Authors: Daniel Groos
Abstract: Fantasy Premier League engages the football community in selecting the Premier League players who will perform best from gameweek to gameweek. Access to accurate performance forecasts gives participants an edge over competitors by guiding expectations about player outcomes and reducing uncertainty in squad selection. However, high-accuracy forecasts are currently limited to commercial services whose inner workings are undisclosed and that rely on proprietary data. This paper aims to democratize access to highly accurate forecasts of player performance by presenting OpenFPL, an open-source Fantasy Premier League forecasting method developed exclusively from public data. Comprising position-specific ensemble models optimized on Fantasy Premier League and Understat data from four previous seasons (2020-21 to 2023-24), OpenFPL achieves accuracy comparable to a leading commercial service when tested prospectively on data from the 2024-25 season. OpenFPL also surpasses the commercial benchmark for high-return players ($>$ 2 points), which are most influential for rank gains. These findings hold across one-, two-, and three-gameweek forecast horizons, supporting long-term planning of transfers and strategies while also informing final-day decisions.
Authors: Daniel Beaglehole, David Holzm\"uller, Adityanarayanan Radhakrishnan, Mikhail Belkin
Abstract: Inference from tabular data, collections of continuous and categorical variables organized into matrices, is a foundation for modern technology and science. Yet, in contrast to the explosive changes in the rest of AI, the best practice for these predictive tasks has been relatively unchanged and is still primarily based on variations of Gradient Boosted Decision Trees (GBDTs). Very recently, there has been renewed interest in developing state-of-the-art methods for tabular data based on recent developments in neural networks and feature learning methods. In this work, we introduce xRFM, an algorithm that combines feature learning kernel machines with a tree structure to both adapt to the local structure of the data and scale to essentially unlimited amounts of training data. We show that compared to $31$ other methods, including recently introduced tabular foundation models (TabPFNv2) and GBDTs, xRFM achieves best performance across $100$ regression datasets and is competitive to the best methods across $200$ classification datasets outperforming GBDTs. Additionally, xRFM provides interpretability natively through the Average Gradient Outer Product.
Authors: Amy Armento Lee, Narayan Hegde, Nina Deliu, Emily Rosenzweig, Arun Suggala, Sriram Lakshminarasimhan, Qian He, John Hernandez, Martin Seneviratne, Rahul Singh, Pradnesh Kalkar, Karthikeyan Shanmugam, Aravindan Raghuveer, Abhimanyu Singh, My Nguyen, James Taylor, Jatin Alla, Sofia S. Villar, Hulya Emir-Farinas
Abstract: Consistent physical inactivity poses a major global health challenge. Mobile health (mHealth) interventions, particularly Just-in-Time Adaptive Interventions (JITAIs), offer a promising avenue for scalable, personalized physical activity (PA) promotion. However, developing and evaluating such interventions at scale, while integrating robust behavioral science, presents methodological hurdles. The PEARL study was the first large-scale, four-arm randomized controlled trial to assess a reinforcement learning (RL) algorithm, informed by health behavior change theory, to personalize the content and timing of PA nudges via a Fitbit app. We enrolled and randomized 13,463 Fitbit users into four study arms: control, random, fixed, and RL. The control arm received no nudges. The other three arms received nudges from a bank of 155 nudges based on behavioral science principles. The random arm received nudges selected at random. The fixed arm received nudges based on a pre-set logic from survey responses about PA barriers. The RL group received nudges selected by an adaptive RL algorithm. We included 7,711 participants in primary analyses (mean age 42.1, 86.3% female, baseline steps 5,618.2). We observed an increase in PA for the RL group compared to all other groups from baseline to 1 and 2 months. The RL group had significantly increased average daily step count at 1 month compared to all other groups: control (+296 steps, p=0.0002), random (+218 steps, p=0.005), and fixed (+238 steps, p=0.002). At 2 months, the RL group sustained a significant increase compared to the control group (+210 steps, p=0.0122). Generalized estimating equation models also revealed a sustained increase in daily steps in the RL group vs. control (+208 steps, p=0.002). These findings demonstrate the potential of a scalable, behaviorally-informed RL approach to personalize digital health interventions for PA.
Authors: Steven Klee, Yuntian Xia
Abstract: Time series forecasting is a critical first step in generating demand plans for supply chains. Experiments on time series models typically focus on demonstrating improvements in forecast accuracy over existing/baseline solutions, quantified according to some accuracy metric. There is no doubt that forecast accuracy is important; however in production systems, demand planners often value consistency and stability over incremental accuracy improvements. Assuming that the inputs have not changed significantly, forecasts that vary drastically from one planning cycle to the next require high amounts of human intervention, which frustrates demand planners and can even cause them to lose trust in ML forecasting models. We study model-induced stochasticity, which quantifies the variance of a set of forecasts produced by a single model when the set of inputs is fixed. Models with lower variance are more stable. Recently the forecasting community has seen significant advances in forecast accuracy through the development of deep machine learning models for time series forecasting. We perform a case study measuring the stability and accuracy of state-of-the-art forecasting models (Chronos, DeepAR, PatchTST, Temporal Fusion Transformer, TiDE, and the AutoGluon best quality ensemble) on public data sets from the M5 competition and Favorita grocery sales. We show that ensemble models improve stability without significantly deteriorating (or even improving) forecast accuracy. While these results may not be surprising, the main point of this paper is to propose the need for further study of forecast stability for models that are being deployed in production systems.
