby rapidly recovering knowledge that was supposedly erased. This exposes a critical robustness gap between apparent unlearning and actual knowledge removal. To address this issue, we propose StableUN, a bi-level feedback-guided optimization framework that explicitly seeks more stable parameter regions via neighborhood-aware optimization. It integrates forgetting feedback, which uses adversarial perturbations to probe parameter neighborhoods, with remembering feedback to preserve model utility, aligning the two objectives through gradient projection. Experiments on WMDP and MUSE benchmarks demonstrate that our method is significantly more robust against both relearning and jailbreaking attacks while maintaining competitive utility performance.
Authors: Xin An, Ruijie Li, Qiao Ning, Hui Li, Qian Ma, Shikai Guo
Abstract: Non-coding RNAs (ncRNAs) play pivotal roles in gene expression regulation and the pathogenesis of various diseases. Accurate classification of ncRNAs is essential for functional annotation and disease diagnosis. To address existing limitations in feature extraction depth and multimodal fusion, we propose HGMamba-ncRNA, a HyperGraphMamba-based multichannel adaptive model, which integrates sequence, secondary structure, and optionally available expression features of ncRNAs to enhance classification performance. Specifically, the sequence of ncRNA is modeled using a parallel Multi-scale Convolution and LSTM architecture (MKC-L) to capture both local patterns and long-range dependencies of nucleotides. The structure modality employs a multi-scale graph transformer (MSGraphTransformer) to represent the multi-level topological characteristics of ncRNA secondary structures. The expression modality utilizes a Chebyshev Polynomial-based Kolmogorov-Arnold Network (CPKAN) to effectively model and interpret high-dimensional expression profiles. Finally, by incorporating virtual nodes to facilitate efficient and comprehensive multimodal interaction, HyperGraphMamba is proposed to adaptively align and integrate multichannel heterogeneous modality features. Experiments conducted on three public datasets demonstrate that HGMamba-ncRNA consistently outperforms state-of-the-art methods in terms of accuracy and other metrics. Extensive empirical studies further confirm the model's robustness, effectiveness, and strong transferability, offering a novel and reliable strategy for complex ncRNA functional classification. Code and datasets are available at https://anonymous.4open.science/r/HGMamba-ncRNA-94D0.
Authors: Felipe Oviedo, Fiodar Kazhamiaka, Esha Choukse, Allen Kim, Amy Luers, Melanie Nakagawa, Ricardo Bianchini, Juan M. Lavista Ferres
Abstract: As AI inference scales to billions of queries and emerging reasoning and agentic workflows increase token demand, reliable estimates of per-query energy use are increasingly important for capacity planning, emissions accounting, and efficiency prioritization. Many public estimates are inconsistent and overstate energy use, because they extrapolate from limited benchmarks and fail to reflect efficiency gains achievable at scale. In this perspective, we introduce a bottom-up methodology to estimate the per-query energy of large-scale LLM systems based on token throughput. For models running on an H100 node under realistic workloads, GPU utilization and PUE constraints, we estimate a median energy per query of 0.34 Wh (IQR: 0.18-0.67) for frontier-scale models (>200 billion parameters). These results are consistent with measurements using production-scale configurations and show that non-production estimates and assumptions can overstate energy use by 4-20x. Extending to test-time scaling scenarios with 15x more tokens per typical query, the median energy rises 13x to 4.32 Wh, indicating that targeting efficiency in this regime will deliver the largest fleet-wide savings. We quantify achievable efficiency gains at the model, serving platform, and hardware levels, finding individual median reductions of 1.5-3.5x in energy per query, while combined advances can plausibly deliver 8-20x reductions. To illustrate the system-level impact, we estimate the baseline daily energy use of a deployment serving 1 billion queries to be 0.8 GWh/day. If 10% are long queries, demand could grow to 1.8 GWh/day. With targeted efficiency interventions, it falls to 0.9 GWh/day, similar to the energy footprint of web search at that scale. This echoes how data centers historically tempered energy growth through efficiency gains during the internet and cloud build-up.
Authors: Abhishek Sharma, Anat Parush, Sumit Wadhwa, Amihai Savir, Anne Guinard, Prateek Srivastava
Abstract: Accurate forecasting in the e-commerce finance domain is particularly challenging due to irregular invoice schedules, payment deferrals, and user-specific behavioral variability. These factors, combined with sparse datasets and short historical windows, limit the effectiveness of conventional time-series methods. While deep learning and Transformer-based models have shown promise in other domains, their performance deteriorates under partial observability and limited historical data. To address these challenges, we propose a hybrid forecasting framework that integrates dynamic lagged feature engineering and adaptive rolling-window representations with classical statistical models and ensemble learners. Our approach explicitly incorporates invoice-level behavioral modeling, structured lag of support data, and custom stability-aware loss functions, enabling robust forecasts in sparse and irregular financial settings. Empirical results demonstrate an approximate 5% reduction in MAPE compared to baseline models, translating into substantial financial savings. Furthermore, the framework enhances forecast stability over quarterly horizons and strengthens feature target correlation by capturing both short- and long-term patterns, leveraging user profile attributes, and simulating upcoming invoice behaviors. These findings underscore the value of combining structured lagging, invoice-level closure modeling, and behavioral insights to advance predictive accuracy in sparse financial time-series forecasting.
Authors: \"Omer Veysel \c{C}a\u{g}atan, Bar{\i}\c{s} Akg\"un
Abstract: In this paper, we show that Simple Preference Optimization (SimPO) can be derived as Maximum Entropy Reinforcement Learning with length-normalized temperature, providing a theoretical foundation for this reference-free method. Motivated by SimPO's strong performance in offline preference optimization, we investigate whether Maximum Entropy RL can achieve similar results in online RLHF settings. Our experiments find that Maximum Entropy RL consistently exhibits overoptimization and unstable KL dynamics, even at very low learning rates. Unlike KL-constrained methods that maintain stable training, entropy regularization fails to prevent reward hacking and appears to correlate with overoptimization. Lastly, we discuss possible explanations for why SimPO succeeds in offline settings while Maximum Entropy RL struggles in online scenarios. Our findings suggest that reference-free approaches may face distinct challenges when applied to online or offline preference learning.
Authors: Matteo Cardoni, Sam Leroux
Abstract: As deep neural networks are increasingly deployed in dynamic, real-world environments, relying on a single static model is often insufficient. Changes in input data distributions caused by sensor drift or lighting variations necessitate continual model adaptation. In this paper, we propose a hybrid training methodology that enables efficient on-device domain adaptation by combining the strengths of Backpropagation and Predictive Coding. The method begins with a deep neural network trained offline using Backpropagation to achieve high initial performance. Subsequently, Predictive Coding is employed for online adaptation, allowing the model to recover accuracy lost due to shifts in the input data distribution. This approach leverages the robustness of Backpropagation for initial representation learning and the computational efficiency of Predictive Coding for continual learning, making it particularly well-suited for resource-constrained edge devices or future neuromorphic accelerators. Experimental results on the MNIST and CIFAR-10 datasets demonstrate that this hybrid strategy enables effective adaptation with a reduced computational overhead, offering a promising solution for maintaining model performance in dynamic environments.
Authors: Prabhat Karmakar, Sayan Gupta, Ilaksh Adlakha
Abstract: Predicting how the microstructure governs the mechanical response of heterogeneous materials is essential for optimizing design and performance. Yet this task remains difficult due to the complex, high dimensional nature of microstructural features. Relying on physics based simulations to probe the microstructural space is computationally prohibitive. This motivates the development of computational tools to efficiently learn structure property linkages governing mechanical behavior. While contemporary data driven approaches offer new possibilities, they often require large datasets. To address this challenge, this work presents the Extended Low Rank Approximation (xLRA), a framework that employs canonical polyadic tensor decomposition. It efficiently maps high dimensional microstructural information to the local elastic response by adaptively incorporating higher rank terms. xLRA accurately predicts the local elastic strain fields in porous microstructures, requiring a maximum rank of only 4. The compact formulation of xLRA achieves accurate predictions when trained on just 5% of the dataset, demonstrating significant data efficiency. Moreover, xLRA proves transferability by delivering results across representative material systems, including two phase composites and single and dual phase polycrystals. Despite being compact, xLRA retains essential microstructural details, enabling accurate predictions on unseen microstructures. Benchmarking shows that xLRA outperforms contemporary methods in predictive accuracy, generalizability, and computational efficiency, while requiring 6 orders of magnitude fewer floating point operations. In summary, xLRA provides an efficient framework for predicting the elastic response from microstructures, enabling scalable mapping of structure property linkages.
Authors: Dayu Tan, Jing Chen, Xiaoping Zhou, Yansen Su, Chunhou Zheng
Abstract: Infectious diseases continue to pose a serious threat to public health, underscoring the urgent need for effective computational approaches to screen novel anti-infective agents. Oligopeptides have emerged as promising candidates in antimicrobial research due to their structural simplicity, high bioavailability, and low susceptibility to resistance. Despite their potential, computational models specifically designed to predict associations between oligopeptides and infectious diseases remain scarce. This study introduces a prompt-guided graph-based contrastive learning framework (PGCLODA) to uncover potential associations. A tripartite graph is constructed with oligopeptides, microbes, and diseases as nodes, incorporating both structural and semantic information. To preserve critical regions during contrastive learning, a prompt-guided graph augmentation strategy is employed to generate meaningful paired views. A dual encoder architecture, integrating Graph Convolutional Network (GCN) and Transformer, is used to jointly capture local and global features. The fused embeddings are subsequently input into a multilayer perceptron (MLP) classifier for final prediction. Experimental results on a benchmark dataset indicate that PGCLODA consistently outperforms state-of-the-art models in AUROC, AUPRC, and accuracy. Ablation and hyperparameter studies confirm the contribution of each module. Case studies further validate the generalization ability of PGCLODA and its potential to uncover novel, biologically relevant associations. These findings offer valuable insights for mechanism-driven discovery and oligopeptide-based drug development. The source code of PGCLODA is available online at https://github.com/jjnlcode/PGCLODA.
Authors: Benjamin Feuer, Chiung-Yi Tseng, Astitwa Sarthak Lathe, Oussama Elachqar, John P Dickerson
Abstract: LLM-judged benchmarks are increasingly used to evaluate complex model behaviors, yet their design introduces failure modes absent in conventional ground-truth based benchmarks. We argue that without tight objectives and verifiable constructions, benchmark rankings can produce high-confidence rankings that are in fact largely noise. We introduce two mechanisms to diagnose these issues. Schematic adherence quantifies how much of a judge's overall verdict is explained by the explicit evaluation schema, revealing unexplained variance when judges deviate from their own rubric. Psychometric validity aggregates internal consistency and discriminant validity signals to quantify irreducible uncertainty in any benchmarking run. Applying these tools to Arena-Hard Auto, we find severe schema incoherence and factor collapse across popular judges: for example, unexplained variance exceeding 90 percent for DeepSeek-R1-32B and factor correlations above 0.93 for most criteria. We also show that the ELO-style aggregation used by Arena-Hard Auto collapses and masks genuine ranking uncertainty. Our results highlight design failures that undermine validity and offer actionable principles for building better-scoped, reliability-aware LLM-judged benchmarks. We release our code at https://anonymous.4open.science/r/judgment-to-noise-947D/README.md
URLs: https://anonymous.4open.science/r/judgment-to-noise-947D/README.md
Authors: Dongming Huang, Zhifan Li, Yicheng Li, Qian Lin
Abstract: We study spectral algorithms in the setting where kernels are learned from data. We introduce the effective span dimension (ESD), an alignment-sensitive complexity measure that depends jointly on the signal, spectrum, and noise level $\sigma^2$. The ESD is well-defined for arbitrary kernels and signals without requiring eigen-decay conditions or source conditions. We prove that for sequence models whose ESD is at most $K$, the minimax excess risk scales as $\sigma^2 K$. Furthermore, we analyze over-parameterized gradient flow and prove that it can reduce the ESD. This finding establishes a connection between adaptive feature learning and provable improvements in generalization of spectral algorithms. We demonstrate the generality of the ESD framework by extending it to linear models and RKHS regression, and we support the theory with numerical experiments. This framework provides a novel perspective on generalization beyond traditional fixed-kernel theories.
Authors: Om Roy, Yashar Moshfeghi, Keith Smith
Abstract: Modelling dynamically evolving spatio-temporal signals is a prominent challenge in the Graph Neural Network (GNN) literature. Notably, GNNs assume an existing underlying graph structure. While this underlying structure may not always exist or is derived independently from the signal, a temporally evolving functional network can always be constructed from multi-channel data. Graph Variate Signal Analysis (GVSA) defines a unified framework consisting of a network tensor of instantaneous connectivity profiles against a stable support usually constructed from the signal itself. Building on GVSA and tools from graph signal processing, we introduce Graph-Variate Neural Networks (GVNNs): layers that convolve spatio-temporal signals with a signal-dependent connectivity tensor combining a stable long-term support with instantaneous, data-driven interactions. This design captures dynamic statistical interdependencies at each time step without ad hoc sliding windows and admits an efficient implementation with linear complexity in sequence length. Across forecasting benchmarks, GVNNs consistently outperform strong graph-based baselines and are competitive with widely used sequence models such as LSTMs and Transformers. On EEG motor-imagery classification, GVNNs achieve strong accuracy highlighting their potential for brain-computer interface applications.
Authors: Sara Fridovich-Keil, Mert Pilanci
Abstract: We prove the first guarantees of sparse recovery for ReLU neural networks, where the sparse network weights constitute the signal to be recovered. Specifically, we study structural properties of the sparse network weights for two-layer, scalar-output networks under which a simple iterative hard thresholding algorithm recovers these weights exactly, using memory that grows linearly in the number of nonzero weights. We validate this theoretical result with simple experiments on recovery of sparse planted MLPs, MNIST classification, and implicit neural representations. Experimentally, we find performance that is competitive with, and often exceeds, a high-performing but memory-inefficient baseline based on iterative magnitude pruning.
Authors: Thadd\"aus Wiedemer, Yuxuan Li, Paul Vicol, Shixiang Shane Gu, Nick Matarese, Kevin Swersky, Been Kim, Priyank Jaini, Robert Geirhos
Abstract: The remarkable zero-shot capabilities of Large Language Models (LLMs) have propelled natural language processing from task-specific models to unified, generalist foundation models. This transformation emerged from simple primitives: large, generative models trained on web-scale data. Curiously, the same primitives apply to today's generative video models. Could video models be on a trajectory towards general-purpose vision understanding, much like LLMs developed general-purpose language understanding? We demonstrate that Veo 3 can solve a broad variety of tasks it wasn't explicitly trained for: segmenting objects, detecting edges, editing images, understanding physical properties, recognizing object affordances, simulating tool use, and more. These abilities to perceive, model, and manipulate the visual world enable early forms of visual reasoning like maze and symmetry solving. Veo's emergent zero-shot capabilities indicate that video models are on a path to becoming unified, generalist vision foundation models.
Authors: Tianyu Ruan, Kuo Gai, Shihua Zhang
Abstract: Why do deep networks generalize well? In contrast to classical generalization theory, we approach this fundamental question by examining not only inputs and outputs, but the evolution of internal features. Our study suggests a phenomenon of temporal consistency that predictions remain stable when shallow features from earlier checkpoints combine with deeper features from later ones. This stability is not a trivial convergence artifact. It acts as a form of implicit, structured augmentation that supports generalization. We show that temporal consistency extends to unseen and corrupted data, but collapses when semantic structure is destroyed (e.g., random labels). Statistical tests further reveal that SGD injects anisotropic noise aligned with a few principal directions, reinforcing its role as a source of structured variability. Together, these findings suggest a conceptual perspective that links feature dynamics to generalization, pointing toward future work on practical surrogates for measuring temporal feature evolution.
Authors: Xinnan Dai, Chung-Hsiang Lo, Kai Guo, Shenglai Zeng, Dongsheng Luo, Jiliang Tang
Abstract: Transformer-based LLMs demonstrate strong performance on graph reasoning tasks, yet their internal mechanisms remain underexplored. To uncover these reasoning process mechanisms in a fundamental and unified view, we set the basic decoder-only transformers and explain them using the circuit-tracer framework. Through this lens, we visualize reasoning traces and identify two core mechanisms in graph reasoning: token merging and structural memorization, which underlie both path reasoning and substructure extraction tasks. We further quantify these behaviors and analyze how they are influenced by graph density and model size. Our study provides a unified interpretability framework for understanding structural reasoning in decoder-only Transformers.
Authors: Mohsen Nayebi Kerdabadi, William Andrew Byron, Xin Sun, Amirfarrokh Iranitalab
Abstract: Account Takeover (ATO) fraud poses a significant challenge in consumer banking, requiring high recall under strict latency while minimizing friction for legitimate users. Production systems typically rely on tabular gradient-boosted decision trees (e.g., XGBoost) that score sessions independently, overlooking the relational and temporal structure of online activity that characterizes coordinated attacks and "fraud rings." We introduce ATLAS (Account Takeover Learning Across Spatio-Temporal Directed Graph), a framework that reformulates ATO detection as spatio-temporal node classification on a time-respecting directed session graph. ATLAS links entities via shared identifiers (account, device, IP) and regulates connectivity with time-window and recency constraints, enabling causal, time-respecting message passing and latency-aware label propagation that uses only labels available at scoring time, non-anticipative and leakage-free. We operationalize ATLAS with inductive GraphSAGE variants trained via neighbor sampling, at scale on a sessions graph with more than 100M nodes and around 1B edges. On a high-risk digital product at Capital One, ATLAS delivers 6.38 percent AUC improvement and more than 50 percent reduction in customer friction, improving fraud capture while reducing user friction.
Authors: Ramona Rubini, Siavash Khodakarami, Aniruddha Bora, George Em Karniadakis, Michele Dassisti
Abstract: Accurate time-series forecasting for complex physical systems is the backbone of modern industrial monitoring and control. While deep learning models excel at capturing complex dynamics, currently, their deployment is limited due to physical inconsistency and robustness, hence constraining their reliability in regulated environments. We introduce process-informed forecasting (PIF) models for temperature in pharmaceutical lyophilization. We investigate a wide range of models, from classical ones such as Autoregressive Integrated Moving Average Model (ARIMA) and Exponential Smoothing Model (ETS), to modern deep learning architectures, including Kolmogorov-Arnold Networks (KANs). We compare three different loss function formulations that integrate a process-informed trajectory prior: a fixed-weight loss, a dynamic uncertainty-based loss, and a Residual-Based Attention (RBA) mechanism. We evaluate all models not only for accuracy and physical consistency but also for robustness to sensor noise. Furthermore, we test the practical generalizability of the best model in a transfer learning scenario on a new process. Our results show that PIF models outperform their data-driven counterparts in terms of accuracy, physical plausibility and noise resilience. This work provides a roadmap for developing reliable and generalizable forecasting solutions for critical applications in the pharmaceutical manufacturing landscape.
Authors: Yue Zhang, Jiaxin Zhang, Qiuyu Ren, Tahsin Saffat, Xiaoxuan Liu, Zitong Yang, Banghua Zhu, Yi Ma
Abstract: We introduce \textbf{GAUSS} (\textbf{G}eneral \textbf{A}ssessment of \textbf{U}nderlying \textbf{S}tructured \textbf{S}kills in Mathematics), a benchmark that evaluates LLMs' mathematical abilities across twelve core skill dimensions, grouped into three domains: knowledge and understanding, problem solving and communication, and meta-skills and creativity. By categorizing problems according to cognitive skills and designing tasks that isolate specific abilities, GAUSS constructs comprehensive, fine-grained, and interpretable profiles of models' mathematical abilities. These profiles faithfully represent their underlying mathematical intelligence. To exemplify how to use the \textsc{GAUSS} benchmark, we have derived the skill profile of \textsc{GPT-5-thinking}, revealing its strengths and weaknesses as well as its differences relative to \textsc{o4-mini-high}, thereby underscoring the value of multidimensional, skill-based evaluation.
Authors: Chang Wang, Ming Zhu, Shahram Latifi, Buddhadeb Dawn, Shengjie Zhai
Abstract: Congenital Heart Disease (CHD) is the most common neonatal anomaly, highlighting the urgent need for early detection to improve outcomes. Yet, fetal ECG (fECG) signals in abdominal ECG (aECG) are often masked by maternal ECG and noise, challenging conventional methods under low signal-to-noise ratio (SNR) conditions. We propose FetalHealthNet (FHNet), a deep learning framework that integrates Graph Neural Networks with a multi-scale enhanced transformer to dynamically model spatiotemporal inter-lead correlations and extract clean fECG signals. On benchmark aECG datasets, FHNet consistently outperforms long short-term memory (LSTM) models, standard transformers, and state-of-the-art models, achieving R2>0.99 and RMSE = 0.015 even under severe noise. Interpretability analyses highlight physiologically meaningful temporal and lead contributions, supporting model transparency and clinical trust. FHNet illustrates the potential of AI-driven modeling to advance fetal monitoring and enable early CHD screening, underscoring the transformative impact of next-generation biomedical signal processing.
Authors: Minghui Wu, Zhen Gao
Abstract: Massive multiple-input multiple-output (MIMO) promises high spectral efficiency but also leads to high-dimensional downlink channel state information (CSI), which complicates real-time channel acquisition and precoding. To address this, we propose an end-to-end (E2E) uplink-downlink CSI fusion precoding network that jointly models downlink CSI reference signal (CSI-RS) design, CSI feedback, and base-station (BS) precoding within a single E2E neural architecture. Concretely, a projection network built on the MAXIM architecture takes uplink sounding reference signals (SRS) as input and outputs frequency-, beam-, and port-domain projection matrices for designing downlink CSI-RS. User equipment (UE) then compresses/quantizes the resulting CSI-RS observations and feeds back a compact representation. At the base station (BS), two complementary branches produce candidate precoders: one is a feedback-only precoding network driven by quantized downlink observations, and the other is an SRS-only precoding network driven by uplink SRS. These candidate precoders are subsequently combined by a fusion precoding network to yield the final transmit precoder. All the modules are trained with a spectral-efficiency-oriented loss under a three-stage schedule. Simulation results show that the proposed approach effectively harnesses both SRS-derived information and UE feedback, achieving markedly better performance than conventional baselines.
Authors: Huipeng Liu, Zhichao Zhu, Yuan Zhou, Changlu Li
Abstract: As the consumption of traditional energy sources intensifies and their adverse environmental impacts become more pronounced, wave energy stands out as a highly promising member of the renewable energy family due to its high energy density, stability, widespread distribution, and environmental friendliness. The key to its development lies in the precise prediction of Significant Wave Height (WVHT). However, wave energy signals exhibit strong nonlinearity, abrupt changes, multi-scale periodicity, data sparsity, and high-frequency noise interference; additionally, physical models for wave energy prediction incur extremely high computational costs. To address these challenges, this study proposes a hybrid model combining STL-FFT-STFT-TCN-LSTM. This model exploits the Seasonal-Trend Decomposition Procedure based on Loess (STL), Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), Temporal Convolutional Network (TCN), and Long Short-Term Memory (LSTM) technologies. The model aims to optimize multi-scale feature fusion, capture extreme wave heights, and address issues related to high-frequency noise and periodic signals, thereby achieving efficient and accurate prediction of significant wave height. Experiments were conducted using hourly data from NOAA Station 41008 and 41047 spanning 2019 to 2022. The results showed that compared with other single models and hybrid models, the STL-FFT-STFT-TCN-LSTM model achieved significantly higher prediction accuracy in capturing extreme wave heights and suppressing high-frequency noise, with MAE reduced by 15.8\%-40.5\%, SMAPE reduced by 8.3\%-20.3\%, and R increased by 1.31\%-2.9\%; in ablation experiments, the model also demonstrated the indispensability of each component step, validating its superiority in multi-scale feature fusion.
Authors: Yiqiao Chen, Zijian Huang, Zhenghui Feng
Abstract: Pediatric arrhythmias are a major risk factor for disability and sudden cardiac death, yet their automated classification remains challenging due to class imbalance, few-shot categories, and complex signal characteristics, which severely limit the efficiency and reliability of early screening and clinical intervention. To address this problem, we propose a multimodal end-to-end deep learning framework that combines dual-branch convolutional encoders for ECG and IEGM, semantic attention for cross-modal feature alignment, and a lightweight Transformer encoder for global dependency modeling. In addition, we introduce a new contrastive loss fucntion named Adaptive Global Class-Aware Contrastive Loss (AGCACL) to enhance intra-class compactness and inter-class separability through class prototypes and a global similarity matrix. To the best of our knowledge, this is the first systematic study based on the Leipzig Heart Center pediatric/congenital ECG+IEGM dataset, for which we also provide a complete and reproducible preprocessing pipeline. Experimental results demonstrate that the proposed method achieves the overall best performance on this dataset, including 97.76\% Top-1 Accuracy, 94.08\% Macro Precision, 91.97\% Macro Recall, 92.97\% Macro F1, and 92.36\% Macro F2, with improvements of +13.64, +15.96, +19.82, and +19.44 percentage points over the strongest baseline in Macro Precision/Recall/F1/F2, respectively. These findings indicate that the framework significantly improves the detectability and robustness for minority arrhythmia classes, offering potential clinical value for rhythm screening, pre-procedural assessment, and postoperative follow-up in pediatric and congenital heart disease populations.
Authors: Ammar Kamoona, Hui Song, Ali Moradi Amani, Mahdi Jalili, Xinghuo Yu, Peter McTaggart
Abstract: Electric vehicle (EV) charging loads identification from behind smart meter recordings is an indispensable aspect that enables effective decision-making for energy distributors to reach an informed and intelligent decision about the power grid's reliability. When EV charging happens behind the meter (BTM), the charging occurs on the customer side of the meter, which measures the overall electricity consumption. In other words, the charging of the EV is considered part of the customer's load and not separately measured by the Distribution Network Operators (DNOs). DNOs require complete knowledge about the EV presence in their network. Identifying the EV charging demand is essential to better plan and manage the distribution grid. Unlike supervised methods, this paper addresses the problem of EV charging load identification in a non-nonintrusive manner from low-frequency smart meter using an unsupervised learning approach based on anomaly detection technique. Our approach does not require prior knowledge of EV charging profiles. It only requires real power consumption data of non-EV users, which are abundant in practice. We propose a deep temporal convolution encoding decoding (TAE) network. The TAE is applied to power consumption from smart BTM from Victorian households in Australia, and the TAE shows superior performance in identifying households with EVs.
Authors: Sina Montazeri, Waltenegus Dargie, Yunhe Feng, Kewei Sha
Abstract: Human activity recognition is critical for applications such as early intervention and health analytics. Traditional activity recognition relies on inertial measurement units (IMUs), which are resource intensive and require calibration. Although electrocardiogram (ECG)-based methods have been explored, these have typically served as supplements to IMUs or have been limited to broad categorical classification such as fall detection or active vs. inactive in daily activities. In this paper, we advance the field by demonstrating, for the first time, robust recognition of activity only with ECG in six distinct activities, which is beyond the scope of previous work. We design and evaluate three new deep learning models, including a CNN classifier with Squeeze-and-Excitation blocks for channel-wise feature recalibration, a ResNet classifier with dilated convolutions for multiscale temporal dependency capture, and a novel CNNTransformer hybrid combining convolutional feature extraction with attention mechanisms for long-range temporal relationship modeling. Tested on data from 54 subjects for six activities, all three models achieve over 94% accuracy for seen subjects, while CNNTransformer hybrid reaching the best accuracy of 72% for unseen subjects, a result that can be further improved by increasing the training population. This study demonstrates the first successful ECG-only activity classification in multiple physical activities, offering significant potential for developing next-generation wearables capable of simultaneous cardiac monitoring and activity recognition without additional motion sensors.
Authors: Zejun Liu, Yunshan Chen, Chengxi Xie, Huan Liu
Abstract: EEG-based multimodal emotion recognition(EMER) has gained significant attention and witnessed notable advancements, the inherent complexity of human neural systems has motivated substantial efforts toward multimodal approaches. However, this field currently suffers from three critical limitations: (i) the absence of open-source implementations. (ii) the lack of standardized and transparent benchmarks for fair performance analysis. (iii) in-depth discussion regarding main challenges and promising research directions is a notable scarcity. To address these challenges, we introduce LibEMER, a unified evaluation framework that provides fully reproducible PyTorch implementations of curated deep learning methods alongside standardized protocols for data preprocessing, model realization, and experimental setups. This framework enables unbiased performance assessment on three widely-used public datasets across two learning tasks. The open-source library is publicly accessible at: https://anonymous.4open.science/r/2025ULUIUBUEUMUEUR485384
URLs: https://anonymous.4open.science/r/2025ULUIUBUEUMUEUR485384
Authors: Enhao Huang, Zhiyu Zhang, Tianxiang Xu, Chunshu Xia, Kaichun Hu, Yuchen Yang, Tongtong Pan, Dong Dong, Zhan Qin
Abstract: Complex-valued signals encode both amplitude and phase, yet most deep models treat attention as real-valued correlation, overlooking interference effects. We introduce the Holographic Transformer, a physics-inspired architecture that incorporates wave interference principles into self-attention. Holographic attention modulates interactions by relative phase and coherently superimposes values, ensuring consistency between amplitude and phase. A dual-headed decoder simultaneously reconstructs the input and predicts task outputs, preventing phase collapse when losses prioritize magnitude over phase. We demonstrate that holographic attention implements a discrete interference operator and maintains phase consistency under linear mixing. Experiments on PolSAR image classification and wireless channel prediction show strong performance, achieving high classification accuracy and F1 scores, low regression error, and increased robustness to phase perturbations. These results highlight that enforcing physical consistency in attention leads to generalizable improvements in complex-valued learning and provides a unified, physics-based framework for coherent signal modeling. The code is available at https://github.com/EonHao/Holographic-Transformers.
Authors: Shangqing Yuan, Wenshuang Zhai, Shengwen Guo
Abstract: To address the issue of limited channels and insufficient information collection in portable EEG devices, this study explores an EEG virtual channel signal generation network using a novel spatio-temporal feature fusion strategy. Based on the EEG signals from four frontal lobe channels, the network aims to generate virtual channel EEG signals for other 13 important brain regions. The architecture of the network is a two-dimensional convolutional neural network and it includes a parallel module for temporal and spatial domain feature extraction, followed by a feature fusion module. The public PRED+CT database, which includes multi-channel EEG signals from 119 subjects, was selected to verify the constructed network. The results showed that the average correlation coefficient between the generated virtual channel EEG signals and the original real signals was 0.6724, with an average absolute error of 3.9470. Furthermore, the 13 virtual channel EEG signals were combined with the original EEG signals of four brain regions and then used for anxiety classification with a support vector machine. The results indicate that the virtual EEG signals generated by the constructed network not only have a high degree of consistency with the real channel EEG signals but also significantly enhance the performance of machine learning algorithms for anxiety classification. This study effectively alleviates the problem of insufficient information acquisition by portable EEG devices with few channels.
Authors: Yang Fu, Peng Qin, Yueyue Zhang, Yifei Wang
Abstract: 6G networks are envisioned to support on-demand AI model downloading to accommodate diverse inference requirements of end users. By proactively caching models at edge nodes, users can retrieve the requested models with low latency for on-device AI inference. However, the substantial size of contemporary AI models poses significant challenges for edge caching under limited storage capacity, as well as for the concurrent delivery of heterogeneous models over wireless channels. To address these challenges, we propose a fine-grained AI model caching and downloading system that exploits parameter reusability, stemming from the common practice of fine-tuning task-specific models from a shared pre-trained model with frozen parameters. This system selectively caches model parameter blocks (PBs) at edge nodes, eliminating redundant storage of reusable parameters across different cached models. Additionally, it incorporates coordinated multipoint (CoMP) broadcasting to simultaneously deliver reusable PBs to multiple users, thereby enhancing downlink spectrum utilization. Under this arrangement, we formulate a model downloading delay minimization problem to jointly optimize PB caching, migration (among edge nodes), and broadcasting beamforming. To tackle this intractable problem, we develop a distributed multi-agent learning framework that enables edge nodes to explicitly learn mutual influence among their actions, thereby facilitating cooperation. Furthermore, a data augmentation approach is proposed to adaptively generate synthetic training samples through a predictive model, boosting sample efficiency and accelerating policy learning. Both theoretical analysis and simulation experiments validate the superior convergence performance of the proposed learning framework.
Authors: Xinyu Qin, Ye Xue, Qi Yan, Shutao Zhang, Bingsheng Peng, Tsung-Hui Chang
Abstract: Localized statistical channel modeling (LSCM) is crucial for effective performance evaluation in digital twin-assisted network optimization. Solely relying on the multi-beam reference signal receiving power (RSRP), LSCM aims to model the localized statistical propagation environment by estimating the channel angular power spectrum (APS). However, existing methods rely heavily on drive test data with high collection costs and limited spatial coverage. In this paper, we propose a measurement report (MR) data-driven framework for LSCM, exploiting the low-cost and extensive collection of MR data. The framework comprises two novel modules. The MR localization module addresses the issue of missing locations in MR data by introducing a semi-supervised method based on hypergraph neural networks, which exploits multi-modal information via distance-aware hypergraph modeling and hypergraph convolution for location extraction. To enhance the computational efficiency and solution robustness, LSCM operates at the grid level. Compared to independently constructing geographically uniform grids and estimating channel APS, the joint grid construction and channel APS estimation module enhances robustness in complex environments with spatially non-uniform data by exploiting their correlation. This module alternately optimizes grid partitioning and APS estimation using clustering and improved sparse recovery for the ill-conditioned measurement matrix and incomplete observations. Through comprehensive experiments on a real-world MR dataset, we demonstrate the superior performance and robustness of our framework in localization and channel modeling.
Authors: Robert Tjarko Lange, Yuki Imajuku, Edoardo Cetin
Abstract: We introduce ShinkaEvolve: a new open-source framework leveraging large language models (LLMs) to advance scientific discovery with state-of-the-art performance and unprecedented efficiency. Recent advances in scaling inference time compute of LLMs have enabled significant progress in generalized scientific discovery. These approaches rely on evolutionary agentic harnesses that leverage LLMs as mutation operators to generate candidate solutions. However, current code evolution methods suffer from critical limitations: they are sample inefficient, requiring thousands of samples to identify effective solutions, and remain closed-source, hindering broad adoption and extension. ShinkaEvolve addresses these limitations, introducing three key innovations: a parent sampling technique balancing exploration and exploitation, code novelty rejection-sampling for efficient search space exploration, and a bandit-based LLM ensemble selection strategy. We evaluate ShinkaEvolve across diverse tasks, demonstrating consistent improvements in sample efficiency and solution quality. ShinkaEvolve discovers a new state-of-the-art circle packing solution using only 150 samples, designs high-performing agentic harnesses for AIME mathematical reasoning tasks, identifies improvements to ALE-Bench competitive programming solutions, and discovers novel mixture-of-expert load balancing loss functions that illuminate the space of optimization strategies. Our results demonstrate that ShinkaEvolve achieves broad applicability with exceptional sample efficiency. By providing open-source accessibility and cost-efficiency, this work democratizes open-ended discovery across diverse computational problems.
Authors: Katherine Xie, Gabriel Koch Ocker
Abstract: The Drosophila mushroom body (MB) is known to be involved in olfactory learning and memory; the synaptic plasticity of the Kenyon cell (KC) to mushroom body output neuron (MBON) synapses plays a key role in the learning process. Previous research has focused on projection neuron (PN) to Kenyon cell (KC) connectivity within the MB; we examine how perturbations to the mushroom body circuit structure and changes in connectivity, specifically within the KC to mushroom body output neuron (MBON) neural circuit, affect the MBONs' ability to distinguish between odor classes. We constructed a neural network that incorporates the connectivity between PNs, KCs, and MBONs. To train our model, we generated ten artificial input classes, which represent the projection neuron activity in response to different odors. We collected data on the number of KC-to-MBON connections, MBON error rates, and KC-to-MBON synaptic weights, among other metrics. We observed that MBONs with very few presynaptic KCs consistently performed worse than others in the odor classification task. The developmental types of KCs also played a significant role in each MBON's output. We performed random and targeted KC ablation and observed that ablating developmentally mature KCs had a greater negative impact on MBONs' learning capacity than ablating immature KCs. Random and targeted pruning of KC-MBON synaptic connections yielded results largely consistent with the ablation experiments. To further explore the various types of KCs, we also performed rewiring experiments in the PN to KC circuit. Our study furthers our understanding of olfactory neuroplasticity and provides important clues to understanding learning and memory in general. Understanding how the olfactory circuits process and learn can also have potential applications in artificial intelligence and treatments for neurodegenerative diseases.
Authors: Wannes Janssens, Matthias Bogaert, Dirk Van den Poel
Abstract: The BERTopic framework leverages transformer embeddings and hierarchical clustering to extract latent topics from unstructured text corpora. While effective, it often struggles with social media data, which tends to be noisy and sparse, resulting in an excessive number of overlapping topics. Recent work explored the use of large language models for end-to-end topic modelling. However, these approaches typically require significant computational overhead, limiting their scalability in big data contexts. In this work, we propose a framework that combines BERTopic for topic generation with large language models for topic reduction. The method first generates an initial set of topics and constructs a representation for each. These representations are then provided as input to the language model, which iteratively identifies and merges semantically similar topics. We evaluate the approach across three Twitter/X datasets and four different language models. Our method outperforms the baseline approach in enhancing topic diversity and, in many cases, coherence, with some sensitivity to dataset characteristics and initial parameter selection.
Authors: Borhan Uddin Chowdhury, Damian Valles, Md Raf E Ul Shougat
Abstract: We present a sensor-fusion framework for rapid, non-destructive classification and quality control of organic substances, built on a standard Arduino Mega 2560 microcontroller platform equipped with three commercial environmental and gas sensors. All data used in this study were generated in-house: sensor outputs for ten distinct classes - including fresh and expired samples of apple juice, onion, garlic, and ginger, as well as cinnamon and cardamom - were systematically collected and labeled using this hardware setup, resulting in a unique, application-specific dataset. Correlation analysis was employed as part of the preprocessing pipeline for feature selection. After preprocessing and dimensionality reduction (PCA/LDA), multiple supervised learning models - including Support Vector Machine (SVM), Decision Tree (DT), and Random Forest (RF), each with hyperparameter tuning, as well as an Artificial Neural Network (ANN) and an ensemble voting classifier - were trained and cross-validated on the collected dataset. The best-performing models, including tuned Random Forest, ensemble, and ANN, achieved test accuracies in the 93 to 94 percent range. These results demonstrate that low-cost, multisensory platforms based on the Arduino Mega 2560, combined with advanced machine learning and correlation-driven feature engineering, enable reliable identification and quality control of organic compounds.
Authors: Oscar A. Oviedo
Abstract: This study presents the development and optimization of a deep learning model based on Long Short-Term Memory (LSTM) networks to predict short-term hourly electricity demand in C\'ordoba, Argentina. Integrating historical consumption data with exogenous variables (climatic factors, temporal cycles, and demographic statistics), the model achieved high predictive precision, with a mean absolute percentage error of 3.20\% and a determination coefficient of 0.95. The inclusion of periodic temporal encodings and weather variables proved crucial to capture seasonal patterns and extreme consumption events, enhancing the robustness and generalizability of the model. In addition to the design and hyperparameter optimization of the LSTM architecture, two complementary analyses were carried out: (i) an interpretability study using Random Forest regression to quantify the relative importance of exogenous drivers, and (ii) an evaluation of model performance in predicting the timing of daily demand maxima and minima, achieving exact-hour accuracy in more than two-thirds of the test days and within abs(1) hour in over 90\% of cases. Together, these results highlight both the predictive accuracy and operational relevance of the proposed framework, providing valuable insights for grid operators seeking optimized planning and control strategies under diverse demand scenarios.
Authors: Nelson Alves Ferreira Neto
Abstract: Low-latency intelligent systems are required for autonomous driving on non-uniform terrain in open-pit mines and developing countries. This work proposes a perception system for autonomous vehicles on unpaved roads and off-road environments, capable of navigating rough terrain without a predefined trail. The Configurable Modular Segmentation Network (CMSNet) framework is proposed, facilitating different architectural arrangements. CMSNet configurations were trained to segment obstacles and trafficable ground on new images from unpaved/off-road scenarios with adverse conditions (night, rain, dust). We investigated applying deep learning to detect drivable regions without explicit track boundaries, studied algorithm behavior under visibility impairment, and evaluated field tests with real-time semantic segmentation. A new dataset, Kamino, is presented with almost 12,000 images from an operating vehicle with eight synchronized cameras. The Kamino dataset has a high number of labeled pixels compared to similar public collections and includes images from an off-road proving ground emulating a mine under adverse visibility. To achieve real-time inference, CMSNet CNN layers were methodically removed and fused using TensorRT, C++, and CUDA. Empirical experiments on two datasets validated the proposed system's effectiveness.
Authors: Xiaolong Li, Zhi-qin John Xu, Peiting You, Yifei Zhu
Abstract: Passive intermodulation (PIM) has emerged as a critical source of self-interference in modern MIMO-OFDM systems, especially under the stringent requirements of 5G and beyond. Conventional cancellation methods often rely on complex nonlinear models with limited scalability and high computational cost. In this work, we propose a lightweight deep learning framework for PIM cancellation that leverages depthwise separable convolutions and dilated convolutions to efficiently capture nonlinear dependencies across antennas and subcarriers. To further enhance convergence, we adopt a cyclic learning rate schedule and gradient clipping. In a controlled MIMO experimental setup, the method effectively suppresses third-order passive intermodulation (PIM) distortion, achieving up to 29dB of average power error (APE) with only 11k trainable parameters. These results highlight the potential of compact neural architectures for scalable interference mitigation in future wireless communication systems.
Authors: Hongyuan Shi, Yilin Zhai, Ping Dong, Zaijin You, Chao Zhan, Qing Wang
Abstract: Reconstructing high-resolution regional significant wave height fields from sparse and uneven buoy observations remains a core challenge for ocean monitoring and risk-aware operations. We introduce AUWave, a hybrid deep learning framework that fuses a station-wise sequence encoder (MLP) with a multi-scale U-Net enhanced by a bottleneck self-attention layer to recover 32$\times$32 regional SWH fields. A systematic Bayesian hyperparameter search with Optuna identifies the learning rate as the dominant driver of generalization, followed by the scheduler decay and the latent dimension. Using NDBC buoy observations and ERA5 reanalysis over the Hawaii region, AUWave attains a minimum validation loss of 0.043285 and a slightly right-skewed RMSE distribution. Spatial errors are lowest near observation sites and increase with distance, reflecting identifiability limits under sparse sampling. Sensitivity experiments show that AUWave consistently outperforms a representative baseline in data-richer configurations, while the baseline is only marginally competitive in the most underdetermined single-buoy cases. The architecture's multi-scale and attention components translate into accuracy gains when minimal but non-trivial spatial anchoring is available. Error maps and buoy ablations reveal key anchor stations whose removal disproportionately degrades performance, offering actionable guidance for network design. AUWave provides a scalable pathway for gap filling, high-resolution priors for data assimilation, and contingency reconstruction.
Authors: Benjamin J. Choi, Griffin Milsap, Clara A. Scholl, Francesco Tenore, Mattson Ogg
Abstract: Effective control of neural interfaces is limited by poor signal quality. While neural network-based electroencephalography (EEG) denoising methods for electromyogenic (EMG) artifacts have improved in recent years, current state-of-the-art (SOTA) models perform suboptimally in settings with high noise. To address the shortcomings of current machine learning (ML)-based denoising algorithms, we present a signal filtration algorithm driven by a new mixture-of-experts (MoE) framework. Our algorithm leverages three new statistical insights into the EEG-EMG denoising problem: (1) EMG artifacts can be partitioned into quantifiable subtypes to aid downstream MoE classification, (2) local experts trained on narrower signal-to-noise ratio (SNR) ranges can achieve performance increases through specialization, and (3) correlation-based objective functions, in conjunction with rescaling algorithms, can enable faster convergence in a neural network-based denoising context. We empirically demonstrate these three insights into EMG artifact removal and use our findings to create a new downstream MoE denoising algorithm consisting of convolutional (CNN) and recurrent (RNN) neural networks. We tested all results on a major benchmark dataset (EEGdenoiseNet) collected from 67 subjects. We found that our MoE denoising model achieved competitive overall performance with SOTA ML denoising algorithms and superior lower bound performance in high noise settings. These preliminary results highlight the promise of our MoE framework for enabling advances in EMG artifact removal for EEG processing, especially in high noise settings. Further research and development will be necessary to assess our MoE framework on a wider range of real-world test cases and explore its downstream potential to unlock more effective neural interfaces.
Authors: Antonio Quintero Rincon, Nicolas Masino, Veronica Marsico, Hadj Batatia
Abstract: Spike-and-wave discharges (SWDs) are the electroencephalographic hallmark of absence epilepsy, yet their manual identification in multi-day recordings remains labour-intensive and error-prone. We present a lightweight hybrid pipeline that couples analytical features with a shallow artificial neural network (ANN) for accurate, patient-specific SWD detection in long-term, monopolar EEG. A two-sided moving-average (MA) filter first suppresses the high-frequency components of normal background activity. The residual signal is then summarised by the mean and the standard deviation of its normally distributed samples, yielding a compact, two-dimensional feature vector for every 20s window. These features are fed to a single-hidden-layer ANN trained via back-propagation to classify each window as SWD or non-SWD. The method was evaluated on 780 channels sampled at 256 Hz from 12 patients, comprising 392 annotated SWD events. It correctly detected 384 events (sensitivity: 98%) while achieving a specificity of 96.2 % and an overall accuracy of 97.2%. Because feature extraction is analytic, and the classifier is small, the pipeline runs in real-time and requires no manual threshold tuning. These results indicate that normal-distribution descriptors combined with a modest ANN provide an effective and computationally inexpensive solution for automated SWD screening in extended EEG recordings.
Authors: Jiarui Jin, Xiaocheng Fang, Haoyu Wang, Jun Li, Che Liu, Donglin Xie, Hongyan Li, Shenda Hong
Abstract: Myocardial infarction is a critical manifestation of coronary artery disease, yet detecting it from single-lead electrocardiogram (ECG) remains challenging due to limited spatial information. An intuitive idea is to convert single-lead into multiple-lead ECG for classification by pre-trained models, but generative methods optimized at the signal level in most cases leave a large latent space gap, ultimately degrading diagnostic performance. This naturally raises the question of whether latent space alignment could help. However, most prior ECG alignment methods focus on learning transformation invariance, which mismatches the goal of single-lead detection. To address this issue, we propose SelfMIS, a simple yet effective alignment learning framework to improve myocardial infarction detection from single-lead ECG. Discarding manual data augmentations, SelfMIS employs a self-cutting strategy to pair multiple-lead ECG with their corresponding single-lead segments and directly align them in the latent space. This design shifts the learning objective from pursuing transformation invariance to enriching the single-lead representation, explicitly driving the single-lead ECG encoder to learn a representation capable of inferring global cardiac context from the local signal. Experimentally, SelfMIS achieves superior performance over baseline models across nine myocardial infarction types while maintaining a simpler architecture and lower computational overhead, thereby substantiating the efficacy of direct latent space alignment. Our code and checkpoint will be publicly available after acceptance.
Authors: Jiazhen Hong, Geoff Mackellar, Soheila Ghane
Abstract: Electroencephalogram (EEG)-based P300 speller brain-computer interfaces (BCIs) face three main challenges: low signal-to-noise ratio (SNR), poor generalization, and time-consuming calibration. We propose SpellerSSL, a framework that combines self-supervised learning (SSL) with P300 aggregation to address these issues. First, we introduce an aggregation strategy to enhance SNR. Second, to achieve generalization in training, we employ a customized 1D U-Net backbone and pretrain the model on both cross-domain and in-domain EEG data. The pretrained model is subsequently fine-tuned with a lightweight ERP-Head classifier for P300 detection, which adapts the learned representations to subject-specific data. Our evaluations on calibration time demonstrate that combining the aggregation strategy with SSL significantly reduces the calibration burden per subject and improves robustness across subjects. Experimental results show that SSL learns effective EEG representations in both in-domain and cross-domain, with in-domain achieving a state-of-the-art character recognition rate of 94% with only 7 repetitions and the highest information transfer rate (ITR) of 21.86 bits/min on the public II-B dataset. Moreover, in-domain SSL with P300 aggregation reduces the required calibration size by 60% while maintaining a comparable character recognition rate. To the best of our knowledge, this is the first study to apply SSL to P300 spellers, highlighting its potential to improve both efficiency and generalization in speller BCIs and paving the way toward an EEG foundation model for P300 speller BCIs.
Authors: Sheng-Bin Duan, Jian-Long Hao, Tian-Yu Xiang, Xiao-Hu Zhou, Mei-Jiang Gui, Xiao-Liang Xie, Shi-Qi Liu, Zeng-Guang Hou
Abstract: Individual differences in brain activity hinder the online application of electroencephalogram (EEG)-based brain computer interface (BCI) systems. To overcome this limitation, this study proposes an online adaptation algorithm for unseen subjects via dual-stage alignment and self-supervision. The alignment process begins by applying Euclidean alignment in the EEG data space and then updates batch normalization statistics in the representation space. Moreover, a self-supervised loss is designed to update the decoder. The loss is computed by soft pseudo-labels derived from the decoder as a proxy for the unknown ground truth, and is calibrated by Shannon entropy to facilitate self-supervised training. Experiments across five public datasets and seven decoders show the proposed algorithm can be integrated seamlessly regardless of BCI paradigm and decoder architecture. In each iteration, the decoder is updated with a single online trial, which yields average accuracy gains of 4.9% on steady-state visual evoked potentials (SSVEP) and 3.6% on motor imagery. These results support fast-calibration operation and show that the proposed algorithm has great potential for BCI applications.
Authors: Vasilis Andritsoudis, Pavlos Sermpezis, Ilias Dimitriadis, Athena Vakali
Abstract: The Internet Yellow Pages (IYP) aggregates information from multiple sources about Internet routing into a unified, graph-based knowledge base. However, querying it requires knowledge of the Cypher language and the exact IYP schema, thus limiting usability for non-experts. In this paper, we propose ChatIYP, a domain-specific Retrieval-Augmented Generation (RAG) system that enables users to query IYP through natural language questions. Our evaluation demonstrates solid performance on simple queries, as well as directions for improvement, and provides insights for selecting evaluation metrics that are better fit for IYP querying AI agents.
Authors: Rikab Gambhir, Matt LeBlanc, Yuanchen Zhou
Abstract: Classifying hadronic jets using their constituents' kinematic information is a critical task in modern high-energy collider physics. Often, classifiers are designed by targeting the best performance using metrics such as accuracy, AUC, or rejection rates. However, the use of a single metric can lead to the use of architectures that are more model-dependent than competitive alternatives, leading to potential uncertainty and bias in analysis. We explore such trade-offs and demonstrate the consequences of using networks with high performance metrics but low resilience.
Authors: Alessandro Saviolo, Jeffrey Mao, Giuseppe Loianno
Abstract: Search and rescue operations require unmanned aerial vehicles to both traverse unknown unstructured environments at high speed and track targets once detected. Achieving both capabilities under degraded sensing and without global localization remains an open challenge. Recent works on relative navigation have shown robust tracking by anchoring planning and control to a visible detected object, but cannot address navigation when no target is in the field of view. We present HUNT (High-speed UAV Navigation and Tracking), a real-time framework that unifies traversal, acquisition, and tracking within a single relative formulation. HUNT defines navigation objectives directly from onboard instantaneous observables such as attitude, altitude, and velocity, enabling reactive high-speed flight during search. Once a target is detected, the same perception-control pipeline transitions seamlessly to tracking. Outdoor experiments in dense forests, container compounds, and search-and-rescue operations with vehicles and mannequins demonstrate robust autonomy where global methods fail.
Authors: UniverseTBD, :, Kshitij Duraphe, Michael J. Smith, Shashwat Sourav, John F. Wu
Abstract: We test the Platonic Representation Hypothesis (PRH) in astronomy by measuring representational convergence across a range of foundation models trained on different data types. Using spectroscopic and imaging observations from JWST, HSC, Legacy Survey, and DESI, we compare representations from vision transformers, self-supervised models, and astronomy-specific architectures via mutual $k$-nearest neighbour analysis. We observe consistent scaling: representational alignment generally increases with model capacity across our tested architectures, supporting convergence toward a shared representation of galaxy astrophysics. Our results suggest that astronomical foundation models can use pre-trained general-purpose architectures, allowing us to capitalise on the broader machine learning community's already-spent computational investment.
Authors: Jason Chen, I-Chun Arthur Liu, Gaurav Sukhatme, Daniel Seita
Abstract: Training robust bimanual manipulation policies via imitation learning requires demonstration data with broad coverage over robot poses, contacts, and scene contexts. However, collecting diverse and precise real-world demonstrations is costly and time-consuming, which hinders scalability. Prior works have addressed this with data augmentation, typically for either eye-in-hand (wrist camera) setups with RGB inputs or for generating novel images without paired actions, leaving augmentation for eye-to-hand (third-person) RGB-D training with new action labels less explored. In this paper, we propose Synthetic Robot Pose Generation for RGB-D Bimanual Data Augmentation (ROPA), an offline imitation learning data augmentation method that fine-tunes Stable Diffusion to synthesize third-person RGB and RGB-D observations of novel robot poses. Our approach simultaneously generates corresponding joint-space action labels while employing constrained optimization to enforce physical consistency through appropriate gripper-to-object contact constraints in bimanual scenarios. We evaluate our method on 5 simulated and 3 real-world tasks. Our results across 2625 simulation trials and 300 real-world trials demonstrate that ROPA outperforms baselines and ablations, showing its potential for scalable RGB and RGB-D data augmentation in eye-to-hand bimanual manipulation. Our project website is available at: https://ropaaug.github.io/.
Authors: Mert Gurbuzbalaban, Hoang M. Nguyen, Xicheng Zhang, Lingjiong Zhu
Abstract: Standard first-order Langevin algorithms such as the unadjusted Langevin algorithm (ULA) are obtained by discretizing the Langevin diffusion and are widely used for sampling in machine learning because they scale to high dimensions and large datasets. However, they face two key limitations: (i) they require differentiable log-densities, excluding targets with non-differentiable components; and (ii) they generally fail to sample heavy-tailed targets. We propose anchored Langevin dynamics, a unified approach that accommodates non-differentiable targets and certain classes of heavy-tailed distributions. The method replaces the original potential with a smooth reference potential and modifies the Langevin diffusion via multiplicative scaling. We establish non-asymptotic guarantees in the 2-Wasserstein distance to the target distribution and provide an equivalent formulation derived via a random time change of the Langevin diffusion. We provide numerical experiments to illustrate the theory and practical performance of our proposed approach.
Authors: Yifan Ye, Jun Cen, Jing Chen, Zhihe Lu
Abstract: Imitation learning has been a trend recently, yet training a generalist agent across multiple tasks still requires large-scale expert demonstrations, which are costly and labor-intensive to collect. To address the challenge of limited supervision, we propose Self-Evolved Imitation Learning (SEIL), a framework that progressively improves a few-shot model through simulator interactions. The model first attempts tasksin the simulator, from which successful trajectories are collected as new demonstrations for iterative refinement. To enhance the diversity of these demonstrations, SEIL employs dual-level augmentation: (i) Model-level, using an Exponential Moving Average (EMA) model to collaborate with the primary model, and (ii) Environment-level, introducing slight variations in initial object positions. We further introduce a lightweight selector that filters complementary and informative trajectories from the generated pool to ensure demonstration quality. These curated samples enable the model to achieve competitive performance with far fewer training examples. Extensive experiments on the LIBERO benchmark show that SEIL achieves a new state-of-the-art performance in few-shot imitation learning scenarios. Code is available at https://github.com/Jasper-aaa/SEIL.git.
Authors: Shripad Vilasrao Deshmukh, Will Schwarzer, Scott Niekum
Abstract: Policy evaluation is often a prerequisite for deploying safety- and performance-critical systems. Existing evaluation approaches frequently suffer from high variance due to limited data and long-horizon tasks, or high bias due to unequal support or inaccurate environmental models. We posit that these challenges arise, in part, from the standard reinforcement learning (RL) paradigm of policy learning without explicit consideration of evaluation. As an alternative, we propose evaluation-aware reinforcement learning (EvA-RL), in which a policy is trained to maximize expected return while simultaneously minimizing expected evaluation error under a given value prediction scheme -- in other words, being "easy" to evaluate. We formalize a framework for EvA-RL and design an instantiation that enables accurate policy evaluation, conditioned on a small number of rollouts in an assessment environment that can be different than the deployment environment. However, our theoretical analysis and empirical results show that there is often a tradeoff between evaluation accuracy and policy performance when using a fixed value-prediction scheme within EvA-RL. To mitigate this tradeoff, we extend our approach to co-learn an assessment-conditioned state-value predictor alongside the policy. Empirical results across diverse discrete and continuous action domains demonstrate that EvA-RL can substantially reduce evaluation error while maintaining competitive returns. This work lays the foundation for a broad new class of RL methods that treat reliable evaluation as a first-class principle during training.
Authors: Monika Doerfler, Franz Luef, Henry McNulty
Abstract: In this short note, we study the impact of data augmentation on the smoothness of principal components of high-dimensional datasets. Using tools from quantum harmonic analysis, we show that eigenfunctions of operators corresponding to augmented data sets lie in the modulation space $M^1(\mathbb{R}^d)$, guaranteeing smoothness and continuity. Numerical examples on synthetic and audio data confirm the theoretical findings. While interesting in itself, the results suggest that manifold learning and feature extraction algorithms can benefit from systematic and informed augmentation principles.
Authors: Noriaki Hirose, Catherine Glossop, Dhruv Shah, Sergey Levine
Abstract: Humans can flexibly interpret and compose different goal specifications, such as language instructions, spatial coordinates, or visual references, when navigating to a destination. In contrast, most existing robotic navigation policies are trained on a single modality, limiting their adaptability to real-world scenarios where different forms of goal specification are natural and complementary. In this work, we present a training framework for robotic foundation models that enables omni-modal goal conditioning for vision-based navigation. Our approach leverages a high-capacity vision-language-action (VLA) backbone and trains with three primary goal modalities: 2D poses, egocentric images, and natural language, as well as their combinations, through a randomized modality fusion strategy. This design not only expands the pool of usable datasets but also encourages the policy to develop richer geometric, semantic, and visual representations. The resulting model, OmniVLA, achieves strong generalization to unseen environments, robustness to scarce modalities, and the ability to follow novel natural language instructions. We demonstrate that OmniVLA outperforms specialist baselines across modalities and offers a flexible foundation for fine-tuning to new modalities and tasks. We believe OmniVLA provides a step toward broadly generalizable and flexible navigation policies, and a scalable path for building omni-modal robotic foundation models. We present videos showcasing OmniVLA performance and will release its checkpoints and training code on our project page.
Authors: Cem Ashbaugh, Leon Baumg\"artner, Tim Gress, Nikita Sidorov, Daniel Werner
Abstract: Linking implicit scientific claims made on social media to their original publications is crucial for evidence-based fact-checking and scholarly discourse, yet it is hindered by lexical sparsity, very short queries, and domain-specific language. Team AIRwaves ranked second in Subtask 4b of the CLEF-2025 CheckThat! Lab with an evidence-retrieval approach that markedly outperforms the competition baseline. The optimized sparse-retrieval baseline(BM25) achieves MRR@5 = 0.5025 on the gold label blind test set. To surpass this baseline, a two-stage retrieval pipeline is introduced: (i) a first stage that uses a dual encoder based on E5-large, fine-tuned using in-batch and mined hard negatives and enhanced through chunked tokenization and rich document metadata; and (ii) a neural re-ranking stage using a SciBERT cross-encoder. Replacing purely lexical matching with neural representations lifts performance to MRR@5 = 0.6174, and the complete pipeline further improves to MRR@5 = 0.6828. The findings demonstrate that coupling dense retrieval with neural re-rankers delivers a powerful and efficient solution for tweet-to-study matching and provides a practical blueprint for future evidence-retrieval pipelines.