Authors: Niels M\"undler, Jasper Dekoninck, Martin Vechev
Abstract: Large language models (LLMs) have shown promising performance across diverse domains. Many practical applications of LLMs, such as code completion and structured data extraction, require adherence to syntactic constraints specified by a formal language. Yet, due to their probabilistic nature, LLM output is not guaranteed to adhere to such formal languages. Prior work has proposed constrained decoding as a means to restrict LLM generation to particular formal languages. However, existing works are not applicable to the emerging paradigm of diffusion LLMs, when used in practical scenarios such as the generation of formally correct C++ or JSON output. In this paper we address this challenge and present the first constrained decoding method for diffusion models, one that can handle formal languages captured by context-free grammars. We begin by reducing constrained decoding to the more general additive infilling problem, which asks whether a partial output can be completed to a valid word in the target language. This problem also naturally subsumes the previously unaddressed multi-region infilling constrained decoding. We then reduce this problem to the task of deciding whether the intersection of the target language and a regular language is empty and present an efficient algorithm to solve it for context-free languages. Empirical results on various applications, such as C++ code infilling and structured data extraction in JSON, demonstrate that our method achieves near-perfect syntactic correctness while consistently preserving or improving functional correctness. Importantly, our efficiency optimizations ensure that the computational overhead remains practical.
Authors: Sola Shirai, Kavitha Srinivas, Julian Dolby, Michael Katz, Horst Samulowitz, Shirin Sohrabi
Abstract: For large language models (LLMs), reasoning over graphs could help solve many problems. Prior work has tried to improve LLM graph reasoning by examining how best to serialize graphs as text and by combining GNNs and LLMs. However, the merits of such approaches remain unclear, so we empirically answer the following research questions: (1) Can LLMs learn to solve fundamental graph tasks without specialized graph encoding models?, (2) Can LLMs generalize learned solutions to unseen graph structures or tasks?, and (3) What are the merits of competing approaches to learn graph tasks? We show that even small LLMs can learn to solve graph tasks by training them with instructive chain-of-thought solutions, and this training generalizes, without specialized graph encoders, to new tasks and graph structures.
Authors: Ke Niu, Haiyang Yu, Zhuofan Chen, Mengyang Zhao, Teng Fu, Bin Li, Xiangyang Xue
Abstract: Computer-Aided Design (CAD) plays a vital role in engineering and manufacturing, yet current CAD workflows require extensive domain expertise and manual modeling effort. Recent advances in large language models (LLMs) have made it possible to generate code from natural language, opening new opportunities for automating parametric 3D modeling. However, directly translating human design intent into executable CAD code remains highly challenging, due to the need for logical reasoning, syntactic correctness, and numerical precision. In this work, we propose CAD-RL, a multimodal Chain-of-Thought (CoT) guided reinforcement learning post training framework for CAD modeling code generation. Our method combines CoT-based Cold Start with goal-driven reinforcement learning post training using three task-specific rewards: executability reward, geometric accuracy reward, and external evaluation reward. To ensure stable policy learning under sparse and high-variance reward conditions, we introduce three targeted optimization strategies: Trust Region Stretch for improved exploration, Precision Token Loss for enhanced dimensions parameter accuracy, and Overlong Filtering to reduce noisy supervision. To support training and benchmarking, we release ExeCAD, a noval dataset comprising 16,540 real-world CAD examples with paired natural language and structured design language descriptions, executable CADQuery scripts, and rendered 3D models. Experiments demonstrate that CAD-RL achieves significant improvements in reasoning quality, output precision, and code executability over existing VLMs.
Authors: Maxime Heuillet, Yufei Cui, Boxing Chen, Audrey Durand, Prasanna Parthasarathi
Abstract: Advanced reasoning in LLMs on challenging domains like mathematical reasoning can be tackled using verifiable rewards based reinforced fine-tuning (ReFT). In standard ReFT frameworks, a behavior model generates multiple completions with answers per problem, for the answer to be then scored by a reward function. While such RL post-training methods demonstrate significant performance improvements across challenging reasoning domains, the computational cost of generating completions during training with multiple inference steps makes the training cost non-trivial. To address this, we draw inspiration from off-policy RL, and speculative decoding to introduce a novel ReFT framework, dubbed Nested-ReFT, where a subset of layers of the target model acts as the behavior model to generate off-policy completions during training. The behavior model configured with dynamic layer skipping per batch during training decreases the inference cost compared to the standard ReFT frameworks. Our theoretical analysis shows that Nested-ReFT yields unbiased gradient estimates with controlled variance. Our empirical analysis demonstrates improved computational efficiency measured as tokens/sec across multiple math reasoning benchmarks and model sizes. Additionally, we explore three variants of bias mitigation to minimize the off-policyness in the gradient updates that allows for maintaining performance that matches the baseline ReFT performance.