Authors: Sai Teja Reddy Adapala
Abstract: The scaling of Large Language Models (LLMs) has exposed a critical gap between their performance on static benchmarks and their fragility in dynamic, information-rich environments. While models excel at isolated tasks, the computational limits that govern their reasoning under cognitive load remain poorly understood. In this work, we introduce a formal theory of computational cognitive load, positing that extraneous, task-irrelevant information (Context Saturation) and interference from task-switching (Attentional Residue) are key mechanisms that degrade performance. We designed the Interleaved Cognitive Evaluation (ICE), a deconfounded benchmark to systematically manipulate these load factors on challenging multi-hop reasoning tasks. A comprehensive study (N = 10 replications per item across 200 questions) revealed significant performance variations across five instruction-tuned models. Smaller open-source architectures (Llama-3-8B-Instruct, Mistral-7B-Instruct-v0.2) exhibited baseline brittleness, achieving 0% accuracy (SEM = 0.0) across all conditions, including clean controls, on this high-intrinsic-load task. In contrast, Gemini-2.0-Flash-001 showed partial resilience, achieving 85% accuracy in control conditions, with a statistically significant degradation under context saturation ($\beta = -0.003$ per % load, $p < 0.001$). These findings provide preliminary evidence that cognitive load is a key contributor to reasoning failures, supporting theories of hallucination-as-guessing under uncertainty. We conclude that dynamic, cognitive-aware stress testing, as exemplified by the ICE benchmark, is essential for evaluating the true resilience and safety of advanced AI systems.
Authors: Sankalp Agrawal, Junwon Seo, Kensuke Nakamura, Ran Tian, Andrea Bajcsy
Abstract: Recent works have shown that foundational safe control methods, such as Hamilton-Jacobi (HJ) reachability analysis, can be applied in the latent space of world models. While this enables the synthesis of latent safety filters for hard-to-model vision-based tasks, they assume that the safety constraint is known a priori and remains fixed during deployment, limiting the safety filter's adaptability across scenarios. To address this, we propose constraint-parameterized latent safety filters that can adapt to user-specified safety constraints at runtime. Our key idea is to define safety constraints by conditioning on an encoding of an image that represents a constraint, using a latent-space similarity measure. The notion of similarity to failure is aligned in a principled way through conformal calibration, which controls how closely the system may approach the constraint representation. The parameterized safety filter is trained entirely within the world model's imagination, treating any image seen by the model as a potential test-time constraint, thereby enabling runtime adaptation to arbitrary safety constraints. In simulation and hardware experiments on vision-based control tasks with a Franka manipulator, we show that our method adapts at runtime by conditioning on the encoding of user-specified constraint images, without sacrificing performance. Video results can be found on https://any-safe.github.io
Authors: Iris Kamsteeg, Juan Cardenas-Cartagena, Floris van Beers, Gineke ten Holt, Tsegaye Misikir Tashu, Matias Valdenegro-Toro
Abstract: This research aims to explore the intersection of Large Language Models and confidence calibration in Entity Matching. To this end, we perform an empirical study to compare baseline RoBERTa confidences for an Entity Matching task against confidences that are calibrated using Temperature Scaling, Monte Carlo Dropout and Ensembles. We use the Abt-Buy, DBLP-ACM, iTunes-Amazon and Company datasets. The findings indicate that the proposed modified RoBERTa model exhibits a slight overconfidence, with Expected Calibration Error scores ranging from 0.0043 to 0.0552 across datasets. We find that this overconfidence can be mitigated using Temperature Scaling, reducing Expected Calibration Error scores by up to 23.83%.
Authors: Li Zhou, Elvan Ceyhan
Abstract: We introduce the Stochastic Correlated Obstacle Scene (SCOS) problem, a navigation setting with spatially correlated obstacles of uncertain blockage status, realistically constrained sensors that provide noisy readings and costly disambiguation. Modeling the spatial correlation with Gaussian Random Field (GRF), we develop Bayesian belief updates that refine blockage probabilities, and use the posteriors to reduce search space for efficiency. To find the optimal traversal policy, we propose a novel two-stage learning framework. An offline phase learns a robust base policy via optimistic policy iteration augmented with information bonus to encourage exploration in informative regions, followed by an online rollout policy with periodic base updates via a Bayesian mechanism for information adaptation. This framework supports both Monte Carlo point estimation and distributional reinforcement learning (RL) to learn full cost distributions, leading to stronger uncertainty quantification. We establish theoretical benefits of correlation-aware updating and convergence property under posterior sampling. Comprehensive empirical evaluations across varying obstacle densities, sensor capabilities demonstrate consistent performance gains over baselines. This framework addresses navigation challenges in environments with adversarial interruptions or clustered natural hazards.
Authors: Stefania Radu, Marco Zullich, Matias Valdenegro-Toro
Abstract: Pixel-based language models aim to solve the vocabulary bottleneck problem in language modeling, but the challenge of uncertainty quantification remains open. The novelty of this work consists of analysing uncertainty and confidence in pixel-based language models across 18 languages and 7 scripts, all part of 3 semantically challenging tasks. This is achieved through several methods such as Monte Carlo Dropout, Transformer Attention, and Ensemble Learning. The results suggest that pixel-based models underestimate uncertainty when reconstructing patches. The uncertainty is also influenced by the script, with Latin languages displaying lower uncertainty. The findings on ensemble learning show better performance when applying hyperparameter tuning during the named entity recognition and question-answering tasks across 16 languages.
Authors: Dohyun Ku, Catherine D. Chong, Visar Berisha, Todd J. Schwedt, Jing Li
Abstract: Time series analysis has emerged as an important tool for improving patient diagnosis and management in healthcare applications. However, these applications commonly face two critical challenges: time misalignment and data sparsity. Traditional approaches address these issues through a two-step process of imputation followed by prediction. We propose MAGIC (Multi-tAsk Gaussian Process for Imputation and Classification), a novel unified framework that simultaneously performs class-informed missing value imputation and label prediction within a hierarchical multi-task Gaussian process coupled with functional logistic regression. To handle intractable likelihood components, MAGIC employs Taylor expansion approximations with bounded error analysis, and parameter estimation is performed using EM algorithm with block coordinate optimization supported by convergence analysis. We validate MAGIC through two healthcare applications: prediction of post-traumatic headache improvement following mild traumatic brain injury and prediction of in-hospital mortality within 48 hours after ICU admission. In both applications, MAGIC achieves superior predictive accuracy compared to existing methods. The ability to generate real-time and accurate predictions with limited samples facilitates early clinical assessment and treatment planning, enabling healthcare providers to make more informed treatment decisions.
Authors: Adrien Goldszal, Diego Calanzone, Vincent Taboga, Pierre-Luc Bacon
Abstract: Most refrigerants currently used in air-conditioning systems, such as hydrofluorocarbons, are potent greenhouse gases and are being phased down. Large-scale molecular screening has been applied to the search for alternatives, but in practice only about 300 refrigerants are known, and only a few additional candidates have been suggested without experimental validation. This scarcity of reliable data limits the effectiveness of purely data-driven methods. We present Refgen, a generative pipeline that integrates machine learning with physics-grounded inductive biases. Alongside fine-tuning for valid molecular generation, Refgen incorporates predictive models for critical properties, equations of state, thermochemical polynomials, and full vapor compression cycle simulations. These models enable reinforcement learning fine-tuning under thermodynamic constraints, enforcing consistency and guiding discovery toward molecules that balance efficiency, safety, and environmental impact. By embedding physics into the learning process, Refgen leverages scarce data effectively and enables de novo refrigerant discovery beyond the known set of compounds.
Authors: Nathanael Jo, Ashia Wilson
Abstract: Evaluations of generative models on benchmark data are now ubiquitous, and their outcomes critically shape public and scientific expectations of AI's capabilities. Yet growing skepticism surrounds their reliability. How can we know that a reported accuracy genuinely reflects a model's true performance? Evaluations are often presented as simple measurements, but in reality they are inferences: to treat benchmark scores as evidence of capability is already to assume a theory of what capability is and how it manifests in a test. We make this step explicit by proposing a principled framework for evaluation as inference: begin from a theory of capability, and then derive methods for estimating it. This perspective, familiar in fields such as psychometrics, has not yet become commonplace in AI evaluation. As a proof of concept, we address a central challenge that undermines reliability: sensitivity to perturbations. After formulating a model of ability, we introduce methods that infer ability while accounting for uncertainty from sensitivity and finite samples, including an adaptive algorithm that significantly reduces sample complexity. Together, these contributions lay the groundwork for more reliable and trustworthy estimates of AI capabilities as measured through benchmarks.
Authors: Daniel Holmberg, Ivan Zaitsev, Markku Alho, Ioanna Bouri, Fanni Franssila, Haewon Jeong, Minna Palmroth, Teemu Roos
Abstract: Accurate space weather forecasting is crucial for protecting our increasingly digital infrastructure. Hybrid-Vlasov models, like Vlasiator, offer physical realism beyond that of current operational systems, but are too computationally expensive for real-time use. We introduce a graph-based neural emulator trained on Vlasiator data to autoregressively predict near-Earth space conditions driven by an upstream solar wind. We show how to achieve both fast deterministic forecasts and, by using a generative model, produce ensembles to capture forecast uncertainty. This work demonstrates that machine learning offers a way to add uncertainty quantification capability to existing space weather prediction systems, and make hybrid-Vlasov simulation tractable for operational use.
Authors: Ryan Punamiya, Dhruv Patel, Patcharapong Aphiwetsa, Pranav Kuppili, Lawrence Y. Zhu, Simar Kareer, Judy Hoffman, Danfei Xu
Abstract: Egocentric human experience data presents a vast resource for scaling up end-to-end imitation learning for robotic manipulation. However, significant domain gaps in visual appearance, sensor modalities, and kinematics between human and robot impede knowledge transfer. This paper presents EgoBridge, a unified co-training framework that explicitly aligns the policy latent spaces between human and robot data using domain adaptation. Through a measure of discrepancy on the joint policy latent features and actions based on Optimal Transport (OT), we learn observation representations that not only align between the human and robot domain but also preserve the action-relevant information critical for policy learning. EgoBridge achieves a significant absolute policy success rate improvement by 44% over human-augmented cross-embodiment baselines in three real-world single-arm and bimanual manipulation tasks. EgoBridge also generalizes to new objects, scenes, and tasks seen only in human data, where baselines fail entirely. Videos and additional information can be found at https://ego-bridge.github.io
Authors: Manuel Perez-Carrasco, Maya Nasr, Sebastien Roche, Chris Chan Miller, Zhan Zhang, Core Francisco Park, Eleanor Walker, Cecilia Garraffo, Douglas Finkbeiner, Ritesh Gautam, Steven Wofsy
Abstract: Effective cloud and cloud shadow detection is a critical prerequisite for accurate retrieval of concentrations of atmospheric methane or other trace gases in hyperspectral remote sensing. This challenge is especially pertinent for MethaneSAT and for its airborne companion mission, MethaneAIR. In this study, we use machine learning methods to address the cloud and cloud shadow detection problem for sensors with these high spatial resolutions instruments. Cloud and cloud shadows in remote sensing data need to be effectively screened out as they bias methane retrievals in remote sensing imagery and impact the quantification of emissions. We deploy and evaluate conventional techniques including Iterative Logistic Regression (ILR) and Multilayer Perceptron (MLP), with advanced deep learning architectures, namely UNet and a Spectral Channel Attention Network (SCAN) method. Our results show that conventional methods struggle with spatial coherence and boundary definition, affecting the detection of clouds and cloud shadows. Deep learning models substantially improve detection quality: UNet performs best in preserving spatial structure, while SCAN excels at capturing fine boundary details. Notably, SCAN surpasses UNet on MethaneSAT data, underscoring the benefits of incorporating spectral attention for satellite specific features. This in depth assessment of various disparate machine learning techniques demonstrates the strengths and effectiveness of advanced deep learning architectures in providing robust, scalable solutions for clouds and cloud shadow screening towards enhancing methane emission quantification capacity of existing and next generation hyperspectral missions. Our data and code is publicly available at https://doi.org/10.7910/DVN/IKLZOJ
Authors: Nam Ho-Nguyen, Fatma K{\i}l{\i}n\c{c}-Karzan, Ellie Nguyen, Lingqing Shen
Abstract: Online classification is a central problem in optimization, statistical learning and data science. Classical algorithms such as the perceptron offer efficient updates and finite mistake guarantees on linearly separable data, but they do not exploit the underlying geometric structure of the classification problem. We study the offline maximum margin problem through its dual formulation and use the resulting geometric insights to design a principled and efficient algorithm for the online setting. A key feature of our method is its translation invariance, inherited from the offline formulation, which plays a central role in its performance analysis. Our theoretical analysis yields improved mistake and margin bounds that depend only on translation-invariant quantities, offering stronger guarantees than existing algorithms under the same assumptions in favorable settings. In particular, we identify a parameter regime where our algorithm makes at most two mistakes per sequence, whereas the perceptron can be forced to make arbitrarily many mistakes. Our numerical study on real data further demonstrates that our method matches the computational efficiency of existing online algorithms, while significantly outperforming them in accuracy.
Authors: Prateek Verma, Mert Pilanci
Abstract: We propose a framework that enables neural models to "think while listening" to everyday sounds, thereby enhancing audio classification performance. Motivated by recent advances in the reasoning capabilities of large language models, we address two central questions: (i) how can thinking be incorporated into existing audio classification pipelines to enable reasoning in the category space and improve performance, and (ii) can a new architecture be designed from the ground up to support both thinking and test-time scaling? We demonstrate that in both settings, our models exhibit improved classification accuracy. Leveraging test-time scaling, we observe consistent gains as the number of sampled traces increases. Furthermore, we evaluate two open-source reasoning models, GPT-OSS-20B and Qwen3-14B, showing that while such models are capable of zero-shot reasoning, a lightweight approach--retraining only the embedding matrix of a frozen, smaller model like GPT-2--can surpass the performance of billion-parameter text-based reasoning models.
Authors: Devesh Nath, Haoran Yin, Glen Chou
Abstract: We present a method for formal safety verification of learning-based generative motion planners. Generative motion planners (GMPs) offer advantages over traditional planners, but verifying the safety and dynamic feasibility of their outputs is difficult since neural network verification (NNV) tools scale only to a few hundred neurons, while GMPs often contain millions. To preserve GMP expressiveness while enabling verification, our key insight is to imitate the GMP by stabilizing references sampled from the GMP with a small neural tracking controller and then applying NNV to the closed-loop dynamics. This yields reachable sets that rigorously certify closed-loop safety, while the controller enforces dynamic feasibility. Building on this, we construct a library of verified GMP references and deploy them online in a way that imitates the original GMP distribution whenever it is safe to do so, improving safety without retraining. We evaluate across diverse planners, including diffusion, flow matching, and vision-language models, improving safety in simulation (on ground robots and quadcopters) and on hardware (differential-drive robot).
Authors: Noah Geiger, Tamim Asfour, Neville Hogan, Johannes Lachner
Abstract: Learning methods excel at motion generation in the information domain but are not primarily designed for physical interaction in the energy domain. Impedance Control shapes physical interaction but requires task-aware tuning by selecting feasible impedance parameters. We present Diffusion-Based Impedance Learning, a framework that combines both domains. A Transformer-based Diffusion Model with cross-attention to external wrenches reconstructs a simulated Zero-Force Trajectory (sZFT). This captures both translational and rotational task-space behavior. For rotations, we introduce a novel SLERP-based quaternion noise scheduler that ensures geometric consistency. The reconstructed sZFT is then passed to an energy-based estimator that updates stiffness and damping parameters. A directional rule is applied that reduces impedance along non task axes while preserving rigidity along task directions. Training data were collected for a parkour scenario and robotic-assisted therapy tasks using teleoperation with Apple Vision Pro. With only tens of thousands of samples, the model achieved sub-millimeter positional accuracy and sub-degree rotational accuracy. Its compact model size enabled real-time torque control and autonomous stiffness adaptation on a KUKA LBR iiwa robot. The controller achieved smooth parkour traversal within force and velocity limits and 30/30 success rates for cylindrical, square, and star peg insertions without any peg-specific demonstrations in the training data set. All code for the Transformer-based Diffusion Model, the robot controller, and the Apple Vision Pro telemanipulation framework is publicly available. These results mark an important step towards Physical AI, fusing model-based control for physical interaction with learning-based methods for trajectory generation.
Authors: David Huk, Theodoros Damoulas
Abstract: Copulas are a fundamental tool for modelling multivariate dependencies in data, forming the method of choice in diverse fields and applications. However, the adoption of existing models for multimodal and high-dimensional dependencies is hindered by restrictive assumptions and poor scaling. In this work, we present methods for modelling copulas based on the principles of diffusions and flows. We design two processes that progressively forget inter-variable dependencies while leaving dimension-wise distributions unaffected, provably defining valid copulas at all times. We show how to obtain copula models by learning to remember the forgotten dependencies from each process, theoretically recovering the true copula at optimality. The first instantiation of our framework focuses on direct density estimation, while the second specialises in expedient sampling. Empirically, we demonstrate the superior performance of our proposed methods over state-of-the-art copula approaches in modelling complex and high-dimensional dependencies from scientific datasets and images. Our work enhances the representational power of copula models, empowering applications and paving the way for their adoption on larger scales and more challenging domains.
Authors: Anand Kumar, Vishal Khare, Deepak Sharma, Satyam Kumar, Vijay Saini, Anshul Yadav, Sachendra Jain, Ankit Rana, Pratham Verma, Vaibhav Meena, Avinash Edubilli
Abstract: We present a comprehensive real-world evaluation of AI-assisted software development tools deployed at enterprise scale. Over one year, 300 engineers across multiple teams integrated an in-house AI platform (DeputyDev) that combines code generation and automated review capabilities into their daily workflows. Through rigorous cohort analysis, our study demonstrates statistically significant productivity improvements, including an overall 31.8% reduction in PR review cycle time. Developer adoption was strong, with 85% satisfaction for code review features and 93% expressing a desire to continue using the platform. Adoption patterns showed systematic scaling from 4% engagement in month 1 to 83% peak usage by month 6, stabilizing at 60% active engagement. Top adopters achieved a 61% increase in code volume pushed to production, contributing to approximately 30 to 40% of code shipped to production through this tool, accounting for an overall 28% increase in code shipment volume. Unlike controlled benchmark evaluations, our longitudinal analysis provides empirical evidence from production environments, revealing both the transformative potential and practical deployment challenges of integrating AI into enterprise software development workflows.
Authors: Cheng Qian, Zuxin Liu, Akshara Prabhakar, Jielin Qiu, Zhiwei Liu, Haolin Chen, Shirley Kokane, Heng Ji, Weiran Yao, Shelby Heinecke, Silvio Savarese, Caiming Xiong, Huan Wang
Abstract: Reinforcement learning (RL) has shown promise in training agentic models that move beyond static benchmarks to engage in dynamic, multi-turn interactions. Yet, the ultimate value of such agents lies in their ability to assist users, a setting where diversity and dynamics of user interaction pose challenges. In this work, we propose UserRL, a unified framework for training and evaluating user-centric abilities through standardized gym environments paired with simulated users. We systematically vary turn-level reward assignment and trajectory-level score calculation to analyze how different formulations affect learning under the GRPO algorithm. Our experiments across Qwen3 models reveal three key findings: (i) SFT cold start is critical for unlocking initial interaction ability and enabling sustained RL improvements; (ii) deliberate trajectory scoring yields more efficient and effective multi-turn interactions; and (iii) while stronger simulated users (e.g., GPT-4o) facilitates training, open-source simulators (e.g., Qwen3-32B) remain a cost-effective and transferable option. Together, these results highlight that careful design of reward shaping and user simulation choice is as crucial as model scale, and establish UserRL as a practical pathway for developing robust user-centric agentic models. All codes and data are public for future research.
Authors: Eunji Lim
Abstract: A common way to estimate an unknown convex regression function $f_0: \Omega \subset \mathbb{R}^d \rightarrow \mathbb{R}$ from a set of $n$ noisy observations is to fit a convex function that minimizes the sum of squared errors. However, this estimator is known for its tendency to overfit near the boundary of $\Omega$, posing significant challenges in real-world applications. In this paper, we introduce a new estimator of $f_0$ that avoids this overfitting by minimizing a penalty on the subgradient while enforcing an upper bound $s_n$ on the sum of squared errors. The key advantage of this method is that $s_n$ can be directly estimated from the data. We establish the uniform almost sure consistency of the proposed estimator and its subgradient over $\Omega$ as $n \rightarrow \infty$ and derive convergence rates. The effectiveness of our estimator is illustrated through its application to estimating waiting times in a single-server queue.
Authors: Burak I. Tas, Enrique del Castillo
Abstract: We address the Statistical Process Control (SPC) of high-dimensional, dynamic industrial processes from two complementary perspectives: manifold fitting and manifold learning, both of which assume data lies on an underlying nonlinear, lower dimensional space. We propose two distinct monitoring frameworks for online or 'phase II' Statistical Process Control (SPC). The first method leverages state-of-the-art techniques in manifold fitting to accurately approximate the manifold where the data resides within the ambient high-dimensional space. It then monitors deviations from this manifold using a novel scalar distribution-free control chart. In contrast, the second method adopts a more traditional approach, akin to those used in linear dimensionality reduction SPC techniques, by first embedding the data into a lower-dimensional space before monitoring the embedded observations. We prove how both methods provide a controllable Type I error probability, after which they are contrasted for their corresponding fault detection ability. Extensive numerical experiments on a synthetic process and on a replicated Tennessee Eastman Process show that the conceptually simpler manifold-fitting approach achieves performance competitive with, and sometimes superior to, the more classical lower-dimensional manifold monitoring methods. In addition, we demonstrate the practical applicability of the proposed manifold-fitting approach by successfully detecting surface anomalies in a real image dataset of electrical commutators.
Authors: Teruki Kato, Ryotaro Shima, Kenji Kashima
Abstract: Deep learning is increasingly used for complex, large-scale systems where first-principles modeling is difficult. However, standard deep learning models often fail to enforce physical structure or preserve convexity in downstream control, leading to physically inconsistent predictions and discontinuous inputs owing to nonconvexity. We introduce sign constraints--sign restrictions on Jacobian entries--that unify monotonicity, positivity, and sign-definiteness; additionally, we develop model-construction methods that enforce them, together with a control-synthesis procedure. In particular, we design exactly linearizable deep models satisfying these constraints and formulate model predictive control as a convex quadratic program, which yields a unique optimizer and a Lipschitz continuous control law. On a two-tank system and a hybrid powertrain, the proposed approach improves prediction accuracy and produces smoother control inputs than existing methods.
Authors: Arnaud Vadeboncoeur, Gregory Duth\'e, Mark Girolami, Eleni Chatzi
Abstract: Uncertainty Quantification (UQ) is paramount for inference in engineering applications. A common inference task is to recover full-field information of physical systems from a small number of noisy observations, a usually highly ill-posed problem. Critically, engineering systems often have complicated and variable geometries prohibiting the use of standard Bayesian UQ. In this work, we introduce Geometric Autoencoders for Bayesian Inversion (GABI), a framework for learning geometry-aware generative models of physical responses that serve as highly informative geometry-conditioned priors for Bayesian inversion. Following a ''learn first, observe later'' paradigm, GABI distills information from large datasets of systems with varying geometries, without requiring knowledge of governing PDEs, boundary conditions, or observation processes, into a rich latent prior. At inference time, this prior is seamlessly combined with the likelihood of the specific observation process, yielding a geometry-adapted posterior distribution. Our proposed framework is architecture agnostic. A creative use of Approximate Bayesian Computation (ABC) sampling yields an efficient implementation that utilizes modern GPU hardware. We test our method on: steady-state heat over rectangular domains; Reynold-Averaged Navier-Stokes (RANS) flow around airfoils; Helmholtz resonance and source localization on 3D car bodies; RANS airflow over terrain. We find: the predictive accuracy to be comparable to deterministic supervised learning approaches in the restricted setting where supervised learning is applicable; UQ to be well calibrated and robust on challenging problems with complex geometries. The method provides a flexible geometry-aware train-once-use-anywhere foundation model which is independent of any particular observation process.
Authors: Yanke Li, Tianyu Cui, Tommaso Mansi, Mangal Prakash, Rui Liao
Abstract: Efficient design of genomic perturbation experiments is crucial for accelerating drug discovery and therapeutic target identification, yet exhaustive perturbation of the human genome remains infeasible due to the vast search space of potential genetic interactions and experimental constraints. Bayesian optimization (BO) has emerged as a powerful framework for selecting informative interventions, but existing approaches often fail to exploit domain-specific biological prior knowledge. We propose Biology-Informed Bayesian Optimization (BioBO), a method that integrates Bayesian optimization with multimodal gene embeddings and enrichment analysis, a widely used tool for gene prioritization in biology, to enhance surrogate modeling and acquisition strategies. BioBO combines biologically grounded priors with acquisition functions in a principled framework, which biases the search toward promising genes while maintaining the ability to explore uncertain regions. Through experiments on established public benchmarks and datasets, we demonstrate that BioBO improves labeling efficiency by 25-40%, and consistently outperforms conventional BO by identifying top-performing perturbations more effectively. Moreover, by incorporating enrichment analysis, BioBO yields pathway-level explanations for selected perturbations, offering mechanistic interpretability that links designs to biologically coherent regulatory circuits.
Authors: Nico Schulthess, Ender Konukoglu
Abstract: In this work, we leverage informative embeddings from foundational models for unsupervised anomaly detection in medical imaging. For small datasets, a memory-bank of normative features can directly be used for anomaly detection which has been demonstrated recently. However, this is unsuitable for large medical datasets as the computational burden increases substantially. Therefore, we propose to model the distribution of normative DINOv2 embeddings with a Dirichlet Process Mixture model (DPMM), a non-parametric mixture model that automatically adjusts the number of mixture components to the data at hand. Rather than using a memory bank, we use the similarity between the component centers and the embeddings as anomaly score function to create a coarse anomaly segmentation mask. Our experiments show that through DPMM embeddings of DINOv2, despite being trained on natural images, achieve very competitive anomaly detection performance on medical imaging benchmarks and can do this while at least halving the computation time at inference. Our analysis further indicates that normalized DINOv2 embeddings are generally more aligned with anatomical structures than unnormalized features, even in the presence of anomalies, making them great representations for anomaly detection. The code is available at https://github.com/NicoSchulthess/anomalydino-dpmm.
Authors: Somraj Gautam, Nachiketa Purohit, Gaurav Harit
Abstract: Efficient data annotation remains a critical challenge in machine learning, particularly for object detection tasks requiring extensive labeled data. Active learning (AL) has emerged as a promising solution to minimize annotation costs by selecting the most informative samples. While traditional AL approaches primarily rely on uncertainty-based selection, recent advances suggest that incorporating diversity-based strategies can enhance sampling efficiency in object detection tasks. Our approach ensures the selection of representative examples that improve model generalization. We evaluate our method on two benchmark datasets (TableBank-LaTeX, TableBank-Word) using state-of-the-art table detection architectures, CascadeTabNet and YOLOv9. Our results demonstrate that AL-based example selection significantly outperforms random sampling, reducing annotation effort given a limited budget while maintaining comparable performance to fully supervised models. Our method achieves higher mAP scores within the same annotation budget.
Authors: Maurice Wenig, Paul G. Rump, Mark Blacher, Joachim Giesen
Abstract: In 2011, einsum was introduced to NumPy as a practical and convenient notation for tensor expressions in machine learning, quantum circuit simulation, and other fields. It has since been implemented in additional Python frameworks such as PyTorch and TensorFlow, as well as in other programming languages such as Julia. Despite its practical success, the einsum notation still lacks a solid theoretical basis, and is not unified across the different frameworks, limiting opportunities for formal reasoning and systematic optimization. In this work, we discuss the terminology of tensor expressions and provide a formal definition of the einsum language. Based on this definition, we formalize and prove important equivalence rules for tensor expressions and highlight their relevance in practical applications.
Authors: Cas Steigstra, Sergey Milyaev, Shaodi You
Abstract: The reliability of secure graphic verification, a key anti-counterfeiting tool, is undermined by poor image acquisition on smartphones. Uncontrolled user captures of these high-entropy patterns cause high false rejection rates, creating a significant 'reliability gap'. To bridge this gap, we depart from traditional perceptual IQA and introduce a framework that predictively estimates a frame's utility for the downstream verification task. We propose a lightweight model to predict a quality score for a video frame, determining its suitability for a resource-intensive oracle model. Our framework is validated using re-contextualized FNMR and ISRR metrics on a large-scale dataset of 32,000+ images from 105 smartphones. Furthermore, a novel cross-domain analysis on graphics from different industrial printing presses reveals a key finding: a lightweight probe on a frozen, ImageNet-pretrained network generalizes better to an unseen printing technology than a fully fine-tuned model. This provides a key insight for real-world generalization: for domain shifts from physical manufacturing, a frozen general-purpose backbone can be more robust than full fine-tuning, which can overfit to source-domain artifacts.
Authors: Alastair Poole, Stig McArthur, Saravan Kumar
Abstract: Kolmogorov-Arnold Networks (KANs) relocate learnable nonlinearities from nodes to edges, demonstrating remarkable capabilities in scientific machine learning and interpretable modeling. However, current KAN implementations suffer from fundamental inefficiencies due to redundancy in high-dimensional spline parameter spaces, where numerous distinct parameterisations yield functionally equivalent behaviors. This redundancy manifests as a "nuisance space" in the model's Jacobian, leading to susceptibility to overfitting and poor generalization. We introduce Projective Kolmogorov-Arnold Networks (P-KANs), a novel training framework that guides edge function discovery towards interpretable functional representations through entropy-minimisation techniques from signal analysis and sparse dictionary learning. Rather than constraining functions to predetermined spaces, our approach maintains spline space flexibility while introducing "gravitational" terms that encourage convergence towards optimal functional representations. Our key insight recognizes that optimal representations can be identified through entropy analysis of projection coefficients, compressing edge functions to lower-parameter projective spaces (Fourier, Chebyshev, Bessel). P-KANs demonstrate superior performance across multiple domains, achieving up to 80% parameter reduction while maintaining representational capacity, significantly improved robustness to noise compared to standard KANs, and successful application to industrial automated fiber placement prediction. Our approach enables automatic discovery of mixed functional representations where different edges converge to different optimal spaces, providing both compression benefits and enhanced interpretability for scientific machine learning applications.
Authors: Bilal Dalgic (Manisa Celal Bayar University, Turkey), Betul Sen (Manisa Celal Bayar University, Turkey), Muge Erel-Ozcevik (Manisa Celal Bayar University, Turkey)
Abstract: Secure monitoring and dynamic control in an IIoT environment are major requirements for current development goals. We believe that dynamic, secure monitoring of the IIoT environment can be achieved through integration with the Software-Defined Network (SDN) and Digital Twin (DT) paradigms. The current literature lacks implementation details for SDN-based DT and time-aware intelligent model training for short-term anomaly detection against IIoT threats. Therefore, we have proposed a novel framework for short-term anomaly detection that uses an SDN-based DT. Using a comprehensive dataset, time-aware labeling of features, and a comprehensive evaluation of various machine learning models, we propose a novel SD-TWIN-based anomaly detection algorithm. According to the performance of a new real-time SD-TWIN deployment, the GPU- accelerated LightGBM model is particularly effective, achieving a balance of high recall and strong classification performance.
Authors: Matteo Benati, Alessandro Londei, Denise Lanzieri, Vittorio Loreto
Abstract: Extinction times in resampling processes are fundamental yet often intractable, as previous formulas scale as $2^M$ with the number of states $M$ present in the initial probability distribution. We solve this by treating multinomial updates as independent square-root diffusions of zero drift, yielding a closed-form law for the first-extinction time. We prove that the mean coincides exactly with the Wright-Fisher result of Baxter et al., thereby replacing exponential-cost evaluations with a linear-cost expression, and we validate this result through extensive simulations. Finally, we demonstrate predictive power for model collapse in a simple self-training setup: the onset of collapse coincides with the resampling-driven first-extinction time computed from the model's initial stationary distribution. These results hint to a unified view of resampling extinction dynamics.
Authors: JuanaJuana Valeria Hurtado, Rohit Mohan, Abhinav Valada
Abstract: Hyperspectral imaging (HSI) captures spatial information along with dense spectral measurements across numerous narrow wavelength bands. This rich spectral content has the potential to facilitate robust robotic perception, particularly in environments with complex material compositions, varying illumination, or other visually challenging conditions. However, current HSI semantic segmentation methods underperform due to their reliance on architectures and learning frameworks optimized for RGB inputs. In this work, we propose a novel hyperspectral adapter that leverages pretrained vision foundation models to effectively learn from hyperspectral data. Our architecture incorporates a spectral transformer and a spectrum-aware spatial prior module to extract rich spatial-spectral features. Additionally, we introduce a modality-aware interaction block that facilitates effective integration of hyperspectral representations and frozen vision Transformer features through dedicated extraction and injection mechanisms. Extensive evaluations on three benchmark autonomous driving datasets demonstrate that our architecture achieves state-of-the-art semantic segmentation performance while directly using HSI inputs, outperforming both vision-based and hyperspectral segmentation methods. We make the code available at https://hyperspectraladapter.cs.uni-freiburg.de.
Authors: Jarne Mathi Decker
Abstract: The "No Free Lunch" theorem dictates that no single recommender algorithm is optimal for all users, creating a significant Algorithm Selection Problem. Standard meta-learning approaches aim to solve this by selecting an algorithm based on user features, but treat the fundamentally diverse algorithms themselves as equivalent, "black-box" choices. This thesis investigates the impact of overcoming this limitation by engineering a comprehensive feature set to explicitly characterize the algorithms themselves. We combine static code metrics, Abstract Syntax Tree properties, behavioral performance landmarks, and high-level conceptual features. We evaluate two meta-learners across five datasets: a baseline using only user features and our proposed model using both user and algorithm features. Our results show that the meta-learner augmented with algorithm features achieves an average NDCG@10 of 0.143, a statistically significant improvement of 11.7% over the Single Best Algorithm baseline (0.128). However, we found that the inclusion of algorithm features did not lead to an improvement in overall NDCG@10 over the meta learner using only user features (0.144). While adding algorithm features to the meta-learner did improve its Top-1 selection accuracy (+16.1%), this was counterbalanced by leading to a lower Top-3 accuracy (-10.7%). We conclude that for the per-user algorithm selection task in recommender systems, the predictive power of user features is overwhelmingly dominant. While algorithm features improve selection precision, unlocking their potential to boost overall performance remains a non-trivial challenge.
Authors: Siddharth Chandak, Ilai Bistritz, Nicholas Bambos
Abstract: Consider N players and K games taking place simultaneously. Each of these games is modeled as a Tug-of-War (ToW) game where increasing the action of one player decreases the reward for all other players. Each player participates in only one game at any given time. At each time step, a player decides the game in which they wish to participate in and the action they take in that game. Their reward depends on the actions of all players that are in the same game. This system of K games is termed `Meta Tug-of-War' (Meta-ToW) game. These games can model scenarios such as power control, distributed task allocation, and activation in sensor networks. We propose the Meta Tug-of-Peace algorithm, a distributed algorithm where the action updates are done using a simple stochastic approximation algorithm, and the decision to switch games is made using an infrequent 1-bit communication between the players. We prove that in Meta-ToW games, our algorithm converges to an equilibrium that satisfies a target Quality of Service reward vector for the players. We then demonstrate the efficacy of our algorithm through simulations for the scenarios mentioned above.
Authors: Daniel Maninger, Leon Chemnitz, Amir Molzam Sharifloo, Jannis Brugger, Mira Mezini
Abstract: API integration is a cornerstone of our digital infrastructure, enabling software systems to connect and interact. However, as shown by many studies, writing or generating correct code to invoke APIs, particularly web APIs, is challenging. Although large language models~(LLMs) have become popular in software development, their effectiveness in automating the generation of web API integration code remains unexplored. In order to address this, we present a dataset and evaluation pipeline designed to assess the ability of LLMs to generate web API invocation code. Our experiments with several open-source LLMs reveal that generating API invocations poses a significant challenge, resulting in hallucinated endpoints, incorrect argument usage, and other errors. None of the evaluated open-source models were able to solve more than 40% of the tasks.
Authors: Liang Wang, Nan Yang, Shaohan Huang, Li Dong, Furu Wei
Abstract: This paper introduces a simple and scalable approach to improve the data efficiency of large language model (LLM) training by augmenting existing text data with thinking trajectories. The compute for pre-training LLMs has been growing at an unprecedented rate, while the availability of high-quality data remains limited. Consequently, maximizing the utility of available data constitutes a significant research challenge. A primary impediment is that certain high-quality tokens are difficult to learn given a fixed model capacity, as the underlying rationale for a single token can be exceptionally complex and deep. To address this issue, we propose Thinking augmented Pre-Training (TPT), a universal methodology that augments text with automatically generated thinking trajectories. Such augmentation effectively increases the volume of the training data and makes high-quality tokens more learnable through step-by-step reasoning and decomposition. We apply TPT across diverse training configurations up to $100$B tokens, encompassing pre-training with both constrained and abundant data, as well as mid-training from strong open-source checkpoints. Experimental results indicate that our method substantially improves the performance of LLMs across various model sizes and families. Notably, TPT enhances the data efficiency of LLM pre-training by a factor of $3$. For a $3$B parameter model, it improves the post-training performance by over $10\%$ on several challenging reasoning benchmarks.
Authors: Aleksandra Jekic, Afroditi Natsaridou, Signe Riemer-S{\o}rensen, Helge Langseth, Odd Erik Gundersen
Abstract: Approximating solutions to partial differential equations (PDEs) is fundamental for the modeling of dynamical systems in science and engineering. Physics-informed neural networks (PINNs) are a recent machine learning-based approach, for which many properties and limitations remain unknown. PINNs are widely accepted as inferior to traditional methods for solving PDEs, such as the finite element method, both with regard to computation time and accuracy. However, PINNs are commonly claimed to show promise in solving inverse problems and handling noisy or incomplete data. We compare the performance of PINNs in solving inverse problems with that of a traditional approach using the finite element method combined with a numerical optimizer. The models are tested on a series of increasingly difficult fluid mechanics problems, with and without noise. We find that while PINNs may require less human effort and specialized knowledge, they are outperformed by the traditional approach. However, the difference appears to decrease with higher dimensions and more data. We identify common failures during training to be addressed if the performance of PINNs on noisy inverse problems is to become more competitive.
Authors: Dehong Kong, Sifan Yu, Siyuan Liang, Jiawei Liang, Jianhou Gan, Aishan Liu, Wenqi Ren
Abstract: Visual language modeling for automated driving is emerging as a promising research direction with substantial improvements in multimodal reasoning capabilities. Despite its advanced reasoning abilities, VLM-AD remains vulnerable to serious security threats from adversarial attacks, which involve misleading model decisions through carefully crafted perturbations. Existing attacks have obvious challenges: 1) Physical adversarial attacks primarily target vision modules. They are difficult to directly transfer to VLM-AD systems because they typically attack low-level perceptual components. 2) Adversarial attacks against VLM-AD have largely concentrated on the digital level. To address these challenges, we propose the first Universal Camouflage Attack (UCA) framework for VLM-AD. Unlike previous methods that focus on optimizing the logit layer, UCA operates in the feature space to generate physically realizable camouflage textures that exhibit strong generalization across different user commands and model architectures. Motivated by the observed vulnerability of encoder and projection layers in VLM-AD, UCA introduces a feature divergence loss (FDL) that maximizes the representational discrepancy between clean and adversarial images. In addition, UCA incorporates a multi-scale learning strategy and adjusts the sampling ratio to enhance its adaptability to changes in scale and viewpoint diversity in real-world scenarios, thereby improving training stability. Extensive experiments demonstrate that UCA can induce incorrect driving commands across various VLM-AD models and driving scenarios, significantly surpassing existing state-of-the-art attack methods (improving 30\% in 3-P metrics). Furthermore, UCA exhibits strong attack robustness under diverse viewpoints and dynamic conditions, indicating high potential for practical deployment.
Authors: Mohamed Manzour, Catherine M. Elias, Omar M. Shehata, Rub\'en Izquierdo, Miguel \'Angel Sotelo
Abstract: Research on lane change prediction has gained attention in the last few years. Most existing works in this area have been conducted in simulation environments or with pre-recorded datasets, these works often rely on simplified assumptions about sensing, communication, and traffic behavior that do not always hold in practice. Real-world deployments of lane-change prediction systems are relatively rare, and when they are reported, the practical challenges, limitations, and lessons learned are often under-documented. This study explores cooperative lane-change prediction through a real hardware deployment in mixed traffic and shares the insights that emerged during implementation and testing. We highlight the practical challenges we faced, including bottlenecks, reliability issues, and operational constraints that shaped the behavior of the system. By documenting these experiences, the study provides guidance for others working on similar pipelines.
Authors: Tom Burgert, Oliver Stoll, Paolo Rota, Beg\"um Demir
Abstract: The hypothesis that Convolutional Neural Networks (CNNs) are inherently texture-biased has shaped much of the discourse on feature use in deep learning. We revisit this hypothesis by examining limitations in the cue-conflict experiment by Geirhos et al. To address these limitations, we propose a domain-agnostic framework that quantifies feature reliance through systematic suppression of shape, texture, and color cues, avoiding the confounds of forced-choice conflicts. By evaluating humans and neural networks under controlled suppression conditions, we find that CNNs are not inherently texture-biased but predominantly rely on local shape features. Nonetheless, this reliance can be substantially mitigated through modern training strategies or architectures (ConvNeXt, ViTs). We further extend the analysis across computer vision, medical imaging, and remote sensing, revealing that reliance patterns differ systematically: computer vision models prioritize shape, medical imaging models emphasize color, and remote sensing models exhibit a stronger reliance towards texture. Code is available at https://github.com/tomburgert/feature-reliance.
Authors: Andrea Della Vecchia, Damir Filipovi\'c
Abstract: This paper investigates theoretical and methodological foundations for stochastic optimal control (SOC) in discrete time. We start formulating the control problem in a general dynamic programming framework, introducing the mathematical structure needed for a detailed convergence analysis. The associate value function is estimated through a sequence of approximations combining nonparametric regression methods and Monte Carlo subsampling. The regression step is performed within reproducing kernel Hilbert spaces (RKHSs), exploiting the classical KRR algorithm, while Monte Carlo sampling methods are introduced to estimate the continuation value. To assess the accuracy of our value function estimator, we propose a natural error decomposition and rigorously control the resulting error terms at each time step. We then analyze how this error propagates backward in time-from maturity to the initial stage-a relatively underexplored aspect of the SOC literature. Finally, we illustrate how our analysis naturally applies to a key financial application: the pricing of American options.
Authors: Behzad Shayegh, Jan-Thorsten Peter, David Vilar, Tobias Domhan, Juraj Juraska, Markus Freitag, Lili Mou
Abstract: We investigate the tradeoff between adequacy and fluency in machine translation. We show the severity of this tradeoff at the evaluation level and analyze where popular metrics fall within it. Essentially, current metrics generally lean toward adequacy, meaning that their scores correlate more strongly with the adequacy of translations than with fluency. More importantly, we find that this tradeoff also persists at the meta-evaluation level, and that the standard WMT meta-evaluation favors adequacy-oriented metrics over fluency-oriented ones. We show that this bias is partially attributed to the composition of the systems included in the meta-evaluation datasets. To control this bias, we propose a method that synthesizes translation systems in meta-evaluation. Our findings highlight the importance of understanding this tradeoff in meta-evaluation and its impact on metric rankings.
Authors: Kailash Gopal Darmasubramanian, Akash Pareek, Arindam Khan, Arpit Agarwal
Abstract: Optimizing the timing and frequency of ads is a central problem in digital advertising, with significant economic consequences. Existing scheduling policies rely on simple heuristics, such as uniform spacing and frequency caps, that overlook long-term user interest. However, it is well-known that users' long-term interest and engagement result from the interplay of several psychological effects (Curmei, Haupt, Recht, Hadfield-Menell, ACM CRS, 2022). In this work, we model change in user interest upon showing ads based on three key psychological principles: mere exposure, hedonic adaptation, and operant conditioning. The first two effects are modeled using a concave function of user interest with repeated exposure, while the third effect is modeled using a temporal decay function, which explains the decline in user interest due to overexposure. Under our psychological behavior model, we ask the following question: Given a continuous time interval $T$, how many ads should be shown, and at what times, to maximize the user interest towards the ads? Towards answering this question, we first show that, if the number of displayed ads is fixed, then the optimal ad-schedule only depends on the operant conditioning function. Our main result is a quasi-linear time algorithm that outputs a near-optimal ad-schedule, i.e., the difference in the performance of our schedule and the optimal schedule is exponentially small. Our algorithm leads to significant insights about optimal ad placement and shows that simple heuristics such as uniform spacing are sub-optimal under many natural settings. The optimal number of ads to display, which also depends on the mere exposure and hedonistic adaptation functions, can be found through a simple linear search given the above algorithm. We further support our findings with experimental results, demonstrating that our strategy outperforms various baselines.
Authors: Th\'eo Bodrito, Olivier Flasseur, Julien Mairal, Jean Ponce, Maud Langlois, Anne-Marie Lagrange
Abstract: Exoplanet imaging is a major challenge in astrophysics due to the need for high angular resolution and high contrast. We present a multi-scale statistical model for the nuisance component corrupting multivariate image series at high contrast. Integrated into a learnable architecture, it leverages the physics of the problem and enables the fusion of multiple observations of the same star in a way that is optimal in terms of detection signal-to-noise ratio. Applied to data from the VLT/SPHERE instrument, the method significantly improves the detection sensitivity and the accuracy of astrometric and photometric estimation.
Authors: T. O. Abiola, K. D. Abiodun, O. E. Olumide, O. O. Adebanji, O. Hiram Calvo, Grigori Sidorov
Abstract: Hope speech language that fosters encouragement and optimism plays a vital role in promoting positive discourse online. However, its detection remains challenging, especially in multilingual and low-resource settings. This paper presents a multilingual framework for hope speech detection using an active learning approach and transformer-based models, including mBERT and XLM-RoBERTa. Experiments were conducted on datasets in English, Spanish, German, and Urdu, including benchmark test sets from recent shared tasks. Our results show that transformer models significantly outperform traditional baselines, with XLM-RoBERTa achieving the highest overall accuracy. Furthermore, our active learning strategy maintained strong performance even with small annotated datasets. This study highlights the effectiveness of combining multilingual transformers with data-efficient training strategies for hope speech detection.
Authors: Shaofeng Yin, Yanjie Ze, Hong-Xing Yu, C. Karen Liu, Jiajun Wu
Abstract: Humanoid loco-manipulation in unstructured environments demands tight integration of egocentric perception and whole-body control. However, existing approaches either depend on external motion capture systems or fail to generalize across diverse tasks. We introduce VisualMimic, a visual sim-to-real framework that unifies egocentric vision with hierarchical whole-body control for humanoid robots. VisualMimic combines a task-agnostic low-level keypoint tracker -- trained from human motion data via a teacher-student scheme -- with a task-specific high-level policy that generates keypoint commands from visual and proprioceptive input. To ensure stable training, we inject noise into the low-level policy and clip high-level actions using human motion statistics. VisualMimic enables zero-shot transfer of visuomotor policies trained in simulation to real humanoid robots, accomplishing a wide range of loco-manipulation tasks such as box lifting, pushing, football dribbling, and kicking. Beyond controlled laboratory settings, our policies also generalize robustly to outdoor environments. Videos are available at: https://visualmimic.github.io .
Authors: Meshi Bashari, Yonghoon Lee, Roy Maor Lotan, Edgar Dobriban, Yaniv Romano
Abstract: The rapid proliferation of high-quality synthetic data -- generated by advanced AI models or collected as auxiliary data from related tasks -- presents both opportunities and challenges for statistical inference. This paper introduces a GEneral Synthetic-Powered Inference (GESPI) framework that wraps around any statistical inference procedure to safely enhance sample efficiency by combining synthetic and real data. Our framework leverages high-quality synthetic data to boost statistical power, yet adaptively defaults to the standard inference method using only real data when synthetic data is of low quality. The error of our method remains below a user-specified bound without any distributional assumptions on the synthetic data, and decreases as the quality of the synthetic data improves. This flexibility enables seamless integration with conformal prediction, risk control, hypothesis testing, and multiple testing procedures, all without modifying the base inference method. We demonstrate the benefits of our method on challenging tasks with limited labeled data, including AlphaFold protein structure prediction, and comparing large reasoning models on complex math problems.
Authors: Andi Nika, Sepehr Elahi, \c{C}a\u{g}{\i}n Ararat, Cem Tekin
Abstract: We consider the problem of optimizing a vector-valued objective function $\boldsymbol{f}$ sampled from a Gaussian Process (GP) whose index set is a well-behaved, compact metric space $({\cal X},d)$ of designs. We assume that $\boldsymbol{f}$ is not known beforehand and that evaluating $\boldsymbol{f}$ at design $x$ results in a noisy observation of $\boldsymbol{f}(x)$. Since identifying the Pareto optimal designs via exhaustive search is infeasible when the cardinality of ${\cal X}$ is large, we propose an algorithm, called Adaptive $\boldsymbol{\epsilon}$-PAL, that exploits the smoothness of the GP-sampled function and the structure of $({\cal X},d)$ to learn fast. In essence, Adaptive $\boldsymbol{\epsilon}$-PAL employs a tree-based adaptive discretization technique to identify an $\boldsymbol{\epsilon}$-accurate Pareto set of designs in as few evaluations as possible. We provide both information-type and metric dimension-type bounds on the sample complexity of $\boldsymbol{\epsilon}$-accurate Pareto set identification. We also experimentally show that our algorithm outperforms other Pareto set identification methods.
Authors: Ranga Shaarad Ayyagari, Revanth Raj Eega, Ambedkar Dukkipati
Abstract: Reinforcement Learning Algorithms are predominantly developed for stationary environments, and the limited literature that considers nonstationary environments often involves specific assumptions about changes that can occur in transition probability matrices and reward functions. Considering that real-world applications involve environments that continuously evolve due to various external events, and humans make decisions by discerning patterns in historical events, this study investigates Markov Decision Processes under the influence of an external temporal process. We establish the conditions under which the problem becomes tractable, allowing it to be addressed by considering only a finite history of events, based on the properties of the perturbations introduced by the exogenous process. We propose and theoretically analyze a policy iteration algorithm to tackle this problem, which learns policies contingent upon the current state of the environment, as well as a finite history of prior events of the exogenous process. We show that such an algorithm is not guaranteed to converge. However, we provide a guarantee for policy improvement in regions of the state space determined by the approximation error induced by considering tractable policies and value functions. We also establish the sample complexity of least-squares policy evaluation and policy improvement algorithms that consider approximations due to the incorporation of only a finite history of temporal events. While our results are applicable to general discrete-time processes satisfying certain conditions on the rate of decay of the influence of their events, we further analyze the case of discrete-time Hawkes processes with Gaussian marks. We performed experiments to demonstrate our findings for policy evaluation and deployment in traditional control environments.
Authors: Charlie Hou, Kiran Koshy Thekumparampil, Michael Shavlovsky, Giulia Fanti, Yesh Dattatreya, Sujay Sanghavi
Abstract: On tabular data, a significant body of literature has shown that current deep learning (DL) models perform at best similarly to Gradient Boosted Decision Trees (GBDTs), while significantly underperforming them on outlier data. However, these works often study idealized problem settings which may fail to capture complexities of real-world scenarios. We identify a natural tabular data setting where DL models can outperform GBDTs: tabular Learning-to-Rank (LTR) under label scarcity. Tabular LTR applications, including search and recommendation, often have an abundance of unlabeled data, and scarce labeled data. We show that DL rankers can utilize unsupervised pretraining to exploit this unlabeled data. In extensive experiments over both public and proprietary datasets, we show that pretrained DL rankers consistently outperform GBDT rankers on ranking metrics -- sometimes by as much as 38% -- both overall and on outliers.
Authors: Ilya Kuleshov, Evgenia Romanenkova, Vladislav Zhuzhel, Galina Boeva, Evgeni Vorsin, Alexey Zaytsev
Abstract: Neural CDEs provide a natural way to process the temporal evolution of irregular time series. The number of function evaluations (NFE) is these systems' natural analog of depth (the number of layers in traditional neural networks). It is usually regulated via solver error tolerance: lower tolerance means higher numerical precision, requiring more integration steps. However, lowering tolerances does not adequately increase the models' expressiveness. We propose a simple yet effective alternative: scaling the integration time horizon to increase NFEs and "deepen`` the model. Increasing the integration interval causes uncontrollable growth in conventional vector fields, so we also propose a way to stabilize the dynamics via Negative Feedback (NF). It ensures provable stability without constraining flexibility. It also implies robustness: we provide theoretical bounds for Neural ODE risk using Gaussian process theory. Experiments on four open datasets demonstrate that our method, DeNOTS, outperforms existing approaches~ -- ~including recent Neural RDEs and state space models,~ -- ~achieving up to $20\%$ improvement in metrics. DeNOTS combines expressiveness, stability, and robustness, enabling reliable modelling in continuous-time domains.
Authors: Yisong Fu, Fei Wang, Zezhi Shao, Boyu Diao, Lin Wu, Zhulin An, Chengqing Yu, Yujie Li, Yongjun Xu
Abstract: Transformers have gained attention in atmospheric time series forecasting (ATSF) for their ability to capture global spatial-temporal correlations. However, their complex architectures lead to excessive parameter counts and extended training times, limiting their scalability to large-scale forecasting. In this paper, we revisit ATSF from a theoretical perspective of atmospheric dynamics and uncover a key insight: spatial-temporal position embedding (STPE) can inherently model spatial-temporal correlations even without attention mechanisms. Its effectiveness arises from the integration of geographical coordinates and temporal features, which are intrinsically linked to atmospheric dynamics. Based on this, we propose STELLA, a Spatial-Temporal knowledge Embedded Lightweight modeL for ASTF, utilizing only STPE and an MLP architecture in place of Transformer layers. With 10k parameters and one hour of training, STELLA achieves superior performance on five datasets compared to other advanced methods. The paper emphasizes the effectiveness of spatial-temporal knowledge integration over complex architectures, providing novel insights for ATSF. The code is available at https://github.com/GestaltCogTeam/STELLA.
Authors: Vighnesh Birodkar, Gabriel Barcik, James Lyon, Sergey Ioffe, David Minnen, Joshua V. Dillon
Abstract: For learned image representations, basic autoencoders often produce blurry results. Reconstruction quality can be improved by incorporating additional penalties such as adversarial (GAN) and perceptual losses. Arguably, these approaches lack a principled interpretation. Concurrently, in generative settings diffusion has demonstrated a remarkable ability to create crisp, high quality results and has solid theoretical underpinnings (from variational inference to direct study as the Fisher Divergence). Our work combines autoencoder representation learning with diffusion and is, to our knowledge, the first to demonstrate jointly learning a continuous encoder and decoder under a diffusion-based loss and showing that it can lead to higher compression and better generation. We demonstrate that this approach yields better reconstruction quality as compared to GAN-based autoencoders while being easier to tune. We also show that the resulting representation is easier to model with a latent diffusion model as compared to the representation obtained from a state-of-the-art GAN-based loss. Since our decoder is stochastic, it can generate details not encoded in the otherwise deterministic latent representation; we therefore name our approach "Sample what you can't compress", or SWYCC for short.
Authors: Chengxi Ye, Grace Chu, Yanfeng Liu, Yichi Zhang, Lukasz Lew, Li Zhang, Mark Sandler, Andrew Howard
Abstract: The discontinuous operations inherent in quantization and sparsification introduce a long-standing obstacle to backpropagation, particularly in ultra-low precision and sparse regimes. The standard Straight-Through Estimator (STE) is widely used to address this, but the well-understood mismatch between its quantization-aware forward pass and quantization-oblivious backward pass leads to unmanaged error that can corrupt the learning process. We solve this by introducing a denoising dequantization transform derived from a principled ridge regression objective. This transform makes the entire learning process aware of and robust to the quantization error that STE's surrogate gradient bypasses, by creating an explicit, corrective gradient path. We extend this principle to sparsification by viewing it as a special form of quantization that maps insignificant values to zero. Our unified framework allows existing models to be trained at a wide spectrum of precisions and sparsity levels with off-the-shelf recipes, achieving stable training of fully binary (A1W1) and sparse sub-1-bit networks where other methods falter. This approach yields state-of-the-art results and provides a theoretically-grounded path to hyper-efficient neural networks.
Authors: Youzuo Lin, Shihang Feng, James Theiler, Yinpeng Chen, Umberto Villa, Jing Rao, John Greenhall, Cristian Pantea, Mark A. Anastasio, Brendt Wohlberg
Abstract: Computational wave imaging (CWI) extracts hidden structure and physical properties of a volume of material by analyzing wave signals that traverse that volume. Applications include seismic exploration of the Earth's subsurface, acoustic imaging and non-destructive testing in material science, and ultrasound computed tomography in medicine. Current approaches for solving CWI problems can be divided into two categories: those rooted in traditional physics, and those based on deep learning. Physics-based methods stand out for their ability to provide high-resolution and quantitatively accurate estimates of acoustic properties within the medium. However, they can be computationally intensive and are susceptible to ill-posedness and nonconvexity typical of CWI problems. Machine learning-based computational methods have recently emerged, offering a different perspective to address these challenges. Diverse scientific communities have independently pursued the integration of deep learning in CWI. This review discusses how contemporary scientific machine-learning (ML) techniques, and deep neural networks in particular, have been developed to enhance and integrate with traditional physics-based methods for solving CWI problems. We present a structured framework that consolidates existing research spanning multiple domains, including computational imaging, wave physics, and data science. This study concludes with important lessons learned from existing ML-based methods and identifies technical hurdles and emerging trends through a systematic analysis of the extensive literature on this topic.
Authors: Sunwoong Yang, Ricardo Vinuesa, Namwoo Kang
Abstract: This study addresses the critical challenge of error accumulation in spatio-temporal auto-regressive (AR) predictions within scientific machine learning models by exploring temporal integration schemes and adaptive multi-step rollout strategies. We introduce the first implementation of the two-step Adams-Bashforth method specifically tailored for data-driven AR prediction, leveraging historical derivative information to enhance numerical stability without additional computational overhead. To validate our approach, we systematically evaluate time integration schemes across canonical 2D PDEs before extending to complex Navier-Stokes cylinder vortex shedding dynamics. Additionally, we develop three novel adaptive weighting strategies that dynamically adjust the importance of different future time steps during multi-step rollout training. Our analysis reveals that as physical complexity increases, such sophisticated rollout techniques become essential, with the Adams-Bashforth scheme demonstrating consistent robustness across investigated systems and our best adaptive approach delivering an 89% improvement over conventional fixed-weight methods while maintaining similar computational costs. For the complex Navier-Stokes vortex shedding problem, despite using an extremely lightweight graph neural network with just 1,177 trainable parameters and training on only 50 snapshots, our framework accurately predicts 350 future time steps reducing mean squared error from 0.125 (single-step direct prediction) to 0.002 (Adams-Bashforth with proposed multi-step rollout). Our integrated methodology demonstrates an 83% improvement over standard noise injection techniques and maintains robustness under severe spatial constraints; specifically, when trained on only a partial spatial domain, it still achieves 58% and 27% improvements over direct prediction and forward Euler methods, respectively.