Authors: Yuhan Xie, William Cappelletti, Mahsa Shoaran, Pascal Frossard
Abstract: Deep neural networks for time series must capture complex temporal patterns, to effectively represent dynamic data. Self- and semi-supervised learning methods show promising results in pre-training large models, which -- when finetuned for classification -- often outperform their counterparts trained from scratch. Still, the choice of pretext training tasks is often heuristic and their transferability to downstream classification is not granted, thus we propose a novel semi-supervised pre-training strategy to enforce latent representations that satisfy the Neural Collapse phenomenon observed in optimally trained neural classifiers. We use a rotational equiangular tight frame-classifier and pseudo-labeling to pre-train deep encoders with few labeled samples. Furthermore, to effectively capture temporal dynamics while enforcing embedding separability, we integrate generative pretext tasks with our method, and we define a novel sequential augmentation strategy. We show that our method significantly outperforms previous pretext tasks when applied to LSTMs, transformers, and state-space models on three multivariate time series classification datasets. These results highlight the benefit of aligning pre-training objectives with theoretically grounded embedding geometry.
Authors: Maria Stoica, Francesco Leofante, Alessio Lomuscio
Abstract: Accurate and explainable out-of-distribution (OOD) detection is required to use machine learning systems safely. Previous work has shown that feature distance to decision boundaries can be used to identify OOD data effectively. In this paper, we build on this intuition and propose a post-hoc OOD detection method that, given an input, calculates the distance to decision boundaries by leveraging counterfactual explanations. Since computing explanations can be expensive for large architectures, we also propose strategies to improve scalability by computing counterfactuals directly in embedding space. Crucially, as the method employs counterfactual explanations, we can seamlessly use them to help interpret the results of our detector. We show that our method is in line with the state of the art on CIFAR-10, achieving 93.50% AUROC and 25.80% FPR95. Our method outperforms these methods on CIFAR-100 with 97.05% AUROC and 13.79% FPR95 and on ImageNet-200 with 92.55% AUROC and 33.55% FPR95 across four OOD datasets
Authors: Zhankun Luo, Abolfazl Hashemi
Abstract: Mixture models have attracted significant attention due to practical effectiveness and comprehensive theoretical foundations. A persisting challenge is model misspecification, which occurs when the model to be fitted has more mixture components than those in the data distribution. In this paper, we develop a theoretical understanding of the Expectation-Maximization (EM) algorithm's behavior in the context of targeted model misspecification for overspecified two-component Mixed Linear Regression (2MLR) with unknown $d$-dimensional regression parameters and mixing weights. In Theorem 5.1 at the population level, with an unbalanced initial guess for mixing weights, we establish linear convergence of regression parameters in $O(\log(1/\epsilon))$ steps. Conversely, with a balanced initial guess for mixing weights, we observe sublinear convergence in $O(\epsilon^{-2})$ steps to achieve the $\epsilon$-accuracy at Euclidean distance. In Theorem 6.1 at the finite-sample level, for mixtures with sufficiently unbalanced fixed mixing weights, we demonstrate a statistical accuracy of $O((d/n)^{1/2})$, whereas for those with sufficiently balanced fixed mixing weights, the accuracy is $O((d/n)^{1/4})$ given $n$ data samples. Furthermore, we underscore the connection between our population level and finite-sample level results: by setting the desired final accuracy $\epsilon$ in Theorem 5.1 to match that in Theorem 6.1 at the finite-sample level, namely letting $\epsilon = O((d/n)^{1/2})$ for sufficiently unbalanced fixed mixing weights and $\epsilon = O((d/n)^{1/4})$ for sufficiently balanced fixed mixing weights, we intuitively derive iteration complexity bounds $O(\log (1/\epsilon))=O(\log (n/d))$ and $O(\epsilon^{-2})=O((n/d)^{1/2})$ at the finite-sample level for sufficiently unbalanced and balanced initial mixing weights. We further extend our analysis in overspecified setting to low SNR regime.
Authors: Petr Spelda, Vit Stritecky
Abstract: Evaluation of reasoning language models gained importance after it was observed that they can combine their existing capabilities into novel traces of intermediate steps before task completion and that the traces can sometimes help them to generalize better than past models. As reasoning becomes the next scaling dimension of large language models, careful study of their capabilities in critical tasks is needed. We show that better performance is not always caused by test-time algorithmic improvements or model sizes but also by using impactful benchmarks as curricula for learning. We call this benchmark-driven selection of AI and show its effects on DeepSeek-R1 using our sequential decision-making problem from Humanity's Last Exam. Steering development of AI by impactful benchmarks trades evaluation for learning and makes novelty of test tasks key for measuring generalization capabilities of reasoning models. Consequently, some benchmarks could be seen as curricula for training rather than unseen test sets.