Authors: Connor Lawless, Yingxi Li, Anders Wikum, Madeleine Udell, Ellen Vitercik
Abstract: Mixed integer linear programming (MILP) solvers expose hundreds of parameters that have an outsized impact on performance but are difficult to configure for all but expert users. Existing machine learning (ML) approaches require training on thousands of related instances, generalize poorly and can be difficult to integrate into existing solver workflows. We propose a large language model (LLM)-based framework that configures cutting plane separators using problem descriptions and solver-specific separator summaries. To reduce variance in LLM outputs, we introduce an ensembling strategy that clusters and aggregates candidate configurations into a small portfolio of high-performing configurations. Our method requires no custom solver interface, generates configurations in seconds via simple API calls, and requires solving only a small number of instances. Extensive experiments on standard synthetic and real-world MILPs show our approach matches or outperforms state-of-the-art configuration methods with a fraction of the data and computation.
Authors: Zhengpeng Xie, Jiahang Cao, Changwei Wang, Fan Yang, Marco Hutter, Qiang Zhang, Jianxiong Zhang, Renjing Xu
Abstract: In this paper, we argue that mutual distillation between reinforcement learning policies serves as an implicit regularization, preventing them from overfitting to irrelevant features. We highlight two separate contributions: (i) Theoretically, for the first time, we prove that enhancing the policy robustness to irrelevant features leads to improved generalization performance. (ii) Empirically, we demonstrate that mutual distillation between policies contributes to such robustness, enabling the spontaneous emergence of invariant representations over pixel inputs. Ultimately, we do not claim to achieve state-of-the-art performance but rather focus on uncovering the underlying principles of generalization and deepening our understanding of its mechanisms.
Authors: Keisuke Kamo, Hideaki Iiduka
Abstract: Stochastic gradient descent with momentum (SGDM), in which a momentum term is added to SGD, has been well studied in both theory and practice. The theoretical studies show that the settings of the learning rate and momentum weight affect the convergence of SGDM. Meanwhile, the practical studies have shown that the batch-size setting strongly affects the performance of SGDM. In this paper, we focus on mini-batch SGDM with a constant learning rate and constant momentum weight, which is frequently used to train deep neural networks. We show theoretically that using a constant batch size does not always minimize the expectation of the full gradient norm of the empirical loss in training a deep neural network, whereas using an increasing batch size definitely minimizes it; that is, an increasing batch size improves the convergence of mini-batch SGDM. We also provide numerical results supporting our analyses, indicating specifically that mini-batch SGDM with an increasing batch size converges to stationary points faster than with a constant batch size, while also reducing computational cost. Python implementations of the optimizers used in the numerical experiments are available at https://github.com/iiduka-researches/NSHB_increasing_batchsize_acml25/.
URLs: https://github.com/iiduka-researches/NSHB_increasing_batchsize_acml25/.
Authors: Tom Heskes
Abstract: Bias-variance decompositions are widely used to understand the generalization performance of machine learning models. While the squared error loss permits a straightforward decomposition, other loss functions - such as zero-one loss or $L_1$ loss - either fail to sum bias and variance to the expected loss or rely on definitions that lack the essential properties of meaningful bias and variance. Recent research has shown that clean decompositions can be achieved for the broader class of Bregman divergences, with the cross-entropy loss as a special case. However, the necessary and sufficient conditions for these decompositions remain an open question. In this paper, we address this question by studying continuous, nonnegative loss functions that satisfy the identity of indiscernibles (zero loss if and only if the two arguments are identical), under mild regularity conditions. We prove that so-called $g$-Bregman divergences are the only such loss functions that have a clean bias-variance decomposition. A $g$-Bregman divergence can be transformed into a standard Bregman divergence through an invertible change of variables. This makes the squared Mahalanobis distance, up to such a variable transformation, the only symmetric loss function with a clean bias-variance decomposition. Consequently, common metrics such as $0$-$1$ and $L_1$ losses cannot admit a clean bias-variance decomposition, explaining why previous attempts have failed. We also examine the impact of relaxing the restrictions on the loss functions and how this affects our results.
Authors: Javier Fumanal-Idocin, Raquel Fernandez-Peralta, Javier Andreu-Perez
Abstract: Rule-based models are essential for high-stakes decision-making due to their transparency and interpretability, but their discrete nature creates challenges for optimization and scalability. In this work, we present the Fuzzy Rule-based Reasoner (FRR), a novel gradient-based rule learning system that supports strict user constraints over rule-based complexity while achieving competitive performance. To maximize interpretability, the FRR uses semantically meaningful fuzzy logic partitions, unattainable with existing neuro-fuzzy approaches, and sufficient (single-rule) decision-making, which avoids the combinatorial complexity of additive rule ensembles. Through extensive evaluation across 40 datasets, FRR demonstrates: (1) superior performance to traditional rule-based methods (e.g., $5\%$ average accuracy over RIPPER); (2) comparable accuracy to tree-based models (e.g., CART) using rule bases $90\%$ more compact; and (3) achieves $96\%$ of the accuracy of state-of-the-art additive rule-based models while using only sufficient rules and requiring only $3\%$ of their rule base size.
Authors: Andrea Tonini, Luca Dede'
Abstract: Among other uses, neural networks are a powerful tool for solving deterministic and Bayesian inverse problems in real-time, where variational autoencoders, a specialized type of neural network, enable the Bayesian estimation of model parameters and their distribution from observational data allowing real-time inverse uncertainty quantification. In this work, we build upon existing research [Goh, H. et al., Proceedings of Machine Learning Research, 2022] by proposing a novel loss function to train variational autoencoders for Bayesian inverse problems. When the forward map is affine, we provide a theoretical proof of the convergence of the latent states of variational autoencoders to the posterior distribution of the model parameters. We validate this theoretical result through numerical tests and we compare the proposed variational autoencoder with the existing one in the literature both in terms of accuracy and generalization properties. Finally, we test the proposed variational autoencoder on a Laplace equation, with comparison to the original one and Markov Chains Monte Carlo.
Authors: Michael Somma, Thomas Gallien, Branka Stojanovic
Abstract: Anomaly detection in complex dynamical systems is essential for ensuring reliability, safety, and efficiency in industrial and cyber-physical infrastructures. Predictive maintenance helps prevent costly failures, while cybersecurity monitoring has become critical as digitized systems face growing threats. Many of these systems exhibit oscillatory behaviors and bounded motion, requiring anomaly detection methods that capture structured temporal dependencies while adhering to physical consistency principles. In this work, we propose a system-theoretic approach to anomaly detection, grounded in classical embedding theory and physics-inspired consistency principles. We build upon the Fractal Whitney Embedding Prevalence Theorem that extends traditional embedding techniques to complex system dynamics. Additionally, we introduce state-derivative pairs as an embedding strategy to capture system evolution. To enforce temporal coherence, we develop a Temporal Differential Consistency Autoencoder (TDC-AE), incorporating a TDC-Loss that aligns the approximated derivatives of latent variables with their dynamic representations. We evaluate our method on two subsets (FD001, FD003) of the C-MAPSS dataset, a benchmark for turbofan engine degradation. TDC-AE machtes LSTMs and outperforms Transformers while achieving a nearly 100x reduction in MAC operations, making it particularly suited for lightweight edge computing. Our findings support the hypothesis that anomalies disrupt stable system dynamics, providing a robust signal for anomaly detection.
Authors: Teng Pang, Bingzheng Wang, Guoqiang Wu, Yilong Yin
Abstract: Offline preference-based reinforcement learning (PbRL) mitigates the need for reward definition, aligning with human preferences via preference-driven reward feedback without interacting with the environment. However, trajectory-wise preference labels are difficult to meet the precise learning of step-wise reward, thereby affecting the performance of downstream algorithms. To alleviate the insufficient step-wise reward caused by trajectory-wise preferences, we propose a novel preference-based reward acquisition method: Diffusion Preference-based Reward (DPR). DPR directly treats step-wise preference-based reward acquisition as a binary classification and utilizes the robustness of diffusion classifiers to infer step-wise rewards discriminatively. In addition, to further utilize trajectory-wise preference information, we propose Conditional Diffusion Preference-based Reward (C-DPR), which conditions on trajectory-wise preference labels to enhance reward inference. We apply the above methods to existing offline RL algorithms, and a series of experimental results demonstrate that the diffusion classifier-driven reward outperforms the previous reward acquisition method with the Bradley-Terry model.
Authors: Derin Ozer, Sylvain Lamprier, Thomas Cauchy, Nicolas Gutowski, Benoit Da Mota
Abstract: The accurate prediction of chemical reaction outcomes is a major challenge in computational chemistry. Current models rely heavily on either highly specific reaction templates or template-free methods, both of which present limitations. To address these, this work proposes the Broad Reaction Set (BRS), a set featuring 20 generic reaction templates written in SMARTS, a pattern-based notation designed to describe substructures and reactivity. Additionally, we introduce ProPreT5, a T5-based model specifically adapted for chemistry and, to the best of our knowledge, the first language model capable of directly handling and applying SMARTS reaction templates. To further improve generalization, we propose the first augmentation strategy for SMARTS, which injects structural diversity at the pattern level. Trained on augmented templates, ProPreT5 demonstrates strong predictive performance and generalization to unseen reactions. Together, these contributions provide a novel and practical alternative to current methods, advancing the field of template-based reaction prediction.
Authors: Mina Kamao, Hayato Ono, Ayumu Yamashita, Kaoru Amano, Masataka Sawayama
Abstract: Alignment between human brain networks and artificial models has become an active research area in vision science and machine learning. A widely adopted approach is identifying "metamers," stimuli physically different yet perceptually equivalent within a system. However, conventional methods lack a direct approach to searching for the human metameric space. Instead, researchers first develop biologically inspired models and then infer about human metamers indirectly by testing whether model metamers also appear as metamers to humans. Here, we propose the Multidimensional Adaptive Metamer Exploration (MAME) framework, enabling direct, high-dimensional exploration of human metameric spaces through online image generation guided by human perceptual feedback. MAME modulates reference images across multiple dimensions based on hierarchical neural network responses, adaptively updating generation parameters according to participants' perceptual discriminability. Using MAME, we successfully measured multidimensional human metameric spaces within a single psychophysical experiment. Experimental results using a biologically plausible CNN model showed that human discrimination sensitivity was lower for metameric images based on low-level features compared to high-level features, which image contrast metrics could not explain. The finding suggests a relatively worse alignment between the metameric spaces of humans and the CNN model for low-level processing compared to high-level processing. Counterintuitively, given recent discussions on alignment at higher representational levels, our results highlight the importance of early visual computations in shaping biologically plausible models. Our MAME framework can serve as a future scientific tool for directly investigating the functional organization of human vision.
Authors: Xintong Wang, Haihan Nan, Ruidong Li, Huaming Wu
Abstract: Accurately predicting spatio-temporal network traffic is essential for dynamically managing computing resources in modern communication systems and minimizing energy consumption. Although spatio-temporal traffic prediction has received extensive research attention, further improvements in prediction accuracy and computational efficiency remain necessary. In particular, existing decomposition-based methods or hybrid architectures often incur heavy overhead when capturing local and global feature correlations, necessitating novel approaches that optimize accuracy and complexity. In this paper, we propose an efficient spatio-temporal network traffic prediction framework, DP-LET, which consists of a data processing module, a local feature enhancement module, and a Transformer-based prediction module. The data processing module is designed for high-efficiency denoising of network data and spatial decoupling. In contrast, the local feature enhancement module leverages multiple Temporal Convolutional Networks (TCNs) to capture fine-grained local features. Meanwhile, the prediction module utilizes a Transformer encoder to model long-term dependencies and assess feature relevance. A case study on real-world cellular traffic prediction demonstrates the practicality of DP-LET, which maintains low computational complexity while achieving state-of-the-art performance, significantly reducing MSE by 31.8% and MAE by 23.1% compared to baseline models.
Authors: Arash Torabi Goodarzi, Roman Kochnev, Waleed Khalid, Hojjat Torabi Goudarzi, Furui Qin, Tolgay Atinc Uzun, Yashkumar Sanjaybhai Dhameliya, Yash Kanubhai Kathiriya, Zofia Antonina Bentyn, Dmitry Ignatov, Radu Timofte
Abstract: Neural networks are the backbone of modern artificial intelligence, but designing, evaluating, and comparing them remains labor-intensive. While numerous datasets exist for training, there are few standardized collections of the models themselves. We introduce LEMUR, an open-source dataset and framework that provides a large collection of PyTorch-based neural networks across tasks such as classification, segmentation, detection, and natural language processing. Each model follows a unified template, with configurations and results stored in a structured database to ensure consistency and reproducibility. LEMUR integrates automated hyperparameter optimization via Optuna, includes statistical analysis and visualization tools, and offers an API for seamless access to performance data. The framework is extensible, allowing researchers to add new models, datasets, or metrics without breaking compatibility. By standardizing implementations and unifying evaluation, LEMUR aims to accelerate AutoML research, enable fair benchmarking, and reduce barriers to large-scale neural network experimentation. To support adoption and collaboration, LEMUR and its plugins are released under the MIT license at: https://github.com/ABrain-One/nn-dataset https://github.com/ABrain-One/nn-plots https://github.com/ABrain-One/nn-vr
URLs: https://github.com/ABrain-One/nn-dataset, https://github.com/ABrain-One/nn-plots, https://github.com/ABrain-One/nn-vr
Authors: Hu Wang, Congbo Ma, Ian Reid, Mohammad Yaqub
Abstract: The advantage function is a central concept in RL that helps reduce variance in policy gradient estimates. Recently, for language modeling, Group Relative Policy Optimization (GRPO) was proposed to compute the advantage for each output by subtracting the mean reward, as the baseline, for all outputs in the group. However, it can lead to high variance when the reward advantage is inaccurately predicted. In this work, we propose Kalman Filter Enhanced Group Relative Policy Optimization (KRPO) model, by using lightweight Kalman filtering to dynamically estimate the latent reward baseline and uncertainty. This filtering technique replaces the naive group mean, enabling more adaptive advantage normalization. Our method does not require additional learned parameters over GRPO. This approach offers a simple yet effective way to incorporate multiple outputs of GRPO into advantage estimation, improving policy optimization in settings where highly dynamic reward signals are difficult to model for language models. Through the accuracies and rewards obtained from math question answering and reasoning, we show that using a more adaptive advantage estimation model, KRPO can improve the stability and performance of GRPO. The code is available at https://github.com/billhhh/KRPO_LLMs_RL.
Authors: Peter Chen, Xiaopeng Li, Ziniu Li, Xi Chen, Tianyi Lin
Abstract: Reinforcement learning (RL) has proven effective in strengthening the reasoning capabilities of large language models (LLMs). A widely adopted method, Group Relative Policy Optimization (GRPO), has shown strong empirical results in training DeepSeek-R1. However, GRPO fails to update the policy when all responses within a group are incorrect (i.e., \emph{all-negative-sample} groups). This limitation underscores a key gap between artificial and human intelligence: unlike humans, who can learn from mistakes, GRPO discards these signals. Our first contribution is to introduce a simple framework that mitigates the all-negative-sample issue by incorporating response diversity within groups using a \textit{step-wise} judge model, which can be either directly trained or adapted from existing LLMs. We prove that this diversification can accelerate GRPO's learning dynamics in a simplified setting. We also empirically validate the proposed stepwise guided policy optimization (SGPO) method, demonstrating consistent gains across model sizes (7B, 14B, 32B) in offline and online training on 9 benchmarks, including base and distilled variants. Our results highlight two advantages: (i) SGPO surpasses GRPO, especially in the early and mid-training stages where all-negative-sample groups are prevalent; and (ii) SGPO does not require judge models to generate correct answers, differentiating it from knowledge distillation methods.
Authors: Fynn Fromme, Hans Harder, Christine Allen-Blanchette, Sebastian Peitz
Abstract: The use of machine learning for modeling, understanding, and controlling large-scale physics systems is quickly gaining in popularity, with examples ranging from electromagnetism over nuclear fusion reactors and magneto-hydrodynamics to fluid mechanics and climate modeling. These systems - governed by partial differential equations - present unique challenges regarding the large number of degrees of freedom and the complex dynamics over many scales both in space and time, and additional measures to improve accuracy and sample efficiency are highly desirable. We present an end-to-end equivariant surrogate model consisting of an equivariant convolutional autoencoder and an equivariant convolutional LSTM using $G$-steerable kernels. As a case study, we consider the three-dimensional Rayleigh-B\'enard convection, which describes the buoyancy-driven fluid flow between a heated bottom and a cooled top plate. While the system is E(2)-equivariant in the horizontal plane, the boundary conditions break the translational equivariance in the vertical direction. Our architecture leverages vertically stacked layers of $D_4$-steerable kernels, with additional partial kernel sharing in the vertical direction for further efficiency improvement. We demonstrate significant gains in sample and parameter efficiency, as well as a better scaling to more complex dynamics. The accompanying code is available under https://github.com/FynnFromme/equivariant-rb-forecasting.
URLs: https://github.com/FynnFromme/equivariant-rb-forecasting.
Authors: Jian Xiong, Jingbo Zhou, Jingyong Ye, Qiang Huang, Dejing Dou
Abstract: Reinforcement learning (RL) has emerged as an effective approach for enhancing the reasoning capabilities of large language models (LLMs), especially in scenarios where supervised fine-tuning (SFT) falls short due to limited chain-of-thought (CoT) data. Among RL-based post-training methods, group relative advantage estimation, as exemplified by Group Relative Policy Optimization (GRPO), has attracted considerable attention for eliminating the dependency on the value model, thereby simplifying training compared to traditional approaches like Proximal Policy Optimization (PPO). However, we observe that exsiting group relative advantage estimation method still suffers from training inefficiencies, particularly when the estimated advantage approaches zero. To address this limitation, we propose Advantage-Augmented Policy Optimization (AAPO), a novel RL algorithm that optimizes the cross-entropy (CE) loss using advantages enhanced through a momentum-based estimation scheme. This approach effectively mitigates the inefficiencies associated with group relative advantage estimation. Experimental results on multiple mathematical reasoning benchmarks demonstrate the superior performance of AAPO.
Authors: Tong Cheng, Jie Fu, Xinpeng Ling, Huifa Li, Zhili Chen, Haifeng Qian, Junqing Gong
Abstract: Graph Neural Networks (GNNs) have been widely used for graph analysis. Federated Graph Learning (FGL) is an emerging learning framework to collaboratively train graph data from various clients. Although FGL allows client data to remain localized, a malicious server can still steal client private data information through uploaded gradient. In this paper, we for the first time propose label distribution attacks (LDAs) on FGL that aim to infer the label distributions of the client-side data. Firstly, we observe that the effectiveness of LDA is closely related to the variance of node embeddings in GNNs. Next, we analyze the relation between them and propose a new attack named EC-LDA, which significantly improves the attack effectiveness by compressing node embeddings. Then, extensive experiments on node classification and link prediction tasks across six widely used graph datasets show that EC-LDA outperforms the SOTA LDAs. Specifically, EC-LDA can achieve the Cos-sim as high as 1.0 under almost all cases. Finally, we explore the robustness of EC-LDA under differential privacy protection and discuss the potential effective defense methods to EC-LDA. Our code is available at https://github.com/cheng-t/EC-LDA.
Authors: Babak Barazandeh, Subhabrata Majumdar, Om Rajyaguru, George Michailidis
Abstract: Parameter-efficient fine-tuning (PEFT) methods, such as LoRA, offer compact and effective alternatives to full model fine-tuning by introducing low-rank updates to pre-trained weights. However, most existing approaches rely on global low rank structures, which can overlook spatial patterns spread across the parameter space. In this work, we propose Localized LoRA, a generalized framework that models weight updates as a composition of low-rank matrices applied to structured blocks of the weight matrix. This formulation enables dense, localized updates throughout the parameter space without increasing the total number of trainable parameters. We provide a formal comparison between global, diagonal-local, and fully localized low-rank approximations, and show that our method consistently achieves lower approximation error under matched parameter budgets. Experiments on both synthetic and practical settings demonstrate that Localized LoRA offers a more expressive and adaptable alternative to existing methods, enabling efficient fine-tuning with improved performance.
Authors: Daogao Liu, Edith Cohen, Badih Ghazi, Peter Kairouz, Pritish Kamath, Alexander Knop, Ravi Kumar, Pasin Manurangsi, Adam Sealfon, Da Yu, Chiyuan Zhang
Abstract: We introduce $Urania$, a novel framework for generating insights about LLM chatbot interactions with rigorous differential privacy (DP) guarantees. The framework employs a private clustering mechanism and innovative keyword extraction methods, including frequency-based, TF-IDF-based, and LLM-guided approaches. By leveraging DP tools such as clustering, partition selection, and histogram-based summarization, $Urania$ provides end-to-end privacy protection. Our evaluation assesses lexical and semantic content preservation, pair similarity, and LLM-based metrics, benchmarking against a non-private Clio-inspired pipeline (Tamkin et al., 2024). Moreover, we develop a simple empirical privacy evaluation that demonstrates the enhanced robustness of our DP pipeline. The results show the framework's ability to extract meaningful conversational insights while maintaining stringent user privacy, effectively balancing data utility with privacy preservation.
Authors: Mingyu Lu, Ethan Weinberger, Chanwoo Kim, Su-In Lee
Abstract: High-content screening (HCS) assays based on high-throughput microscopy techniques such as Cell Painting have enabled the interrogation of cells' morphological responses to perturbations at an unprecedented scale. The collection of such data promises to facilitate a better understanding of the relationships between different perturbations and their effects on cellular state. Towards achieving this goal, recent advances in cross-modal contrastive learning could, in theory, be leveraged to learn a unified latent space that aligns perturbations with their corresponding morphological effects. However, the application of such methods to HCS data is not straightforward due to substantial differences in the semantics of Cell Painting images compared to natural images, and the difficulty of representing different classes of perturbations (e.g., small molecule vs CRISPR gene knockout) in a single latent space. In response to these challenges, here we introduce CellCLIP, a cross-modal contrastive learning framework for HCS data. CellCLIP leverages pre-trained image encoders coupled with a novel channel encoding scheme to better capture relationships between different microscopy channels in image embeddings, along with natural language encoders for representing perturbations. Our framework outperforms current open-source models, demonstrating the best performance in both cross-modal retrieval and biologically meaningful downstream tasks while also achieving significant reductions in computation time.
Authors: Michael Sullivan, Mareike Hartmann, Alexander Koller
Abstract: Although the power of LLM tool-use agents has ignited a flurry of recent research in this area, the curation of tool-use training data remains an open problem$-$especially for online RL training. Existing approaches to synthetic tool-use data generation tend to be non-interactive, and/or non-compositional. We introduce RandomWorld, a pipeline for the procedural generation of interactive tools and compositional tool-use data. We show that models tuned via SFT and RL on synthetic RandomWorld data improve on a range of tool-use benchmarks, and set the new SoTA for two metrics on the NESTFUL dataset. Further experiments show that downstream performance scales with the amount of RandomWorld-generated training data, opening up the possibility of further improvement through the use of entirely synthetic data.
Authors: Ting-Yun Chang, Muru Zhang, Jesse Thomason, Robin Jia
Abstract: Low-bit weight-only quantization significantly reduces the memory footprint of large language models (LLMs), but disproportionately affects certain examples. We analyze diverse 3-4 bit methods on LLMs ranging from 7B-70B in size and find that the quantization errors of 50 pairs of methods are strongly correlated (avg. 0.82) on FineWeb examples. Moreover, the residual stream magnitudes of full-precision models are indicative of future quantization errors. We further establish a hypothesis that relates the residual stream magnitudes to error amplification and accumulation over layers. Using LLM localization techniques, early exiting, and activation patching, we show that examples with large errors rely on precise residual activations in the late layers, and that the outputs of MLP gates play a crucial role in maintaining the perplexity. Our work reveals why certain examples result in large quantization errors and which model components are most critical for performance preservation.
Authors: Ganyu Wang, Jinjie Fang, Maxwell J. Yin, Bin Gu, Xi Chen, Boyu Wang, Yi Chang, Charles Ling
Abstract: Black-Box Discrete Prompt Learning is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting federated learning to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called \textit{FedOne}, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.
Authors: Wenxin Li, Chuan Wang, Hongdong Zhu, Qi Gao, Yin Ma, Hai Wei, Kai Wen
Abstract: Here in this work, we present a novel Quadratic Binary Optimization (QBO) model for quantized neural network training, enabling the use of arbitrary activation and loss functions through spline interpolation. We introduce Forward Interval Propagation (FIP), a method designed to tackle the challenges of non-linearity and the multi-layer composite structure in neural networks by discretizing activation functions into linear subintervals. This approach preserves the universal approximation properties of neural networks while allowing complex nonlinear functions to be optimized using quantum computers, thus broadening their applicability in artificial intelligence. We provide theoretical upper bounds on the approximation error and the number of Ising spins required, by deriving the sample complexity of the empirical risk minimization problem, from an optimization perspective. A significant challenge in solving the associated Quadratic Constrained Binary Optimization (QCBO) model on a large scale is the presence of numerous constraints. When employing the penalty method to handle these constraints, tuning a large number of penalty coefficients becomes a critical hyperparameter optimization problem, increasing computational complexity and potentially affecting solution quality. To address this, we employ the Quantum Conditional Gradient Descent (QCGD) algorithm, which leverages quantum computing to directly solve the QCBO problem. We prove the convergence of QCGD under a quantum oracle with randomness and bounded variance in objective value, as well as under limited precision constraints in the coefficient matrix. Additionally, we provide an upper bound on the Time-To-Solution for the QCBO solving process. We further propose a training algorithm with single-sample bit-scale optimization.
Authors: Ruokai Yin, Yuhang Li, Donghyun Lee, Priyadarshini Panda
Abstract: Large language models (LLMs) deliver strong performance but are difficult to deploy due to high memory and compute costs. While pruning reduces these demands, most methods ignore activation sparsity observed at runtime. We reinterpret activation sparsity as dynamic structured weight sparsity and propose DuoGPT, a unified framework that constructs dual-sparse (spMspV) workloads by combining unstructured weight pruning with activation sparsity. To preserve accuracy, we extend the Optimal Brain Compression (OBC) framework with activation-aware calibration and introduce output residuals from the dense model as correction terms. We further optimize the solution for efficient GPU execution, enabling scalability to billion-parameter LLMs. Evaluations on LLaMA-2 and LLaMA-3 show that DuoGPT outperforms state-of-the-art structured pruning methods by up to 9.17% accuracy at an iso-speedup of 1.39$\times$ compared to the baseline dense model. Code is available at Github.
Authors: Itay Niv, Neta Rabin
Abstract: Deep learning on graphs has shown remarkable success across numerous applications, including social networks, bio-physics, traffic networks, and recommendation systems. Regardless of their successes, current methods frequently depend on the assumption that training and testing data share the same distribution, a condition rarely met in real-world scenarios. While graph-transformer (GT) backbones have recently outperformed traditional message-passing neural networks (MPNNs) in multiple in-distribution (ID) benchmarks, their effectiveness under distribution shifts remains largely unexplored. In this work, we address the challenge of out-of-distribution (OOD) generalization for graph neural networks, with a special focus on the impact of backbone architecture. We systematically evaluate GT and hybrid backbones in OOD settings and compare them to MPNNs. To do so, we adapt several leading domain generalization (DG) algorithms to work with GTs and assess their performance on a benchmark designed to test a variety of distribution shifts. Our results reveal that GT and hybrid GT-MPNN backbones demonstrate stronger generalization ability compared to MPNNs, even without specialized DG algorithms (on four out of six benchmarks). Additionally, we propose a novel post-training analysis approach that compares the clustering structure of the entire ID and OOD test datasets, specifically examining domain alignment and class separation. Highlighting its model-agnostic design, the method yielded valuable insights into both GT and MPNN backbones and appears well suited for broader DG applications beyond graph learning, offering a deeper perspective on generalization abilities that goes beyond standard accuracy metrics. Together, our findings highlight the promise of graph-transformers for robust, real-world graph learning and set a new direction for future research in OOD generalization.
Authors: Filip Rydin, Attila Lischka, Jiaming Wu, Morteza Haghir Chehreghani, Bal\'azs Kulcs\'ar
Abstract: Learning-based methods for routing have gained significant attention in recent years, both in single-objective and multi-objective contexts. Yet, existing methods are unsuitable for routing on multigraphs, which feature multiple edges with distinct attributes between node pairs, despite their strong relevance in real-world scenarios. In this paper, we propose two graph neural network-based methods to address multi-objective routing on multigraphs. Our first approach operates directly on the multigraph by autoregressively selecting edges until a tour is completed. The second model, which is more scalable, first simplifies the multigraph via a learned pruning strategy and then performs autoregressive routing on the resulting simple graph. We evaluate both models empirically, across a wide range of problems and graph distributions, and demonstrate their competitive performance compared to strong heuristics and neural baselines.
Authors: Dibyajyoti Chakraborty, Haiwen Guan, Jason Stock, Troy Arcomano, Guido Cervone, Romit Maulik
Abstract: Score-based diffusion modeling is a generative machine learning algorithm that can be used to sample from complex distributions. They achieve this by learning a score function, i.e., the gradient of the log-probability density of the data, and reversing a noising process using the same. Once trained, score-based diffusion models not only generate new samples but also enable zero-shot conditioning of the generated samples on observed data. This promises a novel paradigm for data and model fusion, wherein the implicitly learned distributions of pretrained score-based diffusion models can be updated given the availability of online data in a Bayesian formulation. In this article, we apply such a concept to the super-resolution of a high-dimensional dynamical system, given the real-time availability of low-resolution and experimentally observed sparse sensor measurements from multimodal data. Additional analysis on how score-based sampling can be used for uncertainty estimates is also provided. Our experiments are performed for a super-resolution task that generates the ERA5 atmospheric dataset given sparse observations from a coarse-grained representation of the same and/or from unstructured experimental observations of the IGRA radiosonde dataset. We demonstrate accurate recovery of the high dimensional state given multiple sources of low-fidelity measurements. We also discover that the generative model can balance the influence of multiple dataset modalities during spatiotemporal reconstructions.
Authors: Yimeng Min, Carla P. Gomes
Abstract: We propose a non-autoregressive framework for the Travelling Salesman Problem where solutions emerge directly from learned permutations, without requiring explicit search. By applying a similarity transformation to Hamiltonian cycles, the model learns to approximate permutation matrices via continuous relaxations. Our unsupervised approach achieves competitive performance against classical heuristics, demonstrating that the inherent structure of the problem can effectively guide combinatorial optimization without sequential decision-making. Our method offers concrete evidence that neural networks can directly capture and exploit combinatorial structure.
Authors: Xujia Wang, Yunjia Qi, Bin Xu
Abstract: Parameter-Efficient Fine-Tuning (PEFT) methods, such as LoRA, significantly reduce the number of trainable parameters by introducing low-rank decomposition matrices. However, existing methods perform extensive matrix multiplications in domain specialization tasks, resulting in computational inefficiency and sub-optimal fine-tuning performance. Hence, we propose LoSiA(Low-Resources Subnet Integration Adaptation), an innovative method that dynamically localizes and optimizes critical parameters during the training process. Specifically, it identifies a sub-network using gradient sparsity analysis and optimizes it as the trainable target. This design enables effective high-rank adaptation by updating only the sub-network parameters, reducing the additional matrix multiplication. We also present LoSiA-Pro, a faster implementation of LoSiA, which reduces the training latency by about $27\%$ compared to LoRA. Extensive evaluations show that our method achieves minimal performance drop compared to full fine-tuning, while requiring the least training time across domain specialization and common-sense reasoning tasks. Further analysis shows that LoSiA also reduces forgetting during continued training. The source code is available at https://github.com/KlozeWang/LoSiA.
Authors: Yifan Deng, Spencer S. Ericksen, Anthony Gitter
Abstract: Scientific databases aggregate vast amounts of quantitative data alongside descriptive text. In biochemistry, molecule screening assays evaluate candidate molecules' functional responses against disease targets. Unstructured text that describes the biological mechanisms through which these targets operate, experimental screening protocols, and other attributes of assays offer rich information for drug discovery campaigns but has been untapped because of that unstructured format. We present Assay2Mol, a large language model-based workflow that can capitalize on the vast existing biochemical screening assays for early-stage drug discovery. Assay2Mol retrieves existing assay records involving targets similar to the new target and generates candidate molecules using in-context learning with the retrieved assay screening data. Assay2Mol outperforms recent machine learning approaches that generate candidate ligand molecules for target protein structures, while also promoting more synthesizable molecule generation.
Authors: Shreyas Chaudhari, Srinivasa Pranav, Jos\'e M. F. Moura
Abstract: Monotone gradient functions play a central role in solving the Monge formulation of the optimal transport (OT) problem, which arises in modern applications ranging from fluid dynamics to robot swarm control. When the transport cost is the squared Euclidean distance, Brenier's theorem guarantees that the unique optimal transport map satisfies a Monge-Amp\`ere equation and is the gradient of a convex function. In [arXiv:2301.10862] [arXiv:2404.07361], we proposed Monotone Gradient Networks (mGradNets), neural networks that directly parameterize the space of monotone gradient maps. In this work, we leverage mGradNets to directly learn the optimal transport mapping by minimizing a training loss function defined using the Monge-Amp\`ere equation. We empirically show that the structural bias of mGradNets facilitates the learning of optimal transport maps across both image morphing tasks and high-dimensional OT problems.
Authors: Yuqi Jin, Zhenhao Shuai, Zihan Hu, Weiteng Zhang, Weihao Xie, Jianwei Shuai, Xian Shen, Zhen Feng
Abstract: Background and Aims: Large language models (LLMs) have shown remarkable generalization and transfer capabilities by learning from vast corpora of text and web data. Their semantic representations allow cross-task knowledge transfer and reasoning, offering promising opportunities for data-scarce and heterogeneous domains such as clinical medicine. Yet, in diagnostic tasks like sarcopenia, major challenges remain: interpretability, transparency, and deployment efficiency. Traditional machine learning (TML) models provide stable performance and feature-level attribution, ensuring traceable and auditable decision logic, but lack semantic breadth. Conversely, LLMs enable flexible inference but often function as opaque predictors. Existing integration strategies remain shallow, rarely embedding the structured reasoning of TML into LLM inference. Methods: Using sarcopenia diagnosis as a case study, SHapley Additive exPlanations (SHAP) were extracted from a baseline XGBoost model and transformed into structured, LLM-compatible representations. An actor-critic reinforcement learning (RL) strategy guided the LLM to reason over these SHAP-based inputs, producing calibrated rationales and refined decision rules. The distilled reasoning was consolidated into a structured knowledge repository and deployed via retrieval-augmented generation (RAG) for case-based inference. Results: (Omitted here.) Conclusion: By coupling SHAP-derived statistical evidence with reinforcement-trained LLM reasoning, CANDLE mitigates the interpretability-performance trade-off, enhances predictive accuracy, and preserves high decision consistency. The framework offers a scalable approach to knowledge assetization of TML models, enabling interpretable, reproducible, and clinically aligned decision support in sarcopenia and potentially broader medical domains.
Authors: Yixin Shen
Abstract: Fine-tuning massive pre-trained language models across many tasks demands adapters that are both parameter-efficient and expressive. We introduce \textbf{Kron-LoRA}, a hybrid adapter that combines Kronecker-structured factorization with low-rank LoRA compression-an integration that, to our knowledge, has not been explored in parameter-efficient fine-tuning or in matrix approximation literature. Kron-LoRA achieves up to 4$\times$ fewer parameters than standard LoRA while retaining similar expressivity. Experiments on DistilBERT, Mistral-7B, LLaMA-2-7B, and LLaMA-3-8B across eight benchmarks show that Kron-LoRA matches or exceeds LoRA baselines with modest memory savings and only a 5-8\% speed overhead. In sequential fine-tuning, it also delivers competitive cross-task transfer despite using only one-quarter of the adapter parameters. Kron-LoRA thus offers a scalable, sustainable solution for multi-task adaptation of large language models.
Authors: Pritthijit Nath, Sebastian Schemm, Henry Moss, Peter Haynes, Emily Shuckburgh, Mark Webb
Abstract: Sub-grid parameterisations in climate models are traditionally static and tuned offline, limiting adaptability to evolving states. This work introduces FedRAIN-Lite, a federated reinforcement learning (FedRL) framework that mirrors the spatial decomposition used in general circulation models (GCMs) by assigning agents to latitude bands, enabling local parameter learning with periodic global aggregation. Using a hierarchy of simplified energy-balance climate models, from a single-agent baseline (ebm-v1) to multi-agent ensemble (ebm-v2) and GCM-like (ebm-v3) setups, we benchmark three RL algorithms under different FedRL configurations. Results show that Deep Deterministic Policy Gradient (DDPG) consistently outperforms both static and single-agent baselines, with faster convergence and lower area-weighted RMSE in tropical and mid-latitude zones across both ebm-v2 and ebm-v3 setups. DDPG's ability to transfer across hyperparameters and low computational cost make it well-suited for geographically adaptive parameter learning. This capability offers a scalable pathway towards high-complexity GCMs and provides a prototype for physically aligned, online-learning climate models that can evolve with a changing climate. Code accessible at https://github.com/p3jitnath/climate-rl-fedrl.
Authors: Zohair Shafi, Serdar Kadioglu
Abstract: Combinatorial optimization problems are ubiquitous in science and engineering. Still, learning-based approaches to accelerate combinatorial optimization often require solving a large number of difficult instances to collect training data, incurring significant computational cost. Existing learning-based methods require training dedicated models for each problem distribution, for each downstream task, severely limiting their scalability and generalization. We introduce Forge: Foundational Optimization Representations from Graph Embeddings, a framework that pre-trains a vector-quantized graph autoencoder on a large, diverse collection of mixed-integer programming (MIP) instances in an unsupervised manner, without relying on optimization solvers or optimal solutions. Vector quantization produces discrete code assignments that serve as a vocabulary for representing optimization instances. We evaluate Forge in both unsupervised and supervised settings. In the unsupervised setting, Forge embeddings effectively cluster unseen instances across problem domains and sizes. In the supervised setting, we fine-tune Forge embeddings and show that a single pre-trained model helps predicting both the integrality gap for cut-generation and variable hints for search guidance across multiple problem and size distributions. In both tasks, we improve the performance of a commercial optimization solver and outperform state-of-the-art learning-based methods. Finally, we open-source our training code, pre-trained Forge weights, and embeddings for multiple MIP distributions to foster further research in representation learning for optimization problems.
Authors: Maximilian Weimar, Lukas M. Rachbauer, Ilya Starshynov, Daniele Faccio, Linara Adilova, Dorian Bouchet, Stefan Rotter
Abstract: The estimation of continuous parameters from measured data plays a central role in many fields of physics. A key tool in understanding and improving such estimation processes is the concept of Fisher information, which quantifies how information about unknown parameters propagates through a physical system and determines the ultimate limits of precision. With Artificial Neural Networks (ANNs) gradually becoming an integral part of many measurement systems, it is essential to understand how they process and transmit parameter-relevant information internally. Here, we present a method to monitor the flow of Fisher information through an ANN performing a parameter estimation task, tracking it from the input to the output layer. We show that optimal estimation performance corresponds to the maximal transmission of Fisher information, and that training beyond this point results in information loss due to overfitting. This provides a model-free stopping criterion for network training-eliminating the need for a separate validation dataset. To demonstrate the practical relevance of our approach, we apply it to a network trained on data from an imaging experiment, highlighting its effectiveness in a realistic physical setting.
Authors: Senhao Liu, Zhiyu Guo, Zhiyuan Ji, Yueguo Chen, Yateng Tang, Yunhai Wang, Xuehao Zheng, Xiang Ao
Abstract: Typical financial risk management involves distinct phases for pre-service risk assessment and in-service default detection, often modeled separately. This paper proposes a novel framework, Multi-Granularity Knowledge Distillation (abbreviated as MGKD), aimed at improving pre-service risk prediction through the integration of in-service user behavior data. MGKD follows the idea of knowledge distillation, where the teacher model, trained on historical in-service data, guides the student model, which is trained on pre-service data. By using soft labels derived from in-service data, the teacher model helps the student model improve its risk prediction prior to service activation. Meanwhile, a multi-granularity distillation strategy is introduced, including coarse-grained, fine-grained, and self-distillation, to align the representations and predictions of the teacher and student models. This approach not only reinforces the representation of default cases but also enables the transfer of key behavioral patterns associated with defaulters from the teacher to the student model, thereby improving the overall performance of pre-service risk assessment. Moreover, we adopt a re-weighting strategy to mitigate the model's bias towards the minority class. Experimental results on large-scale real-world datasets from Tencent Mobile Payment demonstrate the effectiveness of our proposed approach in both offline and online scenarios.
Authors: Mujie Liu, Chenze Wang, Liping Chen, Nguyen Linh Dan Le, Niharika Tewari, Ting Dang, Jiangang Ma, Feng Xia
Abstract: The limited availability of labeled brain network data makes it challenging to achieve accurate and interpretable psychiatric diagnoses. While self-supervised learning (SSL) offers a promising solution, existing methods often rely on augmentation strategies that can disrupt crucial structural semantics in brain graphs. To address this, we propose SAM-BG, a two-stage framework for learning brain graph representations with structural semantic preservation. In the pre-training stage, an edge masker is trained on a small labeled subset to capture key structural semantics. In the SSL stage, the extracted structural priors guide a structure-aware augmentation process, enabling the model to learn more semantically meaningful and robust representations. Experiments on two real-world psychiatric datasets demonstrate that SAM-BG outperforms state-of-the-art methods, particularly in small-labeled data settings, and uncovers clinically relevant connectivity patterns that enhance interpretability. Our code is available at https://github.com/mjliu99/SAM-BG.
Authors: Zhengxi Lu, Jiabo Ye, Fei Tang, Yongliang Shen, Haiyang Xu, Ziwei Zheng, Weiming Lu, Ming Yan, Fei Huang, Jun Xiao, Yueting Zhuang
Abstract: Graphical User Interface (GUI) agents have demonstrated remarkable progress in automating complex user interface interactions through reinforcement learning. However, current approaches face a fundamental dilemma: offline RL enables stable training on pre-collected trajectories, but struggles with multi-step task execution for lack of trajectory-level reward signals; online RL captures these signals through environment interaction, but suffers from sparse rewards and prohibitive deployment costs. To address it, we present Semi-online Reinforcement Learning, a novel paradigm that simulates online RL on offline trajectories. During each rollout process, we preserve the original model output within the multi-turn dialogue, where a Patch Module adaptively recovers the divergence between rollout and expert trajectories. To capture long-term training signals, Semi-online RL introduces discounted future returns into the reward computation and optimizes the policy with weighted step-level and episode-level advantages. We further introduce Semi-Online Performance (SOP), a metric that aligns better with true online performance, serving as a practical and effective proxy for real-world evaluation. Experiments show that ours Semi-online RL achieves SOTA performance among 7B models across four dynamic benchmarks, with significant gains over the base model (e.g., +12.0% on AndroidWorld, +23.8% on AITW), demonstrating significant progress in bridging the gap between offline training efficiency and online multi-turn reasoning. The code is available at https://github.com/X-PLUG/MobileAgent/tree/main/UI-S1.
URLs: https://github.com/X-PLUG/MobileAgent/tree/main/UI-S1.
Authors: Zhengyao Huang, Daniel Zhengyu Huang, Tiannan Xiao, Dina Ma, Zhenyu Ming, Hao Shi, Yuanhui Wen
Abstract: Symbolic regression aims to discover concise, interpretable mathematical expressions that satisfy desired objectives, such as fitting data, posing a highly combinatorial optimization problem. While genetic programming has been the dominant approach, recent efforts have explored reinforcement learning methods for improving search efficiency. Monte Carlo Tree Search (MCTS), with its ability to balance exploration and exploitation through guided search, has emerged as a promising technique for symbolic expression discovery. However, its traditional bandit strategies and sequential symbol construction often limit performance. In this work, we propose an improved MCTS framework for symbolic regression that addresses these limitations through two key innovations: (1) an extreme bandit allocation strategy tailored for identifying globally optimal expressions, with finite-time performance guarantees under polynomial reward decay assumptions; and (2) evolution-inspired state-jumping actions such as mutation and crossover, which enable non-local transitions to promising regions of the search space. These state-jumping actions also reshape the reward landscape during the search process, improving both robustness and efficiency. We conduct a thorough numerical study to the impact of these improvements and benchmark our approach against existing symbolic regression methods on a variety of datasets, including both ground-truth and black-box datasets. Our approach achieves competitive performance with state-of-the-art libraries in terms of recovery rate, attains favorable positions on the Pareto frontier of accuracy versus model complexity. Code is available at https://github.com/PKU-CMEGroup/MCTS-4-SR.
Authors: Kaichen Xu, Yihang Du, Mianpeng Liu, Zimu Yu, Xiaobo Sun
Abstract: Positional encoding is essential for supplementing transformer with positional information of tokens. Existing positional encoding methods demand predefined token/feature order, rendering them unsuitable for real-world data with non-sequential yet causally-related features. To address this limitation, we propose CAPE, a novel method that identifies underlying causal structure over non-sequential features as a weighted directed acyclic graph (DAG) using generalized structural equation modeling. The DAG is then embedded in hyperbolic space where its geometric structure is well-preserved using a hyperboloid model-based approach that effectively captures two important causal graph properties (causal strength & causal specificity). This step yields causality-aware positional encodings for the features, which are converted into their rotary form for integrating with transformer's self-attention mechanism. Theoretical analysis reveals that CAPE-generated rotary positional encodings possess three valuable properties for enhanced self-attention, including causal distance-induced attenuation, causal generality-induced attenuation, and robustness to positional disturbances. We evaluate CAPE over both synthetic and real-word datasets, empirically demonstrating its theoretical properties and effectiveness in enhancing transformer for data with non-sequential features. Our code is available at https://github.com/Catchxu/CAPE.
Authors: Faramarz Farhangian, Leandro A. Ensina, George D. C. Cavalcanti, Rafael M. O. Cruz
Abstract: The rapid spread of information via social media has made text-based fake news detection critically important due to its societal impact. This paper presents a novel detection method called Dynamic Representation and Ensemble Selection (DRES) for identifying fake news based solely on text. DRES leverages instance hardness measures to estimate the classification difficulty for each news article across multiple textual feature representations. By dynamically selecting the textual representation and the most competent ensemble of classifiers for each instance, DRES significantly enhances prediction accuracy. Extensive experiments show that DRES achieves notable improvements over state-of-the-art methods, confirming the effectiveness of representation selection based on instance hardness and dynamic ensemble selection in boosting performance. Codes and data are available at: https://github.com/FFarhangian/FakeNewsDetection_DRES
Authors: Siming Zheng, Meifang Lan, Tong Wang, Yuanyuan Lin
Abstract: In this paper, we propose a general framework for testing the equality of the conditional distributions in a two-sample problem. This problem is most relevant to transfer learning under covariate shift. Our framework is built on neural network-based generative methods and sample splitting techniques by transforming the conditional distribution testing problem into an unconditional one. We introduce two special tests: the generative permutation-based conditional distribution equality test and the generative classification accuracy-based conditional distribution equality test. Theoretically, we establish a minimax lower bound for statistical inference in testing the equality of two conditional distributions under certain smoothness conditions. We demonstrate that the generative permutation-based conditional distribution equality test and its modified version can attain this lower bound precisely or up to some iterated logarithmic factor. Moreover, we prove the testing consistency of the generative classification accuracy-based conditional distribution equality test. We also establish the convergence rate for the learned conditional generator by deriving new results related to the recently-developed offset Rademacher complexity and approximation properties using neural networks. Empirically, we conduct numerical studies including synthetic datasets and two real-world datasets, demonstrating the effectiveness of our approach.
Authors: Jiaqi Weng, Han Zheng, Hanyu Zhang, Qinqin He, Jialing Tao, Hui Xue, Zhixuan Chu, Xiting Wang
Abstract: Increasing deployment of large language models (LLMs) in real-world applications raises significant safety concerns. Most existing safety research focuses on evaluating LLM outputs or specific safety tasks, limiting their ability to address broader, undefined risks. Sparse Autoencoders (SAEs) facilitate interpretability research to clarify model behavior by explaining single-meaning atomic features decomposed from entangled signals. jHowever, prior applications on SAEs do not interpret features with fine-grained safety-related concepts, thus inadequately addressing safety-critical behaviors, such as generating toxic responses and violating safety regulations. For rigorous safety analysis, we must extract a rich and diverse set of safety-relevant features that effectively capture these high-risk behaviors, yet face two challenges: identifying SAEs with the greatest potential for generating safety concept-specific neurons, and the prohibitively high cost of detailed feature explanation. In this paper, we propose Safe-SAIL, a framework for interpreting SAE features within LLMs to advance mechanistic understanding in safety domains. Our approach systematically identifies SAE with best concept-specific interpretability, explains safety-related neurons, and introduces efficient strategies to scale up the interpretation process. We will release a comprehensive toolkit including SAE checkpoints and human-readable neuron explanations, which supports empirical analysis of safety risks to promote research on LLM safety.
Authors: Jiazheng Kang, Le Huang, Cheng Hou, Zhe Zhao, Zhenxiang Yan, Chuan Shi, Ting Bai
Abstract: In real-world industrial settings, large language models (LLMs) must learn continually to keep pace with diverse and evolving tasks, requiring self-evolution to refine knowledge under dynamic data distributions. However, existing continual learning (CL) approaches, such as replay and parameter isolation, often suffer from catastrophic forgetting: training on new tasks degrades performance on earlier ones by overfitting to the new distribution and weakening generalization.We propose MoE-CL, a parameter-efficient adversarial mixture-of-experts framework for industrial-scale, self-evolving continual instruction tuning of LLMs. MoE-CL uses a dual-expert design: (1) a dedicated LoRA expert per task to preserve task-specific knowledge via parameter independence, mitigating forgetting; and (2) a shared LoRA expert to enable cross-task transfer. To prevent transferring task-irrelevant noise through the shared pathway, we integrate a task-aware discriminator within a GAN. The discriminator encourages the shared expert to pass only task-aligned information during sequential training. Through adversarial learning, the shared expert acquires generalized representations that mimic the discriminator, while dedicated experts retain task-specific details, balancing knowledge retention and cross-task generalization and thereby supporting self-evolution.Extensive experiments on the public MTL5 benchmark and an industrial Tencent3 benchmark validate the effectiveness of MoE-CL for continual instruction tuning. In real-world A/B testing for content compliance review on the Tencent Video platform, MoE-CL reduced manual review costs by 15.3%. These results demonstrate that MoE-CL is practical for large-scale industrial deployment where continual adaptation and stable transfer are critical.
Authors: Sarwan Ali
Abstract: We present a novel information-preserving Chaos Game Representation (CGR) method, also called Reverse-CGR (R-CGR), for biological sequence analysis that addresses the fundamental limitation of traditional CGR approaches - the loss of sequence information during geometric mapping. Our method introduces complete sequence recovery through explicit path encoding combined with rational arithmetic precision control, enabling perfect sequence reconstruction from stored geometric traces. Unlike purely geometric approaches, our reversibility is achieved through comprehensive path storage that maintains both positional and character information at each step. We demonstrate the effectiveness of R-CGR on biological sequence classification tasks, achieving competitive performance compared to traditional sequence-based methods while providing interpretable geometric visualizations. The approach generates feature-rich images suitable for deep learning while maintaining complete sequence information through explicit encoding, opening new avenues for interpretable bioinformatics analysis where both accuracy and sequence recovery are essential.
Authors: Mingchun Sun, Rongqiang Zhao, Hengrui Hu, Songyu Ding, Jie Liu
Abstract: Diffusion models are the mainstream approach for time series generation tasks. However, existing diffusion models for time series generation require retraining the entire framework to introduce specific conditional guidance. There also exists a certain degree of distributional bias between the generated data and the real data, which leads to potential model biases in downstream tasks. Additionally, the complexity of diffusion models and the latent spaces leads to an uninterpretable inference process. To address these issues, we propose the data style-guided diffusion model (DS-Diffusion). In the DS-Diffusion, a diffusion framework based on style-guided kernels is developed to avoid retraining for specific conditions. The time-information based hierarchical denoising mechanism (THD) is developed to reduce the distributional bias between the generated data and the real data. Furthermore, the generated samples can clearly indicate the data style from which they originate. We conduct comprehensive evaluations using multiple public datasets to validate our approach. Experimental results show that, compared to the state-of-the-art model such as ImagenTime, the predictive score and the discriminative score decrease by 5.56% and 61.55%, respectively. The distributional bias between the generated data and the real data is further reduced, the inference process is also more interpretable. Moreover, by eliminating the need to retrain the diffusion model, the flexibility and adaptability of the model to specific conditions are also enhanced.
Authors: Kashaf Ul Emaan
Abstract: Detection of credit card fraud is an acute issue of financial security because transaction datasets are highly lopsided, with fraud cases being only a drop in the ocean. Balancing datasets using the most popular methods of traditional oversampling such as the Synthetic Minority Oversampling Technique (SMOTE) generally create simplistic synthetic samples that are not readily applicable to complex fraud patterns. Recent industry advances that include Conditional Tabular Generative Adversarial Networks (CTGAN) and Tabular Variational Autoencoders (TVAE) have demonstrated increased efficiency in tabular synthesis, yet all these models still exhibit issues with high-dimensional dependence modelling. Now we will present our hybrid approach where we use a Generative Adversarial Network (GAN) with a Transformer encoder block to produce realistic fraudulent transactions samples. The GAN architecture allows training realistic generators adversarial, and the Transformer allows the model to learn rich feature interactions by self-attention. Such a hybrid strategy overcomes the limitations of SMOTE, CTGAN, and TVAE by producing a variety of high-quality synthetic minority classes samples. We test our algorithm on the publicly-available Credit Card Fraud Detection dataset and compare it to conventional and generative resampling strategies with a variety of classifiers, such as Logistic Regression (LR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM). Findings indicate that our Transformer-based GAN shows substantial gains in Recall, F1-score and Area Under the Receiver Operating Characteristic Curve (AUC), which indicates that it is effective in overcoming the severe class imbalance inherent in the task of fraud detection.
Authors: Binghui Li, Fengling Chen, Zixun Huang, Lean Wang, Lei Wu
Abstract: Scaling laws have played a cornerstone role in guiding the training of large language models (LLMs). However, most existing works on scaling laws primarily focus on the final-step loss, overlooking the loss dynamics during the training process and, crucially, the impact of learning rate schedule (LRS). In this paper, we aim to bridge this gap by studying a teacher-student kernel regression setup trained via online stochastic gradient descent (SGD). Leveraging a novel intrinsic time viewpoint and stochastic differential equation (SDE) modeling of SGD, we introduce the Functional Scaling Law (FSL), which characterizes the evolution of population risk during the training process for general LRSs. Remarkably, the impact of the LRSs is captured through an explicit convolution-type functional term, making their effects fully tractable. To illustrate the utility of FSL, we analyze three widely used LRSs -- constant, exponential decay, and warmup-stable-decay (WSD) -- under both data-limited and compute-limited regimes. We provide theoretical justification for widely adopted empirical practices in LLMs pre-training such as (i) higher-capacity models are more data- and compute-efficient; (ii) learning rate decay can improve training efficiency; (iii) WSD-like schedules can outperform direct-decay schedules. Lastly, we explore the practical relevance of FSL as a surrogate model for fitting, predicting and optimizing the loss curves in LLM pre-training, with experiments conducted across model sizes ranging from 0.1B to 1B parameters. We hope our FSL framework can deepen the understanding of LLM pre-training dynamics and provide insights for improving large-scale model training.
Authors: Hyeon Jeon, Micha\"el Aupetit, Soohyun Lee, Kwon Ko, Youngtaek Kim, Ghulam Jilani Quadri, Jinwook Seo
Abstract: Brushing is a common interaction technique in 2D scatterplots, allowing users to select clustered points within a continuous, enclosed region for further analysis or filtering. However, applying conventional brushing to 2D representations of multidimensional (MD) data, i.e., Multidimensional Projections (MDPs), can lead to unreliable cluster analysis due to MDP-induced distortions that inaccurately represent the cluster structure of the original MD data. To alleviate this problem, we introduce a novel brushing technique for MDPs called Distortion-aware brushing. As users perform brushing, Distortion-aware brushing corrects distortions around the currently brushed points by dynamically relocating points in the projection, pulling data points close to the brushed points in MD space while pushing distant ones apart. This dynamic adjustment helps users brush MD clusters more accurately, leading to more reliable cluster analysis. Our user studies with 24 participants show that Distortion-aware brushing significantly outperforms previous brushing techniques for MDPs in accurately separating clusters in the MD space and remains robust against distortions. We further demonstrate the effectiveness of our technique through two use cases: (1) conducting cluster analysis of geospatial data and (2) interactively labeling MD clusters.
Authors: Zhanyu Wang, Guang Cheng, Jordan Awan
Abstract: Differentially private (DP) mechanisms protect individual-level information by introducing randomness into the statistical analysis procedure. Despite the availability of numerous DP tools, there remains a lack of general techniques for conducting statistical inference under DP. We examine a DP bootstrap procedure that releases multiple private bootstrap estimates to infer the sampling distribution and construct confidence intervals (CIs). Our privacy analysis presents new results on the privacy cost of a single DP bootstrap estimate, applicable to any DP mechanism, and identifies some misapplications of the bootstrap in the existing literature. For the composition of the DP bootstrap, we present a numerical method to compute the exact privacy cost of releasing multiple DP bootstrap estimates, and using the Gaussian-DP (GDP) framework (Dong et al., 2022), we show that the release of $B$ DP bootstrap estimates from mechanisms satisfying $(\mu/\sqrt{(2-2/\mathrm{e})B})$-GDP asymptotically satisfies $\mu$-GDP as $B$ goes to infinity. Then, we perform private statistical inference by post-processing the DP bootstrap estimates. We prove that our point estimates are consistent, our standard CIs are asymptotically valid, and both enjoy optimal convergence rates. To further improve the finite performance, we use deconvolution with DP bootstrap estimates to accurately infer the sampling distribution. We derive CIs for tasks such as population mean estimation, logistic regression, and quantile regression, and we compare them to existing methods using simulations and real-world experiments on 2016 Canada Census data. Our private CIs achieve the nominal coverage level and offer the first approach to private inference for quantile regression.
Authors: Yongsong Huang, Tomo Miyazaki, Xiaofeng Liu, Shinichiro Omachi
Abstract: Image Super-Resolution (SR) is essential for a wide range of computer vision and image processing tasks. Investigating infrared (IR) image (or thermal images) super-resolution is a continuing concern within the development of deep learning. This survey aims to provide a comprehensive perspective of IR image super-resolution, including its applications, hardware imaging system dilemmas, and taxonomy of image processing methodologies. In addition, the datasets and evaluation metrics in IR image super-resolution tasks are also discussed. Furthermore, the deficiencies in current technologies and possible promising directions for the community to explore are highlighted. To cope with the rapid development in this field, we intend to regularly update the relevant excellent work at https://github.com/yongsongH/Infrared_Image_SR_Survey.
URLs: https://github.com/yongsongH/Infrared_Image_SR_Survey.
Authors: Yuecheng Li, Lele Fu, Sheng Huang, Chuan Chen, Lei Yang, Zibin Zheng
Abstract: Recently, graph contrastive learning (GCL) has emerged as one of the optimal solutions for node-level and supervised tasks. However, for structure-related and unsupervised tasks such as graph clustering, current GCL algorithms face difficulties acquiring the necessary cluster-level information, resulting in poor performance. In addition, general unsupervised GCL improves the performance of downstream tasks by increasing the number of negative samples, which leads to severe class collision and unfairness of graph clustering. To address the above issues, we propose a Cluster-aware Graph Contrastive Learning Framework (CueGCL) to jointly learn clustering results and node representations. Specifically, we design a personalized self-training (PeST) strategy for unsupervised scenarios, which enables our model to capture precise cluster-level personalized information. With the benefit of the PeST, we alleviate class collision and unfairness without sacrificing the overall model performance. Furthermore, aligned graph clustering (AGC) is employed to obtain the cluster partition, where we align the clustering space of our downstream task with that in PeST to achieve more consistent node embeddings. Finally, we theoretically demonstrate the effectiveness of our model, showing it yields an embedding space with a significantly discernible cluster structure. Extensive experimental results also show our CueGCL exhibits state-of-the-art performance on five benchmark datasets with different scales.