Authors: Rahul Janaa, Shubham Dixit, Mrityunjay Sharma, Ritesh Kumar
Abstract: Monitoring cattle health and optimizing yield are key challenges faced by dairy farmers due to difficulties in tracking all animals on the farm. This work aims to showcase modern data-driven farming practices based on explainable machine learning(ML) methods that explain the activity and behaviour of dairy cattle (cows). Continuous data collection of 3-axis accelerometer sensors and usage of robust ML methodologies and algorithms, provide farmers and researchers with actionable information on cattle activity, allowing farmers to make informed decisions and incorporate sustainable practices. This study utilizes Bluetooth-based Internet of Things (IoT) devices and 4G networks for seamless data transmission, immediate analysis, inference generation, and explains the models performance with explainability frameworks. Special emphasis is put on the pre-processing of the accelerometers time series data, including the extraction of statistical characteristics, signal processing techniques, and lag-based features using the sliding window technique. Various hyperparameter-optimized ML models are evaluated across varying window lengths for activity classification. The k-nearest neighbour Classifier achieved the best performance, with AUC of mean 0.98 and standard deviation of 0.0026 on the training set and 0.99 on testing set). In order to ensure transparency, Explainable AI based frameworks such as SHAP is used to interpret feature importance that can be understood and used by practitioners. A detailed comparison of the important features, along with the stability analysis of selected features, supports development of explainable and practical ML models for sustainable livestock management.
Authors: Will Fein, Ryan J. Horwitz, John E. Brown III, Amit Misra, Felipe Oviedo, Kevin White, Juan M. Lavista Ferres, Samuel K. Wasser
Abstract: The transnational ivory trade continues to drive the decline of elephant populations across Africa, and trafficking networks remain difficult to disrupt. Tusks seized by law enforcement officials carry forensic information on the traffickers responsible for their export, including DNA evidence and handwritten markings made by traffickers. For 20 years, analyses of tusk DNA have identified where elephants were poached and established connections among shipments of ivory. While the links established using genetic evidence are extremely conclusive, genetic data is expensive and sometimes impossible to obtain. But though handwritten markings are easy to photograph, they are rarely documented or analyzed. Here, we present an AI-driven pipeline for extracting and analyzing handwritten markings on seized elephant tusks, offering a novel, scalable, and low-cost source of forensic evidence. Having collected 6,085 photographs from eight large seizures of ivory over a 6-year period (2014-2019), we used an object detection model to extract over 17,000 individual markings, which were then labeled and described using state-of-the-art AI tools. We identified 184 recurring "signature markings" that connect the tusks on which they appear. 20 signature markings were observed in multiple seizures, establishing forensic links between these seizures through traffickers involved in both shipments. This work complements other investigative techniques by filling in gaps where other data sources are unavailable. The study demonstrates the transformative potential of AI in wildlife forensics and highlights practical steps for integrating handwriting analysis into efforts to disrupt organized wildlife crime.
Authors: Abdelmoula El Yazizi, Samee U. Khan, Yaroslav Koshka
Abstract: A local-valley (LV) centered approach to assessing the quality of sampling from Restricted Boltzmann Machines (RBMs) was applied to the latest generation of the D-Wave quantum annealer. D-Wave and Gibbs samples from a classically trained RBM were obtained at conditions relevant to the contrastive-divergence-based RBM learning. The samples were compared for the number of the LVs to which they belonged and the energy of the corresponding local minima. No significant (desirable) increase in the number of the LVs has been achieved by decreasing the D-Wave annealing time. At any training epoch, the states sampled by the D-Wave belonged to a somewhat higher number of LVs than in the Gibbs sampling. However, many of those LVs found by the two techniques differed. For high-probability sampled states, the two techniques were (unfavorably) less complementary and more overlapping. Nevertheless, many potentially "important" local minima, i.e., those having intermediate, even if not high, probability values, were found by only one of the two sampling techniques while missed by the other. The two techniques overlapped less at later than earlier training epochs, which is precisely the stage of the training when modest improvements to the sampling quality could make meaningful differences for the RBM trainability. The results of this work may explain the failure of previous investigations to achieve substantial (or any) improvement when using D-Wave-based sampling. However, the results reveal some potential for improvement, e.g., using a combined classical-quantum approach.
Authors: Li Sun, Shuheng Chen, Junyi Fan, Yong Si, Minoo Ahmadi, Elham Pishgar, Kamiar Alaei, Maryam Pishgar
Abstract: Background: Cirrhosis is a progressive liver disease with high mortality and frequent complications, notably acute kidney injury (AKI), which occurs in up to 50% of hospitalized patients and worsens outcomes. AKI stems from complex hemodynamic, inflammatory, and metabolic changes, making early detection essential. Many predictive tools lack accuracy, interpretability, and alignment with intensive care unit (ICU) workflows. This study developed an interpretable machine learning model for early AKI prediction in critically ill patients with cirrhosis. Methods: We conducted a retrospective analysis of the MIMIC-IV v2.2 database, identifying 1240 adult ICU patients with cirrhosis and excluding those with ICU stays under 48 hours or missing key data. Laboratory and physiological variables from the first 48 hours were extracted. The pipeline included preprocessing, missingness filtering, LASSO feature selection, and SMOTE class balancing. Six algorithms-LightGBM, CatBoost, XGBoost, logistic regression, naive Bayes, and neural networks-were trained and evaluated using AUROC, accuracy, F1-score, sensitivity, specificity, and predictive values. Results: LightGBM achieved the best performance (AUROC 0.808, 95% CI 0.741-0.856; accuracy 0.704; NPV 0.911). Key predictors included prolonged partial thromboplastin time, absence of outside-facility 20G placement, low pH, and altered pO2, consistent with known cirrhosis-AKI mechanisms and suggesting actionable targets. Conclusion: The LightGBM-based model enables accurate early AKI risk stratification in ICU patients with cirrhosis using routine clinical variables. Its high negative predictive value supports safe de-escalation for low-risk patients, and interpretability fosters clinician trust and targeted prevention. External validation and integration into electronic health record systems are warranted.
Authors: Jathin Korrapati, Patrick Mendoza, Aditya Tomar, Abein Abraham
Abstract: In-context learning (ICL) has emerged as a powerful capability of transformer-based language models, enabling them to perform tasks by conditioning on a small number of examples presented at inference time, without any parameter updates. Prior work has shown that transformers can generalize over simple function classes like linear functions, decision trees, even neural networks, purely from context, focusing on numerical or symbolic reasoning over underlying well-structured functions. Instead, we propose a novel application of ICL into the domain of cryptographic function learning, specifically focusing on ciphers such as mono-alphabetic substitution and Vigen\`ere ciphers, two classes of private-key encryption schemes. These ciphers involve a fixed but hidden bijective mapping between plain text and cipher text characters. Given a small set of (cipher text, plain text) pairs, the goal is for the model to infer the underlying substitution and decode a new cipher text word. This setting poses a structured inference challenge, which is well-suited for evaluating the inductive biases and generalization capabilities of transformers under the ICL paradigm. Code is available at https://github.com/adistomar/CS182-project.
Authors: Taibiao Zhao, Mingxuan Sun, Hao Wang, Xiaobing Chen, Xiangwei Zhou
Abstract: Transformer models have demonstrated exceptional performance and have become indispensable in computer vision (CV) and natural language processing (NLP) tasks. However, recent studies reveal that transformers are susceptible to backdoor attacks. Prior backdoor attack methods typically rely on retraining with clean data or altering the model architecture, both of which can be resource-intensive and intrusive. In this paper, we propose Head-wise Pruning and Malicious Injection (HPMI), a novel retraining-free backdoor attack on transformers that does not alter the model's architecture. Our approach requires only a small subset of the original data and basic knowledge of the model architecture, eliminating the need for retraining the target transformer. Technically, HPMI works by pruning the least important head and injecting a pre-trained malicious head to establish the backdoor. We provide a rigorous theoretical justification demonstrating that the implanted backdoor resists detection and removal by state-of-the-art defense techniques, under reasonable assumptions. Experimental evaluations across multiple datasets further validate the effectiveness of HPMI, showing that it 1) incurs negligible clean accuracy loss, 2) achieves at least 99.55% attack success rate, and 3) bypasses four advanced defense mechanisms. Additionally, relative to state-of-the-art retraining-dependent attacks, HPMI achieves greater concealment and robustness against diverse defense strategies, while maintaining minimal impact on clean accuracy.
Authors: Satyaranjan Pradhan, Madan Mohan Soren
Abstract: In this current work, we propose a Max Min approach for approximating functions using exponential neural network operators. We extend this framework to develop the Max Min Kantorovich-type exponential neural network operators and investigate their approximation properties. We study both pointwise and uniform convergence for univariate functions. To analyze the order of convergence, we use the logarithmic modulus of continuity and estimate the corresponding rate of convergence. Furthermore, we examine the convergence behavior of the Max Min Kantorovich type exponential neural network operators within the Orlicz space setting. We provide some graphical representations to illustrate the approximation error of the function through suitable kernel and sigmoidal activation functions.
Authors: Guanzi Yao, Heyao Liu, Linyan Dai
Abstract: This paper addresses the challenges of high resource dynamism and scheduling complexity in cloud-native database systems. It proposes an adaptive resource orchestration method based on multi-agent reinforcement learning. The method introduces a heterogeneous role-based agent modeling mechanism. This allows different resource entities, such as compute nodes, storage nodes, and schedulers, to adopt distinct policy representations. These agents are better able to reflect diverse functional responsibilities and local environmental characteristics within the system. A reward-shaping mechanism is designed to integrate local observations with global feedback. This helps mitigate policy learning bias caused by incomplete state observations. By combining real-time local performance signals with global system value estimation, the mechanism improves coordination among agents and enhances policy convergence stability. A unified multi-agent training framework is developed and evaluated on a representative production scheduling dataset. Experimental results show that the proposed method outperforms traditional approaches across multiple key metrics. These include resource utilization, scheduling latency, policy convergence speed, system stability, and fairness. The results demonstrate strong generalization and practical utility. Across various experimental scenarios, the method proves effective in handling orchestration tasks with high concurrency, high-dimensional state spaces, and complex dependency relationships. This confirms its advantages in real-world, large-scale scheduling environments.