Authors: Sohee Kim, Jisu Kang, Dunam Kim, Seokju Lee
Abstract: In this paper, we demonstrate that CLIP can also be adapted to downstream tasks where its vision-language alignment is suboptimally learned during pre-training on web-crawled data, all without requiring fine-tuning. We explore the case of monocular depth estimation, where CLIP's contrastive prior struggles to generalize, compared to its success in domains such as generative modeling and semantic segmentation. Since CLIP fails to consistently capture similarities between image patches and natural language prompts describing distance, we eliminate the use of its pre-trained natural language token embeddings and distill the semantic prior of its frozen text encoder into a single learnable embedding matrix called "mirror". The main design goal of mirror is to derive a non-human language prompt that approximates an optimal natural language prompt: "How far is this location from the camera?" Using this approach, we jointly train two lightweight modules, a mirror and a compact decoder, on top of a frozen CLIP for dense depth prediction. Compared to conventional depth models, our framework is significantly more efficient in terms of parameters and computation. The resulting model exhibits impressive performance, matching several state-of-the-art vision models on the NYU Depth v2 and KITTI benchmark datasets, while outperforming all vision-language depth models based on a frozen CLIP prior. Experiments demonstrate that the suboptimal depth understanding of CLIP in terms of spatial and temporal consistency can be significantly corrected without either fine-tuning it or concatenating mirror with its pre-trained subword token embeddings. Furthermore, an ablation study on the convergence status of mirror shows that it is implicitly trained to capture objects, such as humans and windows, where semantic cues play an important role in detection.
Authors: Jona te Lintelo, Stefanos Koffas, Stjepan Picek
Abstract: Sponge attacks aim to increase the energy consumption and computation time of neural networks. In this work, we present a novel sponge attack called SkipSponge. SkipSponge is the first sponge attack that is performed directly on the parameters of a pretrained model using only a few data samples. Our experiments show that SkipSponge can successfully increase the energy consumption of image classification models, GANs, and autoencoders, requiring fewer samples than the state-of-the-art sponge attacks (Sponge Poisoning). We show that poisoning defenses are ineffective if not adjusted specifically for the defense against SkipSponge (i.e., they decrease target layer bias values) and that SkipSponge is more effective on the GANs and the autoencoders than Sponge Poisoning. Additionally, SkipSponge is stealthy as it does not require significant changes to the victim model's parameters. Our experiments indicate that SkipSponge can be performed even when an attacker has access to less than 1% of the entire training dataset and reaches up to 13% energy increase.
Authors: Wenpin Tang, Fuzhong Zhou
Abstract: This paper aims to develop and provide a rigorous treatment to the problem of entropy regularized fine-tuning in the context of continuous-time diffusion models, which was recently proposed by Uehara et al. (arXiv:2402.15194, 2024). The idea is to use stochastic control for sample generation, where the entropy regularizer is introduced to mitigate reward collapse. We also show how the analysis can be extended to fine-tuning with a general $f$-divergence regularizer. Numerical experiments on large-scale text-to-image models--Stable Diffusion v1.5 are conducted to validate our approach.
Authors: Xingran Chen, Navid NaderiAlizadeh, Alejandro Ribeiro, Shirin Saeedi Bidokhti
Abstract: We address the challenge of sampling and remote estimation for autoregressive Markovian processes in a multi-hop wireless network with statistically-identical agents. Agents cache the most recent samples from others and communicate over wireless collision channels governed by an underlying graph topology. Our goal is to minimize time-average estimation error and/or age of information with decentralized scalable sampling and transmission policies, considering both oblivious (where decision-making is independent of the physical processes) and non-oblivious policies (where decision-making depends on physical processes). We prove that in oblivious policies, minimizing estimation error is equivalent to minimizing the age of information. The complexity of the problem, especially the multi-dimensional action spaces and arbitrary network topologies, makes theoretical methods for finding optimal transmission policies intractable. We optimize the policies using a graphical multi-agent reinforcement learning framework, where each agent employs a permutation-equivariant graph neural network architecture. Theoretically, we prove that our proposed framework exhibits desirable transferability properties, allowing transmission policies trained on small- or moderate-size networks to be executed effectively on large-scale topologies. Numerical experiments demonstrate that (i) Our proposed framework outperforms state-of-the-art baselines; (ii) The trained policies are transferable to larger networks, and their performance gains increase with the number of agents; (iii) The training procedure withstands non-stationarity even if we utilize independent learning techniques; and, (iv) Recurrence is pivotal in both independent learning and centralized training and decentralized execution, and improves the resilience to non-stationarity in independent learning.
Authors: Andrea Carriero, Davide Pettenuzzo, Shubhranshu Shekhar
Abstract: This paper presents a comparative analysis evaluating the accuracy of Large Language Models (LLMs) against traditional macro time series forecasting approaches. In recent times, LLMs have surged in popularity for forecasting due to their ability to capture intricate patterns in data and quickly adapt across very different domains. However, their effectiveness in forecasting macroeconomic time series data compared to conventional methods remains an area of interest. To address this, we conduct a rigorous evaluation of LLMs against traditional macro forecasting methods, using as common ground the FRED-MD database. Our findings provide valuable insights into the strengths and limitations of LLMs in forecasting macroeconomic time series, shedding light on their applicability in real-world scenarios
Authors: Jing Yu Koh, Stephen McAleer, Daniel Fried, Ruslan Salakhutdinov
Abstract: Autonomous agents powered by language models (LMs) have demonstrated promise in their ability to perform decision-making tasks such as web automation. However, a key limitation remains: LMs, primarily optimized for natural language understanding and generation, struggle with multi-step reasoning, planning, and using environmental feedback when attempting to solve realistic computer tasks. Towards addressing this, we propose an inference-time search algorithm for LM agents to explicitly perform exploration and multi-step planning in interactive web environments. Our approach is a form of best-first tree search that operates within the actual environment space, and is complementary with most existing state-of-the-art agents. It is the first tree search algorithm for LM agents that shows effectiveness on realistic web tasks. On the challenging VisualWebArena benchmark, applying our search algorithm on top of a GPT-4o agent yields a 39.7% relative increase in success rate compared to the same baseline without search, setting a state-of-the-art success rate of 26.4%. On WebArena, search also yields a 28.0% relative improvement over a baseline agent, setting a competitive success rate of 19.2%. Our experiments highlight the effectiveness of search for web agents, and we demonstrate that performance scales with increased test-time compute. We conduct a thorough analysis of our results to highlight improvements from search, limitations, and promising directions for future work. Our code and models are publicly released at https://jykoh.com/search-agents.
Authors: Connor Lawless, Oktay Gunluk
Abstract: Clustering is a well-studied unsupervised learning task that aims to partition data points into a number of clusters. In many applications, these clusters correspond to real-world constructs (e.g., electoral districts, playlists, TV channels), where a group (e.g., social or demographic) benefits only if it reaches a minimum level of representation in the cluster (e.g., 50% to elect their preferred candidate). In this paper, we study the k-means and k-medians clustering problems under the additional fairness constraint that each group must attain a minimum level of representation in at least a specified number of clusters. We formulate this problem as a mixed-integer (nonlinear) optimization problem and propose an alternating minimization algorithm, called MiniReL, to solve it. Although incorporating fairness constraints results in an NP-hard assignment problem within the MiniReL algorithm, we present several heuristic strategies that make the approach practical even for large datasets. Numerical results demonstrate that our method yields fair clusters without increasing clustering cost across standard benchmark datasets.
Authors: Lisa Benato, Wahid Bhimji, Paolo Calafiura, Ragansu Chakkappai, Po-Wen Chang, Yuan-Tang Chou, Sascha Diefenbacher, Jordan Dudley, Ibrahim Elsharkawy, Steven Farrell, Aishik Ghosh, Cristina Giordano, Isabelle Guyon, Chris Harris, Yota Hashizume, Shih-Chieh Hsu, Elham E. Khoda, Claudius Krause, Ang Li, Benjamin Nachman, Peter Nugent, David Rousseau, Robert Schoefbeck, Maryam Shooshtari, Dennis Schwarz, Benjamin Thorne, Ihsan Ullah, Daohan Wang, Yulei Zhang
Abstract: The FAIR Universe HiggsML Uncertainty Challenge focused on measuring the physical properties of elementary particles with imperfect simulators. Participants were required to compute and report confidence intervals for a parameter of interest regarding the Higgs boson while accounting for various systematic (epistemic) uncertainties. The dataset is a tabular dataset of 28 features and 280 million instances. Each instance represents a simulated proton-proton collision as observed at CERN's Large Hadron Collider in Geneva, Switzerland. The features of these simulations were chosen to capture key characteristics of different types of particles. These include primary attributes, such as the energy and three-dimensional momentum of the particles, as well as derived attributes, which are calculated from the primary ones using domain-specific knowledge. Additionally, a label feature designates each instance's type of proton-proton collision, distinguishing the Higgs boson events of interest from three background sources. As outlined in this paper, the permanent release of the dataset allows long-term benchmarking of new techniques. The leading submissions, including Contrastive Normalising Flows and Density Ratios estimation through classification, are described. Our challenge has brought together the physics and machine learning communities to advance our understanding and methodologies in handling systematic uncertainties within AI techniques.
Authors: Jing Xiong, Jianghan Shen, Fanghua Ye, Chaofan Tao, Zhongwei Wan, Jianqiao Lu, Xun Wu, Chuanyang Zheng, Zhijiang Guo, Min Yang, Lingpeng Kong, Ngai Wong
Abstract: Deploying large language models (LLMs) for long-context inference remains challenging due to their substantial memory and computational demands. While techniques such as Key-Value (KV) cache compression are designed to reduce memory usage, they often neglect the structured sparsity inherent in the relationship between hidden states and their corresponding KV cache. In this work, we explore the role of uncertainty as a potential indicator of sparsity within LLMs. We propose UNComp, an uncertainty-aware framework that leverages truncated matrix entropy to identify areas of low information content, thereby revealing sparsity patterns that can be used for adaptive compression. Unlike traditional methods that apply uniform compression, UNComp dynamically adjusts its approach to compression, guided by uncertainty measures that reflect the importance of various model components. Our analysis shows that sparsity patterns, when derived from uncertainty estimates, can be exploited to reveal special long-range dependencies, such as retrieval heads and retrieval layers. This perspective not only enhances our understanding of how compression can be optimized but also provides new insights into the inherent sparsity of LLMs during long-context inference. By focusing on uncertainty to analyze the sparsity pattern in detail, UNComp reduces the KV cache size to 4.74% of the original, achieves a 6% prefill speedup, and improves throughput by 6.4x - not only delivering strong lossless compression performance, but also validating the effectiveness of the underlying theoretical tool. We release the code at https://github.com/menik1126/UNComp.
Authors: Beichen Wang, Juexiao Zhang, Shuwen Dong, Irving Fang, Chen Feng
Abstract: Vision Language Models (VLMs) have recently been adopted in robotics for their capability in common sense reasoning and generalizability. Existing work has applied VLMs to generate task and motion planning from natural language instructions and simulate training data for robot learning. In this work, we explore using VLM to interpret human demonstration videos and generate robot task planning. Our method integrates keyframe selection, visual perception, and VLM reasoning into a pipeline. We named it SeeDo because it enables the VLM to ''see'' human demonstrations and explain the corresponding plans to the robot for it to ''do''. To validate our approach, we collected a set of long-horizon human videos demonstrating pick-and-place tasks in three diverse categories and designed a set of metrics to comprehensively benchmark SeeDo against several baselines, including state-of-the-art video-input VLMs. The experiments demonstrate SeeDo's superior performance. We further deployed the generated task plans in both a simulation environment and on a real robot arm.
Authors: Huancheng Chen, Jingtao Li, Weiming Zhuang, Chen Chen, Lingjuan Lyu
Abstract: We revisit continual learning~(CL), which enables pre-trained vision transformers (ViTs) to sequentially fine-tune on new downstream tasks over time. However, as the scale of these models increases, catastrophic forgetting remains a more serious challenge. Recent studies highlight a crossover between CL techniques and parameter-efficient fine-tuning (PEFT), which focuses on fine-tuning only a small set of trainable parameters to adapt to downstream tasks, such as low-rank adaptation (LoRA). While LoRA achieves faster convergence and requires fewer trainable parameters, it has seldom been explored in the context of continual learning. To address this gap, we propose a novel PEFT-CL method called Dual Low-Rank Adaptation (DualLoRA), which introduces both an orthogonal LoRA adapter and a residual LoRA adapter parallel to pre-trained weights in each layer. These components are orchestrated by a dynamic memory mechanism to strike a balance between stability and plasticity. Additionally, we propose a scheme to predict task identity with confidence and calibrate the model's outputs accordingly. On ViT-based models, we demonstrate that DualLoRA offers significant advantages in accuracy, inference speed, and computation efficiency in training over existing CL methods across multiple benchmarks.
Authors: Haitham Kanj, Seonho Kim, Kiryung Lee
Abstract: This paper presents Sparse Gradient Descent as a solution for variable selection in convex piecewise linear regression, where the model is given as the maximum of $k$-affine functions $ x \mapsto \max_{j \in [k]} \langle a_j^\star, x \rangle + b_j^\star$ for $j = 1,\dots,k$. Here, $\{ a_j^\star\}_{j=1}^k$ and $\{b_j^\star\}_{j=1}^k$ denote the ground-truth weight vectors and intercepts. A non-asymptotic local convergence analysis is provided for Sp-GD under sub-Gaussian noise when the covariate distribution satisfies the sub-Gaussianity and anti-concentration properties. When the model order and parameters are fixed, Sp-GD provides an $\epsilon$-accurate estimate given $\mathcal{O}(\max(\epsilon^{-2}\sigma_z^2,1)s\log(d/s))$ observations where $\sigma_z^2$ denotes the noise variance. This also implies the exact parameter recovery by Sp-GD from $\mathcal{O}(s\log(d/s))$ noise-free observations. The proposed initialization scheme uses sparse principal component analysis to estimate the subspace spanned by $\{ a_j^\star\}_{j=1}^k$, then applies an $r$-covering search to estimate the model parameters. A non-asymptotic analysis is presented for this initialization scheme when the covariates and noise samples follow Gaussian distributions. When the model order and parameters are fixed, this initialization scheme provides an $\epsilon$-accurate estimate given $\mathcal{O}(\epsilon^{-2}\max(\sigma_z^4,\sigma_z^2,1)s^2\log^4(d))$ observations. A new transformation named Real Maslov Dequantization (RMD) is proposed to transform sparse generalized polynomials into sparse max-affine models. The error decay rate of RMD is shown to be exponentially small in its temperature parameter. Furthermore, theoretical guarantees for Sp-GD are extended to the bounded noise model induced by RMD. Numerical Monte Carlo results corroborate theoretical findings for Sp-GD and the initialization scheme.
Authors: Heehwan Wang, Joonwoo Kwon, Sooyoung Kim, Jungwoo Seo, Shinjae Yoo, Yuewei Lin, Jiook Cha
Abstract: Music style transfer enables personalized music creation by blending the structure of a source with the stylistic attributes of a reference. Existing text-conditioned and diffusion-based approaches show promise but often require paired datasets, extensive training, or detailed annotations. We present Stylus, a training-free framework that repurposes a pre-trained Stable Diffusion model for music style transfer in the mel-spectrogram domain. Stylus manipulates self-attention by injecting style key-value features while preserving source queries to maintain musical structure. To improve fidelity, we introduce a phase-preserving reconstruction strategy that avoids artifacts from Griffin-Lim reconstruction, and we adopt classifier-free-guidance-inspired control for adjustable stylization and multi-style blending. In extensive evaluations, Stylus outperforms state-of-the-art baselines, achieving 34.1% higher content preservation and 25.7% better perceptual quality without any additional training.
Authors: Philipp Wolters, Johannes Gilg, Torben Teepe, Fabian Herzog, Felix Fent, Gerhard Rigoll
Abstract: In this work, we present SpaRC, a novel Sparse fusion transformer for 3D perception that integrates multi-view image semantics with Radar and Camera point features. The fusion of radar and camera modalities has emerged as an efficient perception paradigm for autonomous driving systems. While conventional approaches utilize dense Bird's Eye View (BEV)-based architectures for depth estimation, contemporary query-based transformers excel in camera-only detection through object-centric methodology. However, these query-based approaches exhibit limitations in false positive detections and localization precision due to implicit depth modeling. We address these challenges through three key contributions: (1) sparse frustum fusion (SFF) for cross-modal feature alignment, (2) range-adaptive radar aggregation (RAR) for precise object localization, and (3) local self-attention (LSA) for focused query aggregation. In contrast to existing methods requiring computationally intensive BEV-grid rendering, SpaRC operates directly on encoded point features, yielding substantial improvements in efficiency and accuracy. Empirical evaluations on the nuScenes and TruckScenes benchmarks demonstrate that SpaRC significantly outperforms existing dense BEV-based and sparse query-based detectors. Our method achieves state-of-the-art performance metrics of 67.1 NDS and 63.1 AMOTA. The code and pretrained models are available at https://github.com/phi-wol/sparc.
Authors: Saumya Saxena, Blake Buchanan, Chris Paxton, Peiqi Liu, Bingqing Chen, Narunas Vaskevicius, Luigi Palmieri, Jonathan Francis, Oliver Kroemer
Abstract: In Embodied Question Answering (EQA), agents must explore and develop a semantic understanding of an unseen environment to answer a situated question with confidence. This problem remains challenging in robotics, due to the difficulties in obtaining useful semantic representations, updating these representations online, and leveraging prior world knowledge for efficient planning and exploration. To address these limitations, we propose GraphEQA, a novel approach that utilizes real-time 3D metric-semantic scene graphs (3DSGs) and task relevant images as multi-modal memory for grounding Vision-Language Models (VLMs) to perform EQA tasks in unseen environments. We employ a hierarchical planning approach that exploits the hierarchical nature of 3DSGs for structured planning and semantics-guided exploration. We evaluate GraphEQA in simulation on two benchmark datasets, HM-EQA and OpenEQA, and demonstrate that it outperforms key baselines by completing EQA tasks with higher success rates and fewer planning steps. We further demonstrate GraphEQA in multiple real-world home and office environments.
Authors: Thibault de Surrel, Fabien Lotte, Sylvain Chevallier, Florian Yger
Abstract: Circular and non-flat data distributions are prevalent across diverse domains of data science, yet their specific geometric structures often remain underutilized in machine learning frameworks. A principled approach to accounting for the underlying geometry of such data is pivotal, particularly when extending statistical models, like the pervasive Gaussian distribution. In this work, we tackle those issue by focusing on the manifold of symmetric positive definite (SPD) matrices, a key focus in information geometry. We introduce a non-isotropic wrapped Gaussian by leveraging the exponential map, we derive theoretical properties of this distribution and propose a maximum likelihood framework for parameter estimation. Furthermore, we reinterpret established classifiers on SPD through a probabilistic lens and introduce new classifiers based on the wrapped Gaussian model. Experiments on synthetic and real-world datasets demonstrate the robustness and flexibility of this geometry-aware distribution, underscoring its potential to advance manifold-based data analysis. This work lays the groundwork for extending classical machine learning and statistical methods to more complex and structured data.
Authors: Yajie Ji, Jiuyang Liang, Zhenli Xu
Abstract: Machine-learning interatomic potentials have emerged as a revolutionary class of force-field models in molecular simulations, delivering quantum-mechanical accuracy at a fraction of the computational cost and enabling the simulation of large-scale systems over extended timescales. However, they often focus on modeling local environments, neglecting crucial long-range interactions. We propose a Sum-of-Gaussians Neural Network (SOG-Net), a lightweight and versatile framework for integrating long-range interactions into machine learning force field. The SOG-Net employs a latent-variable learning network that seamlessly bridges short-range and long-range components, coupled with an efficient Fourier convolution layer that incorporates long-range effects. By learning sum-of-Gaussians multipliers across different convolution layers, the SOG-Net adaptively captures diverse long-range decay behaviors while maintaining close-to-linear computational complexity during training and simulation via non-uniform fast Fourier transforms. The method is demonstrated effective for a broad range of long-range systems.
Authors: Antoine Chatalic, Marco Letizia, Nicolas Schreuder, Lorenzo Rosasco
Abstract: Two-sample hypothesis testing-determining whether two sets of data are drawn from the same distribution-is a fundamental problem in statistics and machine learning with broad scientific applications. In the context of nonparametric testing, maximum mean discrepancy (MMD) has gained popularity as a test statistic due to its flexibility and strong theoretical foundations. However, its use in large-scale scenarios is plagued by high computational costs. In this work, we use a Nystr\"om approximation of the MMD to design a computationally efficient and practical testing algorithm while preserving statistical guarantees. Our main result is a finite-sample bound on the power of the proposed test for distributions that are sufficiently separated with respect to the MMD. The derived separation rate matches the known minimax optimal rate in this setting. We support our findings with a series of numerical experiments, emphasizing applicability to realistic scientific data.
Authors: Yasin Sonmez, Hanna Krasowski, Murat Arcak
Abstract: Imitation learning is a promising approach for training autonomous vehicles (AV) to navigate complex traffic environments by mimicking expert driver behaviors. While existing imitation learning frameworks focus on leveraging expert demonstrations, they often overlook the potential of additional complex driving data from surrounding traffic participants. In this paper, we study a data augmentation strategy that leverages the observed trajectories of nearby vehicles, captured by the AV's sensors, as additional demonstrations. We introduce a simple vehicle-selection sampling and filtering strategy that prioritizes informative and diverse driving behaviors, contributing to a richer dataset for training. We evaluate this idea with a representative learning-based planner on a large real-world dataset and demonstrate improved performance in complex driving scenarios. Specifically, the approach reduces collision rates and improves safety metrics compared to the baseline. Notably, even when using only 10 percent of the original dataset, the method matches or exceeds the performance of the full dataset. Through ablations, we analyze selection criteria and show that naive random selection can degrade performance. Our findings highlight the value of leveraging diverse real-world trajectory data in imitation learning and provide insights into data augmentation strategies for autonomous driving.
Authors: Ruoxi Cheng, Haoxuan Ma, Weixin Wang, Ranjie Duan, Jiexi Liu, Xiaoshuang Jia, Simeng Qin, Xiaochun Cao, Yang Liu, Xiaojun Jia
Abstract: Alignment is vital for safely deploying large language models (LLMs). Existing techniques are either reward-based (train a reward model on preference pairs and optimize with reinforcement learning) or reward-free (directly fine-tune on ranked outputs). Recent research shows that well-tuned reward-based pipelines remain robust, and single-response demonstrations can outperform pairwise preference data. However, two challenges persist: (1) imbalanced safety datasets that overrepresent common hazards while neglecting long-tail threats; and (2) static reward models that ignore task difficulty, limiting optimization efficiency and attainable gains. We propose DR-IRL (Dynamically adjusting Rewards through Inverse Reinforcement Learning). We first train category-specific reward models using a balanced safety dataset covering seven harmful categories via IRL. Then we enhance Group Relative Policy Optimization (GRPO) by introducing dynamic reward scaling--adjusting rewards by task difficulty--data-level hardness by text encoder cosine similarity, model-level responsiveness by reward gaps. Extensive experiments across various benchmarks and LLMs demonstrate that DR-IRL outperforms all baseline methods in safety alignment while maintaining usefulness.
Authors: Yangxiao Lu, Ruosen Li, Liqiang Jing, Jikai Wang, Xinya Du, Yunhui Guo, Nicholas Ruozzi, Yu Xiang
Abstract: Visual grounding focuses on detecting objects from images based on language expressions. Recent Large Vision-Language Models (LVLMs) have significantly advanced visual grounding performance by training large models with large-scale datasets. However, the problem remains challenging, especially when similar objects appear in the input image. For example, an LVLM may not be able to differentiate Diet Coke and regular Coke in an image. In this case, if additional reference images of Diet Coke and regular Coke are available, it can help the visual grounding of similar objects. In this work, we introduce a new task named Multimodal Reference Visual Grounding (MRVG). In this task, a model has access to a set of reference images of objects in a database. Based on these reference images and a language expression, the model is required to detect a target object from a query image. We first introduce a new dataset to study the MRVG problem. Then we introduce a novel method, named MRVG-Net, to solve this visual grounding problem. We show that by efficiently using reference images with few-shot object detection and using Large Language Models (LLMs) for object matching, our method achieves superior visual grounding performance compared to the state-of-the-art LVLMs such as Qwen2.5-VL-72B. Our approach bridges the gap between few-shot detection and visual grounding, unlocking new capabilities for visual understanding, which has wide applications in robotics. Project page with our video, code, and dataset: https://irvlutd.github.io/MultiGrounding
Authors: Ming Li, Ruiyi Zhang, Jian Chen, Chenguang Wang, Jiuxiang Gu, Yufan Zhou, Franck Dernoncourt, Wanrong Zhu, Tianyi Zhou, Tong Sun
Abstract: Despite the existing evolution of Multimodal Large Language Models (MLLMs), a non-neglectable limitation remains in their struggle with visual text grounding, especially in text-rich images of documents. Document images, such as scanned forms and infographics, highlight critical challenges due to their complex layouts and textual content. However, current benchmarks do not fully address these challenges, as they mostly focus on visual grounding on natural images, rather than text-rich document images. Thus, to bridge this gap, we introduce TRIG, a novel task with a newly designed instruction dataset for benchmarking and improving the Text-Rich Image Grounding capabilities of MLLMs in document question-answering. Specifically, we propose an OCR-LLM-human interaction pipeline to create 800 manually annotated question-answer pairs as a benchmark and a large-scale training set of 90$ synthetic data based on four diverse datasets. A comprehensive evaluation of various MLLMs on our proposed benchmark exposes substantial limitations in their grounding capability on text-rich images. In addition, we propose two simple and effective TRIG methods based on general instruction tuning and plug-and-play efficient embedding, respectively. By finetuning MLLMs on our synthetic dataset, they promisingly improve spatial reasoning and grounding capabilities.
Authors: Margalit Glasgow, Denny Wu, Joan Bruna
Abstract: We study the approximation gap between the dynamics of a polynomial-width neural network and its infinite-width counterpart, both trained using projected gradient descent in the mean-field scaling regime. We demonstrate how to tightly bound this approximation gap through a differential equation governed by the mean-field dynamics. A key factor influencing the growth of this ODE is the local Hessian of each particle, defined as the derivative of the particle's velocity in the mean-field dynamics with respect to its position. We apply our results to the canonical feature learning problem of estimating a well-specified single-index model; we permit the information exponent to be arbitrarily large, leading to convergence times that grow polynomially in the ambient dimension $d$. We show that, due to a certain ``self-concordance'' property in these problems -- where the local Hessian of a particle is bounded by a constant times the particle's velocity -- polynomially many neurons are sufficient to closely approximate the mean-field dynamics throughout training.
Authors: Lovedeep Gondara, Jonathan Simkin, Graham Sayle, Shebnum Devji, Gregory Arbour, Raymond Ng
Abstract: This study aims to guide language model selection by investigating: 1) the necessity of finetuning versus zero-shot usage, 2) the benefits of domain-adjacent versus generic pretrained models, 3) the value of further domain-specific pretraining, and 4) the continued relevance of Small Language Models (SLMs) compared to Large Language Models (LLMs) for specific tasks. Using electronic pathology reports from the British Columbia Cancer Registry (BCCR), three classification scenarios with varying difficulty and data size are evaluated. Models include various SLMs and an LLM. SLMs are evaluated both zero-shot and finetuned; the LLM is evaluated zero-shot only. Finetuning significantly improved SLM performance across all scenarios compared to their zero-shot results. The zero-shot LLM outperformed zero-shot SLMs but was consistently outperformed by finetuned SLMs. Domain-adjacent SLMs generally performed better than the generic SLM after finetuning, especially on harder tasks. Further domain-specific pretraining yielded modest gains on easier tasks but significant improvements on the complex, data-scarce task. The results highlight the critical role of finetuning for SLMs in specialized domains, enabling them to surpass zero-shot LLM performance on targeted classification tasks. Pretraining on domain-adjacent or domain-specific data provides further advantages, particularly for complex problems or limited finetuning data. While LLMs offer strong zero-shot capabilities, their performance on these specific tasks did not match that of appropriately finetuned SLMs. In the era of LLMs, SLMs remain relevant and effective, offering a potentially superior performance-resource trade-off compared to LLMs.
Authors: Junwon Seo, Kensuke Nakamura, Andrea Bajcsy
Abstract: Recent advances in generative world models have enabled classical safe control methods, such as Hamilton-Jacobi (HJ) reachability, to generalize to complex robotic systems operating directly from high-dimensional sensor observations. However, obtaining comprehensive coverage of all safety-critical scenarios during world model training is extremely challenging. As a result, latent safety filters built on top of these models may miss novel hazards and even fail to prevent known ones, overconfidently misclassifying risky out-of-distribution (OOD) situations as safe. To address this, we introduce an uncertainty-aware latent safety filter that proactively steers robots away from both known and unseen failures. Our key idea is to use the world model's epistemic uncertainty as a proxy for identifying unseen potential hazards. We propose a principled method to detect OOD world model predictions by calibrating an uncertainty threshold via conformal prediction. By performing reachability analysis in an augmented state space-spanning both the latent representation and the epistemic uncertainty-we synthesize a latent safety filter that can reliably safeguard arbitrary policies from both known and unseen safety hazards. In simulation and hardware experiments on vision-based control tasks with a Franka manipulator, we show that our uncertainty-aware safety filter preemptively detects potential unsafe scenarios and reliably proposes safe, in-distribution actions. Video results can be found on the project website at https://cmu-intentlab.github.io/UNISafe
Authors: Angelina Wang
Abstract: A key value proposition of machine learning is generalizability: the same methods and model architecture should be able to work across different domains and different contexts. While powerful, this generalization can sometimes go too far, and miss the importance of the specifics. In this work, we look at how fair machine learning has often treated as interchangeable the identity axis along which discrimination occurs. In other words, racism is measured and mitigated the same way as sexism, as ableism, as ageism. Disciplines outside of computer science have pointed out both the similarities and differences between these different forms of oppression, and in this work we draw out the implications for fair machine learning. While certainly not all aspects of fair machine learning need to be tailored to the specific form of oppression, there is a pressing need for greater attention to such specificity than is currently evident. Ultimately, context specificity can deepen our understanding of how to build more fair systems, widen our scope to include currently overlooked harms, and, almost paradoxically, also help to narrow our scope and counter the fear of an infinite number of group-specific methods of analysis.