Authors: Yuxi Wang, Heyao Liu, Nyutian Long, Guanzi Yao
Abstract: This paper proposes an anomaly detection method based on federated learning to address key challenges in multi-tenant cloud environments, including data privacy leakage, heterogeneous resource behavior, and the limitations of centralized modeling. The method establishes a federated training framework involving multiple tenants. Each tenant trains the model locally using private resource usage data. Through parameter aggregation, a global model is optimized, enabling cross-tenant collaborative anomaly detection while preserving data privacy. To improve adaptability to diverse resource usage patterns, a personalized parameter adjustment mechanism is introduced. This allows the model to retain tenant-specific feature representations while sharing global knowledge. In the model output stage, the Mahalanobis distance is used to compute anomaly scores. This enhances both the accuracy and stability of anomaly detection. The experiments use real telemetry data from a cloud platform to construct a simulated multi-tenant environment. The study evaluates the model's performance under varying participation rates and noise injection levels. These comparisons demonstrate the proposed method's robustness and detection accuracy. Experimental results show that the proposed method outperforms existing mainstream models across key metrics such as Precision, Recall, and F1-Score. It also maintains stable performance in various complex scenarios. These findings highlight the method's practical potential for intelligent resource monitoring and anomaly diagnosis in cloud computing environments.
Authors: Ryuta Matsuno
Abstract: This paper addresses source component shift adaptation, aiming to update predictions adapting to source component shifts for incoming data streams based on past training data. Existing online learning methods often fail to utilize recurring shifts effectively, while model-pool-based methods struggle to capture individual source components, leading to poor adaptation. In this paper, we propose a source component shift adaptation method via an offline decomposition and online mixing approach. We theoretically identify that the problem can be divided into two subproblems: offline source component decomposition and online mixing weight adaptation. Based on this, our method first determines prediction models, each of which learns a source component solely based on past training data offline through the EM algorithm. Then, it updates the mixing weight of the prediction models for precise prediction through online convex optimization. Thanks to our theoretical derivation, our method fully leverages the characteristics of the shifts, achieving superior adaptation performance over existing methods. Experiments conducted on various real-world regression datasets demonstrate that our method outperforms baselines, reducing the cumulative test loss by up to 67.4%.
Authors: Ricardo Diaz-Rincon, Muxuan Liang, Adolfo Ramirez-Zamora, Benjamin Shickel
Abstract: Parkinson's Disease (PD) medication management presents unique challenges due to heterogeneous disease progression and treatment response. Neurologists must balance symptom control with optimal dopaminergic dosing based on functional disability while minimizing side effects. This balance is crucial as inadequate or abrupt changes can cause levodopa-induced dyskinesia, wearing off, and neuropsychiatric effects, significantly reducing quality of life. Current approaches rely on trial-and-error decisions without systematic predictive methods. Despite machine learning advances, clinical adoption remains limited due to reliance on point predictions that do not account for prediction uncertainty, undermining clinical trust and utility. Clinicians require not only predictions of future medication needs but also reliable confidence measures. Without quantified uncertainty, adjustments risk premature escalation to maximum doses or prolonged inadequate symptom control. We developed a conformal prediction framework anticipating medication needs up to two years in advance with reliable prediction intervals and statistical guarantees. Our approach addresses zero-inflation in PD inpatient data, where patients maintain stable medication regimens between visits. Using electronic health records from 631 inpatient admissions at University of Florida Health (2011-2021), our two-stage approach identifies patients likely to need medication changes, then predicts required levodopa equivalent daily dose adjustments. Our framework achieved marginal coverage while reducing prediction interval lengths compared to traditional approaches, providing precise predictions for short-term planning and wider ranges for long-term forecasting. By quantifying uncertainty, our approach enables evidence-based decisions about levodopa dosing, optimizing symptom control while minimizing side effects and improving life quality.
Authors: Weijian Mai, Jiamin Wu, Yu Zhu, Zhouheng Yao, Dongzhan Zhou, Andrew F. Luo, Qihao Zheng, Wanli Ouyang, Chunfeng Song
Abstract: Deciphering how visual stimuli are transformed into cortical responses is a fundamental challenge in computational neuroscience. This visual-to-neural mapping is inherently a one-to-many relationship, as identical visual inputs reliably evoke variable hemodynamic responses across trials, contexts, and subjects. However, existing deterministic methods struggle to simultaneously model this biological variability while capturing the underlying functional consistency that encodes stimulus information. To address these limitations, we propose SynBrain, a generative framework that simulates the transformation from visual semantics to neural responses in a probabilistic and biologically interpretable manner. SynBrain introduces two key components: (i) BrainVAE models neural representations as continuous probability distributions via probabilistic learning while maintaining functional consistency through visual semantic constraints; (ii) A Semantic-to-Neural Mapper acts as a semantic transmission pathway, projecting visual semantics into the neural response manifold to facilitate high-fidelity fMRI synthesis. Experimental results demonstrate that SynBrain surpasses state-of-the-art methods in subject-specific visual-to-fMRI encoding performance. Furthermore, SynBrain adapts efficiently to new subjects with few-shot data and synthesizes high-quality fMRI signals that are effective in improving data-limited fMRI-to-image decoding performance. Beyond that, SynBrain reveals functional consistency across trials and subjects, with synthesized signals capturing interpretable patterns shaped by biological neural variability. The code will be made publicly available.
Authors: Danni Peng, Yuan Wang, Kangning Cai, Peiyan Ning, Jiming Xu, Yong Liu, Rick Siow Mong Goh, Qingsong Wei, Huazhu Fu
Abstract: In healthcare, federated learning (FL) is a widely adopted framework that enables privacy-preserving collaboration among medical institutions. With large foundation models (FMs) demonstrating impressive capabilities, using FMs in FL through cost-efficient adapter tuning has become a popular approach. Given the rapidly evolving healthcare environment, it is crucial for individual clients to quickly adapt to new tasks or diseases by tuning adapters while drawing upon past experiences. In this work, we introduce Federated Knowledge-Enhanced Initialization (FedKEI), a novel framework that leverages cross-client and cross-task transfer from past knowledge to generate informed initializations for learning new tasks with adapters. FedKEI begins with a global clustering process at the server to generalize knowledge across tasks, followed by the optimization of aggregation weights across clusters (inter-cluster weights) and within each cluster (intra-cluster weights) to personalize knowledge transfer for each new task. To facilitate more effective learning of the inter- and intra-cluster weights, we adopt a bi-level optimization scheme that collaboratively learns the global intra-cluster weights across clients and optimizes the local inter-cluster weights toward each client's task objective. Extensive experiments on three benchmark datasets of different modalities, including dermatology, chest X-rays, and retinal OCT, demonstrate FedKEI's advantage in adapting to new diseases compared to state-of-the-art methods.
Authors: Keke Gai, Dongjue Wang, Jing Yu, Liehuang Zhu, Qi Wu
Abstract: Existing backdoor defense methods in Federated Learning (FL) rely on the assumption of homogeneous client data distributions or the availability of a clean serve dataset, which limits the practicality and effectiveness. Defending against backdoor attacks under heterogeneous client data distributions while preserving model performance remains a significant challenge. In this paper, we propose a FL backdoor defense framework named CLIP-Fed, which leverages the zero-shot learning capabilities of vision-language pre-training models. By integrating both pre-aggregation and post-aggregation defense strategies, CLIP-Fed overcomes the limitations of Non-IID imposed on defense effectiveness. To address privacy concerns and enhance the coverage of the dataset against diverse triggers, we construct and augment the server dataset using the multimodal large language model and frequency analysis without any client samples. To address class prototype deviations caused by backdoor samples and eliminate the correlation between trigger patterns and target labels, CLIP-Fed aligns the knowledge of the global model and CLIP on the augmented dataset using prototype contrastive loss and Kullback-Leibler divergence. Extensive experiments on representative datasets validate the effectiveness of CLIP-Fed. Compared to state-of-the-art methods, CLIP-Fed achieves an average reduction in ASR, i.e., 2.03\% on CIFAR-10 and 1.35\% on CIFAR-10-LT, while improving average MA by 7.92\% and 0.48\%, respectively.
Authors: Claire Jie Zhang, Seyed A. Esmaeili, Jamie Morgenstern
Abstract: Fair clustering has traditionally focused on ensuring equitable group representation or equalizing group-specific clustering costs. However, Dickerson et al. (2025) recently showed that these fairness notions may yield undesirable or unintuitive clustering outcomes and advocated for a welfare-centric clustering approach that models the utilities of the groups. In this work, we model group utilities based on both distances and proportional representation and formalize two optimization objectives based on welfare-centric clustering: the Rawlsian (Egalitarian) objective and the Utilitarian objective. We introduce novel algorithms for both objectives and prove theoretical guarantees for them. Empirical evaluations on multiple real-world datasets demonstrate that our methods significantly outperform existing fair clustering baselines.
Authors: Md Ashraf Uddin, Nam H. Chu, Reza Rafeh
Abstract: The Internet of Medical Things (IoMT) is driving a healthcare revolution but remains vulnerable to cyberattacks such as denial of service, ransomware, data hijacking, and spoofing. These networks comprise resource constrained, heterogeneous devices (e.g., wearable sensors, smart pills, implantables), making traditional centralized Intrusion Detection Systems (IDSs) unsuitable due to response delays, privacy risks, and added vulnerabilities. Centralized IDSs require all sensors to transmit data to a central server, causing delays or network disruptions in dense environments. Running IDSs locally on IoMT devices is often infeasible due to limited computation, and even lightweight IDS components remain at risk if updated models are delayed leaving them exposed to zero-day attacks that threaten patient health and data security. We propose a multi level IoMT IDS framework capable of detecting zero day attacks and distinguishing between known and unknown threats. The first layer (near Edge) filters traffic at a coarse level (attack or not) using meta-learning or One Class Classification (OCC) with the usfAD algorithm. Subsequent layers (far Edge, Cloud) identify attack type and novelty. Experiments on the CICIoMT2024 dataset show 99.77 percentage accuracy and 97.8 percentage F1-score. The first layer detects zero-day attacks with high accuracy without needing new datasets, ensuring strong applicability in IoMT environments. Additionally, the meta-learning approach achieves high.