Authors: Hongxin Xiang, Ke Li, Mingquan Liu, Zhixiang Cheng, Bin Yao, Wenjie Du, Jun Xia, Li Zeng, Xin Jin, Xiangxiang Zeng
Abstract: Existing molecular machine learning force fields (MLFFs) generally focus on the learning of atoms, molecules, and simple quantum chemical properties (such as energy and force), but ignore the importance of electron density (ED) $\rho(r)$ in accurately understanding molecular force fields (MFFs). ED describes the probability of finding electrons at specific locations around atoms or molecules, which uniquely determines all ground state properties (such as energy, molecular structure, etc.) of interactive multi-particle systems according to the Hohenberg-Kohn theorem. However, the calculation of ED relies on the time-consuming first-principles density functional theory (DFT) which leads to the lack of large-scale ED data and limits its application in MLFFs. In this paper, we introduce EDBench, a large-scale, high-quality dataset of ED designed to advance learning-based research at the electronic scale. Built upon the PCQM4Mv2, EDBench provides accurate ED data, covering 3.3 million molecules. To comprehensively evaluate the ability of models to understand and utilize electronic information, we design a suite of ED-centric benchmark tasks spanning prediction, retrieval, and generation. Our evaluation on several state-of-the-art methods demonstrates that learning from EDBench is not only feasible but also achieves high accuracy. Moreover, we show that learning-based method can efficiently calculate ED with comparable precision while significantly reducing the computational cost relative to traditional DFT calculations. All data and benchmarks from EDBench will be freely available, laying a robust foundation for ED-driven drug discovery and materials science.
Authors: Derrick Gilchrist Edward Manoharan, Anubha Goel, Alexandros Iosifidis, Henri Hansen, Juho Kanniainen
Abstract: The spreading dynamics in social networks are often studied under the assumption that individuals' statuses, whether informed or infected, are fully observable. However, in many real-world situations, such statuses remain unobservable, which is crucial for determining an individual's potential to further spread the infection. While final statuses are hidden, intermediate indicators such as symptoms of infection are observable and provide useful representations of the underlying diffusion process. We propose a partial observability-aware Machine Learning framework to learn the characteristics of the spreading model. We term the method Distribution Classification, which utilizes the power of classifiers to infer the underlying transmission dynamics. Through extensive benchmarking against Approximate Bayesian Computation and GNN-based baselines, our framework consistently outperforms these state-of-the-art methods, delivering accurate parameter estimates across diverse diffusion settings while scaling efficiently to large networks. We validate the method on synthetic networks and extend the study to a real-world insider trading network, demonstrating its effectiveness in analyzing spreading phenomena where direct observation of individual statuses is not possible.
Authors: Yuhang Zhou, Jing Zhu, Shengyi Qian, Zhuokai Zhao, Xiyao Wang, Xiaoyu Liu, Ming Li, Paiheng Xu, Wei Ai, Furong Huang
Abstract: Large Language Models (LLMs) are increasingly aligned with human preferences through Reinforcement Learning from Human Feedback (RLHF). Among RLHF methods, Group Relative Policy Optimization (GRPO) has gained attention for its simplicity and strong performance, notably eliminating the need for a learned value function. However, GRPO implicitly assumes a balanced domain distribution and uniform semantic alignment across groups, assumptions that rarely hold in real-world datasets. When applied to multi-domain, imbalanced data, GRPO disproportionately optimizes for dominant domains, neglecting underrepresented ones and resulting in poor generalization and fairness. We propose Domain-Informed Self-Consistency Policy Optimization (DISCO), a principled extension to GRPO that addresses inter-group imbalance with two key innovations. Domain-aware reward scaling counteracts frequency bias by reweighting optimization based on domain prevalence. Difficulty-aware reward scaling leverages prompt-level self-consistency to identify and prioritize uncertain prompts that offer greater learning value. Together, these strategies promote more equitable and effective policy learning across domains. Extensive experiments across multiple LLMs and skewed training distributions show that DISCO improves generalization, outperforms existing GRPO variants by 5% on Qwen3 models, and sets new state-of-the-art results on multi-domain alignment benchmarks. Our code and data are available at https://github.com/Tonyzhou98/disco_grpo.
Authors: Daniel Jarne Ornia, Nicholas Bishop, Joel Dyer, Wei-Chen Lee, Ani Calinescu, Doyne Farmer, Michael Wooldridge
Abstract: Advanced reasoning models with agentic capabilities (AI agents) are deployed to interact with humans and to solve sequential decision-making problems under (approximate) utility functions and internal models. When such problems have resource or failure constraints where action sequences may be forcibly terminated once resources are exhausted, agents face implicit trade-offs that reshape their utility-driven (rational) behaviour. Additionally, since these agents are typically commissioned by a human principal to act on their behalf, asymmetries in constraint exposure can give rise to previously unanticipated misalignment between human objectives and agent incentives. We formalise this setting through a survival bandit framework, provide theoretical and empirical results that quantify the impact of survival-driven preference shifts, identify conditions under which misalignment emerges and propose mechanisms to mitigate the emergence of risk-seeking or risk-averse behaviours. As a result, this work aims to increase understanding and interpretability of emergent behaviours of AI agents operating under such survival pressure, and offer guidelines for safely deploying such AI systems in critical resource-limited environments.
Authors: Hao Dong, Moru Liu, Jian Liang, Eleni Chatzi, Olga Fink
Abstract: Vision-Language Models (VLMs) have demonstrated strong capabilities in aligning visual and textual modalities, enabling a wide range of applications in multimodal understanding and generation. While they excel in zero-shot and transfer learning scenarios, VLMs remain susceptible to misclassification, often yielding confident yet incorrect predictions. This limitation poses a significant risk in safety-critical domains, where erroneous predictions can lead to severe consequences. In this work, we introduce TrustVLM, a training-free framework designed to address the critical challenge of estimating when VLM's predictions can be trusted. Motivated by the observed modality gap in VLMs and the insight that certain concepts are more distinctly represented in the image embedding space, we propose a novel confidence-scoring function that leverages this space to improve misclassification detection. We rigorously evaluate our approach across 17 diverse datasets, employing 4 architectures and 2 VLMs, and demonstrate state-of-the-art performance, with improvements of up to 51.87% in AURC, 9.14% in AUROC, and 32.42% in FPR95 compared to existing baselines. By improving the reliability of the model without requiring retraining, TrustVLM paves the way for safer deployment of VLMs in real-world applications. The code is available at https://github.com/EPFL-IMOS/TrustVLM.
Authors: Luigi Sigillo, Shengfeng He, Danilo Comminiello
Abstract: High-resolution image synthesis remains a core challenge in generative modeling, particularly in balancing computational efficiency with the preservation of fine-grained visual detail. We present Latent Wavelet Diffusion (LWD), a lightweight training framework that significantly improves detail and texture fidelity in ultra-high-resolution (2K-4K) image synthesis. LWD introduces a novel, frequency-aware masking strategy derived from wavelet energy maps, which dynamically focuses the training process on detail-rich regions of the latent space. This is complemented by a scale-consistent VAE objective to ensure high spectral fidelity. The primary advantage of our approach is its efficiency: LWD requires no architectural modifications and adds zero additional cost during inference, making it a practical solution for scaling existing models. Across multiple strong baselines, LWD consistently improves perceptual quality and FID scores, demonstrating the power of signal-driven supervision as a principled and efficient path toward high-resolution generative modeling.
Authors: Anju Chhetri, Jari Korhonen, Prashnna Gyawali, Binod Bhattarai
Abstract: Ensuring reliability is paramount in deep learning, particularly within the domain of medical imaging, where diagnostic decisions often hinge on model outputs. The capacity to separate out-of-distribution (OOD) samples has proven to be a valuable indicator of a model's reliability in research. In medical imaging, this is especially critical, as identifying OOD inputs can help flag potential anomalies that might otherwise go undetected. While many OOD detection methods rely on feature or logit space representations, recent works suggest these approaches may not fully capture OOD diversity. To address this, we propose a novel OOD scoring mechanism, called NERO, that leverages neuron-level relevance at the feature layer. Specifically, we cluster neuron-level relevance for each in-distribution (ID) class to form representative centroids and introduce a relevance distance metric to quantify a new sample's deviation from these centroids, enhancing OOD separability. Additionally, we refine performance by incorporating scaled relevance in the bias term and combining feature norms. Our framework also enables explainable OOD detection. We validate its effectiveness across multiple deep learning architectures on the gastrointestinal imaging benchmarks Kvasir and GastroVision, achieving improvements over state-of-the-art OOD detection methods.
Authors: Christopher Agia, Rohan Sinha, Jingyun Yang, Rika Antonova, Marco Pavone, Haruki Nishimura, Masha Itkina, Jeannette Bohg
Abstract: In robot imitation learning, policy performance is tightly coupled with the quality and composition of the demonstration data. Yet, developing a precise understanding of how individual demonstrations contribute to downstream outcomes - such as closed-loop task success or failure - remains a persistent challenge. We propose CUPID, a robot data curation method based on a novel influence function-theoretic formulation for imitation learning policies. Given a set of evaluation rollouts, CUPID estimates the influence of each training demonstration on the policy's expected return. This enables ranking and selection of demonstrations according to their impact on the policy's closed-loop performance. We use CUPID to curate data by 1) filtering out training demonstrations that harm policy performance and 2) subselecting newly collected trajectories that will most improve the policy. Extensive simulated and hardware experiments show that our approach consistently identifies which data drives test-time performance. For example, training with less than 33% of curated data can yield state-of-the-art diffusion policies on the simulated RoboMimic benchmark, with similar gains observed in hardware. Furthermore, hardware experiments show that our method can identify robust strategies under distribution shift, isolate spurious correlations, and even enhance the post-training of generalist robot policies. Videos and code are made available at: https://cupid-curation.github.io.
Authors: Nuoye Xiong, Anqi Dong, Ning Wang, Cong Hua, Guangming Zhu, Lin Mei, Peiyi Shen, Liang Zhang
Abstract: Recent advances in deep learning have led to increasingly complex models with deeper layers and more parameters, reducing interpretability and making their decisions harder to understand. While many methods explain black-box reasoning, most lack effective interventions or only operate at sample-level without modifying the model itself. To address this, we propose the Concept Bottleneck Model for Enhancing Human-Neural Network Mutual Understanding (CBM-HNMU). CBM-HNMU leverages the Concept Bottleneck Model (CBM) as an interpretable framework to approximate black-box reasoning and communicate conceptual understanding. Detrimental concepts are automatically identified and refined (removed/replaced) based on global gradient contributions. The modified CBM then distills corrected knowledge back into the black-box model, enhancing both interpretability and accuracy. We evaluate CBM-HNMU on various CNN and transformer-based models across Flower-102, CIFAR-10, CIFAR-100, FGVC-Aircraft, and CUB-200, achieving a maximum accuracy improvement of 2.64% and a maximum increase in average accuracy across 1.03%. Source code is available at: https://github.com/XiGuaBo/CBM-HNMU.
Authors: Keshav Kumar
Abstract: Large Language Models (LLMs) have demonstrated impressive generative capabilities across diverse tasks but remain susceptible to hallucinations, confidently generated yet factually incorrect outputs. We introduce a reference-free, token-level hallucination detection framework that leverages the variance in token log-probabilities across multiple stochastic generations. Unlike prior methods that require ground-truth references or sentence-level verification, our approach is model-agnostic, interpretable, and suited for real-time or post-hoc analysis. We evaluate our method on unanswerable question prompts from the SQuAD v2 dataset and benchmark across three autoregressive models of varying scales: GPT-Neo 125M, Falcon 1B, and Mistral 7B. Through both quantitative metrics and visual diagnostics, we show that token-level variance reliably highlights instability in model outputs and correlates with hallucination patterns. Our framework is lightweight, reproducible, and adaptable to multiple domains, offering a valuable diagnostic tool for analyzing generative reliability in LLMs.
Authors: Roberto Miele, Niklas Linde
Abstract: Diffusion models offer stable training and state-of-the-art performance for deep generative modeling tasks. Here, we consider their use in the context of multivariate subsurface modeling and probabilistic inversion. We first demonstrate that diffusion models enhance multivariate modeling capabilities compared to variational autoencoders and generative adversarial networks. In diffusion modeling, the generative process involves a comparatively large number of time steps with update rules that can be modified to account for conditioning data. We propose different corrections to the popular Diffusion Posterior Sampling approach by Chung et al. (2023). In particular, we introduce a likelihood approximation accounting for the noise-contamination that is inherent in diffusion modeling. We assess performance in a multivariate geological scenario involving facies and correlated acoustic impedance. Conditional modeling is demonstrated using both local hard data (well logs) and nonlinear geophysics (fullstack seismic data). Our tests show significantly improved statistical robustness, enhanced sampling of the posterior probability density function and reduced computational costs, compared to the original approach. The method can be used with both hard and indirect conditioning data, individually or simultaneously. As the inversion is included within the diffusion process, it is faster than other methods requiring an outer-loop around the generative model, such as Markov chain Monte Carlo.
Authors: Anuroop Sriram, Logan M. Brabson, Xiaohan Yu, Sihoon Choi, Kareem Abdelmaqsoud, Elias Moubarak, Pim de Haan, Sindy L\"owe, Johann Brehmer, John R. Kitchin, Max Welling, C. Lawrence Zitnick, Zachary Ulissi, Andrew J. Medford, David S. Sholl
Abstract: Identifying useful sorbent materials for direct air capture (DAC) from humid air remains a challenge. We present the Open DAC 2025 (ODAC25) dataset, a significant expansion and improvement upon ODAC23 (Sriram et al., ACS Central Science, 10 (2024) 923), comprising nearly 60 million DFT single-point calculations for CO$_2$, H$_2$O, N$_2$, and O$_2$ adsorption in 15,000 MOFs. ODAC25 introduces chemical and configurational diversity through functionalized MOFs, high-energy GCMC-derived placements, and synthetically generated frameworks. ODAC25 also significantly improves upon the accuracy of DFT calculations and the treatment of flexible MOFs in ODAC23. Along with the dataset, we release new state-of-the-art machine-learned interatomic potentials trained on ODAC25 and evaluate them on adsorption energy and Henry's law coefficient predictions.
Authors: Abhigya Verma, Sriram Puttagunta, Seganrasan Subramanian, Sravan Ramachandran
Abstract: GRAFT is a structured multimodal benchmark for evaluating models on instruction-following, visual reasoning, and visual-textual alignment tasks. It features programmatically generated charts and synthetically rendered tables, created with Python visualization libraries to ensure control over data semantics, structure, and clarity. Each GRAFT instance pairs a chart or table image with a systematically generated, multi-step analytical question based solely on visual content. Answers are provided in structured formats such as JSON or YAML, supporting consistent evaluation of both reasoning and output format. The benchmark introduces a taxonomy of reasoning types including comparison, trend identification, ranking, aggregation, proportion estimation, and anomaly detection to enable comprehensive assessment. Reference answers follow strict factual and formatting guidelines for precise, aspect-based evaluation. GRAFT offers a unified, scalable framework for fine-grained benchmarking of multimodal models on visually grounded, structured reasoning tasks, setting a new evaluation standard in this field.
Authors: Karan Shah, Attila Cangi
Abstract: Time-dependent density functional theory (TDDFT) is a widely used method to investigate electron dynamics under external time-dependent perturbations such as laser fields. In this work, we present a machine learning approach to accelerate electron dynamics simulations based on real time TDDFT using autoregressive neural operators as time-propagators for the electron density. By leveraging physics-informed constraints and featurization, and high-resolution training data, our model achieves superior accuracy and computational speed compared to traditional numerical solvers. We demonstrate the effectiveness of our model on a class of one-dimensional diatomic molecules under the influence of a range of laser parameters. This method has potential in enabling on-the-fly modeling of laser-irradiated molecules and materials by utilizing fast machine learning predictions in a large space of varying experimental parameters of the laser.
Authors: Srinivasan Arunachalam, Arkopal Dutt, Alexandru Gheorghiu, Michael de Oliveira
Abstract: We initiate the study of quantum agnostic learning of phase states with respect to a function class $\mathsf{C}\subseteq \{c:\{0,1\}^n\rightarrow \{0,1\}\}$: given copies of an unknown $n$-qubit state $|\psi\rangle$ which has fidelity $\textsf{opt}$ with a phase state $|\phi_c\rangle=\frac{1}{\sqrt{2^n}}\sum_{x\in \{0,1\}^n}(-1)^{c(x)}|x\rangle$ for some $c\in \mathsf{C}$, output $|\phi\rangle$ which has fidelity $|\langle \phi | \psi \rangle|^2 \geq \textsf{opt}-\varepsilon$. To this end, we give agnostic learning protocols for the following classes: (i) Size-$t$ decision trees which runs in time $\textsf{poly}(n,t,1/\varepsilon)$. This also implies $k$-juntas can be agnostically learned in time $\textsf{poly}(n,2^k,1/\varepsilon)$. (ii) $s$-term DNF formulas in time $\textsf{poly}(n,(s/\varepsilon)^{\log \log (s/\varepsilon) \cdot \log(1/\varepsilon)})$. Our main technical contribution is a quantum agnostic boosting protocol which converts a weak agnostic learner, which outputs a parity state $|\phi\rangle$ such that $|\langle \phi|\psi\rangle|^2\geq \textsf{opt}/\textsf{poly}(n)$, into a strong learner which outputs a superposition of parity states $|\phi'\rangle$ such that $|\langle \phi'|\psi\rangle|^2\geq \textsf{opt} - \varepsilon$. Using quantum agnostic boosting, we obtain a $n^{O(\log \log n \cdot \log(1/\varepsilon))}$-time algorithm for learning $\textsf{poly}(n)$-sized depth-$3$ circuits (consisting of $\textsf{AND}$, $\textsf{OR}$, $\textsf{NOT}$ gates) in the uniform $\textsf{PAC}$ model given quantum examples, which is near-polynomial time for constant $\varepsilon$. Classically, obtaining an algorithm with a similar complexity has been an open question in the $\textsf{PAC}$ model and our work answers this given quantum examples.
Authors: Julian Konrad, Janina Mittelhaus, David M. Wilkins, Bodo Fiedler, Robert Mei{\ss}ner
Abstract: The vibrational response of polymer networks under load provides a sensitive probe of molecular deformation and a route to non-destructive diagnostics. Here we show that machine-learned force fields reproduce these spectroscopic fingerprints with quantum-level fidelity in realistic epoxy thermosets. Using MACE-OFF23 molecular dynamics, we capture the experimentally observed redshifts of para-phenylene stretching modes under tensile load, in contrast to the harmonic OPLS-AA model. These shifts correlate with molecular elongation and alignment, consistent with Badger's rule, directly linking vibrational features to local stress. To capture IR intensities, we trained a symmetry-adapted dipole moment model on representative epoxy fragments, enabling validation of strain responses. Together, these approaches provide chemically accurate and computationally accessible predictions of strain-dependent vibrational spectra. Our results establish vibrational fingerprints as predictive markers of mechanical state in polymer networks, pointing to new strategies for stress mapping and structural-health diagnostics in advanced materials.
Authors: Kai Jiang, Zhengyan Shi, Dell Zhang, Hongyuan Zhang, Xuelong Li
Abstract: Class Incremental Learning (CIL) aims to continuously learn new categories while retaining the knowledge of old ones. Pre-trained models (PTMs) show promising capabilities in CIL. However, existing approaches that apply lightweight fine-tuning to backbones still induce parameter drift, thereby compromising the generalization capability of pre-trained models. Parameter drift can be conceptualized as a form of noise that obscures critical patterns learned for previous tasks. However, recent researches have shown that noise is not always harmful. For example, the large number of visual patterns learned from pre-training can be easily abused by a single task, and introducing appropriate noise can suppress some low-correlation features, thus leaving a margin for future tasks. To this end, we propose learning beneficial noise for CIL guided by information theory and propose Mixture of Noise (Min), aiming to mitigate the degradation of backbone generalization from adapting new tasks. Specifically, task-specific noise is learned from high-dimension features of new tasks. Then, a set of weights is adjusted dynamically for optimal mixture of different task noise. Finally, Min embeds the beneficial noise into the intermediate features to mask the response of inefficient patterns. Extensive experiments on six benchmark datasets demonstrate that Min achieves state-of-the-art performance in most incremental settings, with particularly outstanding results in 50-steps incremental settings. This shows the significant potential for beneficial noise in continual learning. Code is available at https://github.com/ASCIIJK/MiN-NeurIPS2025.
Authors: Yutong Li, Yu Zhu, Yichen Qiao, Ziyu Guan, Lv Shao, Tong Liu, Bo Zheng
Abstract: The pre-ranking stage in industrial recommendation systems faces a fundamental conflict between efficiency and effectiveness. While powerful models like Target Attention (TA) excel at capturing complex feature interactions in the ranking stage, their high computational cost makes them infeasible for pre-ranking, which often relies on simplistic vector-product models. This disparity creates a significant performance bottleneck for the entire system. To bridge this gap, we propose TARQ, a novel pre-ranking framework. Inspired by generative models, TARQ's key innovation is to equip pre-ranking with an architecture approximate to TA by Residual Quantization. This allows us to bring the modeling power of TA into the latency-critical pre-ranking stage for the first time, establishing a new state-of-the-art trade-off between accuracy and efficiency. Extensive offline experiments and large-scale online A/B tests at Taobao demonstrate TARQ's significant improvements in ranking performance. Consequently, our model has been fully deployed in production, serving tens of millions of daily active users and yielding substantial business improvements.
Authors: Juyeon Yoon, Somin Kim, Robert Feldt, Shin Yoo
Abstract: Software increasingly relies on the emergent capabilities of Large Language Models (LLMs), from natural language understanding to program analysis and generation. Yet testing them on specific tasks remains difficult and costly: many prompts lack ground truth, forcing reliance on human judgment, while existing uncertainty and adequacy measures typically require full inference. A key challenge is to assess input adequacy in a way that reflects the demands of the task, ideally before even generating any output. We introduce CLOTHO, a task-specific, pre-generation adequacy measure that estimates input difficulty directly from hidden LLM states. Given a large pool of unlabelled inputs for a specific task, CLOTHO uses a Gaussian Mixture Model (GMM) to adaptively sample the most informative cases for human labelling. Based on this reference set the GMM can then rank unseen inputs by their likelihood of failure. In our empirical evaluation across eight benchmark tasks and three open-weight LLMs, CLOTHO can predict failures with a ROC-AUC of 0.716, after labelling reference sets that are on average only 5.4% of inputs. It does so without generating any outputs, thereby reducing costs compared to existing uncertainty measures. Comparison of CLOTHO and post-generation uncertainty measures shows that the two approaches complement each other. Crucially, we show that adequacy scores learnt from open-weight LLMs transfer effectively to proprietary models, extending the applicability of the approach. When prioritising test inputs for proprietary models, CLOTHO increases the average number of failing inputs from 18.7 to 42.5 out of 100, compared to random prioritisation.
Authors: Zhuofan Chen, Jiyuan He, Yichi Zhang, Xing Hu, Haoxing Wen, Jun Bai, Wenge Rong
Abstract: Mathematical reasoning poses significant challenges for Large Language Models (LLMs) due to its demand for multi-step reasoning and abstract conceptual integration. While recent test-time scaling techniques rely heavily on high-quality, challenging problems, the scarcity of Olympiad-level math problems remains a bottleneck. We introduce CogAtom, a novel cognitive atom-based framework for synthesizing mathematically rigorous and cognitively diverse problems. Unlike prior approaches, CogAtom models problem construction as a process of selecting and recombining fundamental reasoning units, cognitive atoms, extracted from human-authored solutions. A diversity-promoting random walk algorithm enables exploration of the cognitive atom space, while a constraint-based recombination mechanism ensures logical soundness and structural validity. The combinatorial nature of the graph structure provides a near-infinite space of reasoning paths, and the walk algorithm systematically explores this space to achieve large-scale synthesis of high-quality problems; meanwhile, by controlling the number of cognitive atoms, we can precisely adjust problem difficulty, ensuring diversity, scalability, and controllability of the generated problems. Experimental results demonstrate that CogAtom outperforms existing methods in accuracy, reasoning depth, and diversity, generating problems that closely match the difficulty of AIME while exceeding it in structural variation. Our work offers a cognitively grounded pathway toward scalable, high-quality math problem generation.Our code is publicly available at https://github.com/Icarus-1111/CogAtom.
Authors: Yuanrong Wang, Yingpeng Du
Abstract: Recommender systems (RS) greatly influence users' consumption decisions, making them attractive targets for malicious shilling attacks that inject fake user profiles to manipulate recommendations. Existing shilling methods can generate effective and stealthy fake profiles when training data only contain rating matrix, but they lack comprehensive solutions for scenarios where side features are present and utilized by the recommender. To address this gap, we extend the Leg-UP framework by enhancing the generator architecture to incorporate side features, enabling the generation of side-feature-aware fake user profiles. Experiments on benchmarks show that our method achieves strong attack performance while maintaining stealthiness.
Authors: Akwasi Asare, Ulas Bagci
Abstract: Colorectal cancer (CRC) remains one of the leading causes of cancer-related morbidity and mortality worldwide, with gastrointestinal (GI) polyps serving as critical precursors according to the World Health Organization (WHO). Early and accurate segmentation of polyps during colonoscopy is essential for reducing CRC progression, yet manual delineation is labor-intensive and prone to observer variability. Deep learning methods have demonstrated strong potential for automated polyp analysis, but their limited interpretability remains a barrier to clinical adoption. In this study, we present PolypSeg-GradCAM, an explainable deep learning framework that integrates the U-Net architecture with Gradient-weighted Class Activation Mapping (Grad-CAM) for transparent polyp segmentation. The model was trained and evaluated on the Kvasir-SEG dataset of 1000 annotated endoscopic images. Experimental results demonstrate robust segmentation performance, achieving a mean Intersection over Union (IoU) of 0.9257 on the test set and consistently high Dice coefficients (F-score > 0.96) on training and validation sets. Grad-CAM visualizations further confirmed that predictions were guided by clinically relevant regions, enhancing transparency and trust in the model's decisions. By coupling high segmentation accuracy with interpretability, PolypSeg-GradCAM represents a step toward reliable, trustworthy AI-assisted colonoscopy and improved early colorectal cancer prevention.
Authors: Natasha Butt, Ariel Kwiatkowski, Ismail Labiad, Julia Kempe, Yann Ollivier
Abstract: The use of continuous instead of discrete tokens during the Chain-of-Thought (CoT) phase of reasoning LLMs has garnered attention recently, based on the intuition that a continuous mixture of discrete tokens could simulate a superposition of several reasoning paths simultaneously. Theoretical results have formally proven that continuous tokens have much greater expressivity and can solve specific problems more efficiently. However, practical use of continuous tokens has been limited by strong training difficulties: previous works either just use continuous tokens at inference time on a pre-trained discrete-token model, or must distill the continuous CoT from ground-truth discrete CoTs and face computational costs that limit the CoT to very few tokens. This is the first work introducing a scalable method to learn continuous CoTs via reinforcement learning (RL), without distilling from reference discrete CoTs. We use "soft" tokens: mixtures of tokens together with noise on the input embedding to provide RL exploration. Computational overhead is minimal, enabling us to learn continuous CoTs with hundreds of tokens. On math reasoning benchmarks with Llama and Qwen models up to 8B, training with continuous CoTs match discrete-token CoTs for pass@1 and surpass them for pass@32, showing greater CoT diversity. In systematic comparisons, the best-performing scenario is to train with continuous CoT tokens then use discrete tokens for inference, meaning the "soft" models can be deployed in a standard way. Finally, we show continuous CoT RL training better preserves the predictions of the base model on out-of-domain tasks, thus providing a softer touch to the base model.
Authors: Georgii Kolokolnikov, Marie-Lena Schmalhofer, Sophie Goetz, Lennart Well, Said Farschtschi, Victor-Felix Mautner, Inka Ristow, Rene Werner
Abstract: Background and Objectives: Neurofibromatosis type 1 is a genetic disorder characterized by the development of numerous neurofibromas (NFs) throughout the body. Whole-body MRI (WB-MRI) is the clinical standard for detection and longitudinal surveillance of NF tumor growth. Existing interactive segmentation methods fail to combine high lesion-wise precision with scalability to hundreds of lesions. This study proposes a novel interactive segmentation model tailored to this challenge. Methods: We introduce MOIS-SAM2, a multi-object interactive segmentation model that extends the state-of-the-art, transformer-based, promptable Segment Anything Model 2 (SAM2) with exemplar-based semantic propagation. MOIS-SAM2 was trained and evaluated on 119 WB-MRI scans from 84 NF1 patients acquired using T2-weighted fat-suppressed sequences. The dataset was split at the patient level into a training set and four test sets (one in-domain and three reflecting different domain shift scenarios, e.g., MRI field strength variation, low tumor burden, differences in clinical site and scanner vendor). Results: On the in-domain test set, MOIS-SAM2 achieved a scan-wise DSC of 0.60 against expert manual annotations, outperforming baseline 3D nnU-Net (DSC: 0.54) and SAM2 (DSC: 0.35). Performance of the proposed model was maintained under MRI field strength shift (DSC: 0.53) and scanner vendor variation (DSC: 0.50), and improved in low tumor burden cases (DSC: 0.61). Lesion detection F1 scores ranged from 0.62 to 0.78 across test sets. Preliminary inter-reader variability analysis showed model-to-expert agreement (DSC: 0.62-0.68), comparable to inter-expert agreement (DSC: 0.57-0.69). Conclusions: The proposed MOIS-SAM2 enables efficient and scalable interactive segmentation of NFs in WB-MRI with minimal user input and strong generalization, supporting integration into clinical workflows.