Authors: Samer Lahoud, Kinda Khawam
Abstract: Semantic communication aims to convey meaning rather than bit-perfect reproduction, representing a paradigm shift from traditional communication. This paper investigates distribution learning in semantic communication where receivers must infer the underlying meaning distribution through sequential observations. While semantic communication traditionally optimizes individual meaning transmission, we establish fundamental conditions for learning source statistics when priors are unknown. We prove that learnability requires full rank of the effective transmission matrix, characterize the convergence rate of distribution estimation, and quantify how estimation errors translate to semantic distortion. Our analysis reveals a fundamental trade-off: encoding schemes optimized for immediate semantic performance often sacrifice long-term learnability. Experiments on CIFAR-10 validate our theoretical framework, demonstrating that system conditioning critically impacts both learning rate and achievable performance. These results provide the first rigorous characterization of statistical learning in semantic communication and offer design principles for systems that balance immediate performance with adaptation capability.
Authors: Jiyong Kim, Jaeho Lee, Jiahao Lin, Alish Kanani, Miao Sun, Umit Y. Ogras, Jaehyun Park
Abstract: State Space Model (SSM)-based machine learning architectures have recently gained significant attention for processing sequential data. Mamba, a recent sequence-to-sequence SSM, offers competitive accuracy with superior computational efficiency compared to state-of-the-art transformer models. While this advantage makes Mamba particularly promising for resource-constrained edge devices, no hardware acceleration frameworks are currently optimized for deploying it in such environments. This paper presents eMamba, a comprehensive end-to-end hardware acceleration framework explicitly designed for deploying Mamba models on edge platforms. eMamba maximizes computational efficiency by replacing complex normalization layers with lightweight hardware-aware alternatives and approximating expensive operations, such as SiLU activation and exponentiation, considering the target applications. Then, it performs an approximation-aware neural architecture search (NAS) to tune the learnable parameters used during approximation. Evaluations with Fashion-MNIST, CIFAR-10, and MARS, an open-source human pose estimation dataset, show eMamba achieves comparable accuracy to state-of-the-art techniques using 1.63-19.9$\times$ fewer parameters. In addition, it generalizes well to large-scale natural language tasks, demonstrating stable perplexity across varying sequence lengths on the WikiText2 dataset. We also quantize and implement the entire eMamba pipeline on an AMD ZCU102 FPGA and ASIC using GlobalFoundries (GF) 22 nm technology. Experimental results show 4.95-5.62$\times$ lower latency and 2.22-9.95$\times$ higher throughput, with 4.77$\times$ smaller area, 9.84$\times$ lower power, and 48.6$\times$ lower energy consumption than baseline solutions while maintaining competitive accuracy.
Authors: Liyun Zhang, Jingcheng Ke, Shenli Fan, Xuanmeng Sha, Zheng Lian
Abstract: Recent works have emerged in multi-annotator learning that shift focus from Consensus-oriented Learning (CoL), which aggregates multiple annotations into a single ground-truth prediction, to Individual Tendency Learning (ITL), which models annotator-specific labeling behavior patterns (i.e., tendency) to provide explanation analysis for understanding annotator decisions. However, no evaluation framework currently exists to assess whether ITL methods truly capture individual tendencies and provide meaningful behavioral explanations. To address this gap, we propose the first unified evaluation framework with two novel metrics: (1) Difference of Inter-annotator Consistency (DIC) quantifies how well models capture annotator tendencies by comparing predicted inter-annotator similarity structures with ground-truth; (2) Behavior Alignment Explainability (BAE) evaluates how well model explanations reflect annotator behavior and decision relevance by aligning explainability-derived with ground-truth labeling similarity structures via Multidimensional Scaling (MDS). Extensive experiments validate the effectiveness of our proposed evaluation framework.
Authors: Aditya Tomar, Coleman Hooper, Minjae Lee, Haocheng Xi, Rishabh Tiwari, Wonjun Kang, Luca Manolache, Michael W. Mahoney, Kurt Keutzer, Amir Gholami
Abstract: Although LLM inference has emerged as a critical workload for many downstream applications, efficiently inferring LLMs is challenging due to the substantial memory footprint and bandwidth requirements. In parallel, compute capabilities have steadily outpaced both memory capacity and bandwidth over the last few decades, a trend that remains evident in modern GPU hardware and exacerbates the challenge of LLM inference. As such, new algorithms are emerging that trade increased computation for reduced memory operations. To that end, we present XQuant, which takes advantage of this trend, enabling an order-of-magnitude reduction in memory consumption through low-bit quantization with substantial accuracy benefits relative to state-of-the-art KV cache quantization methods. We accomplish this by quantizing and caching the layer input activations X, instead of using standard KV caching, and then rematerializing the Keys and Values on-the-fly during inference. This results in an immediate 2$\times$ memory savings compared to KV caching. By applying XQuant, we achieve up to $\sim 7.7\times$ memory savings with $<0.1$ perplexity degradation compared to the FP16 baseline. Furthermore, our approach leverages the fact that X values are similar across layers. Building on this observation, we introduce XQuant-CL, which exploits the cross-layer similarity in the X embeddings for extreme compression. Across different models, XQuant-CL attains up to 10$\times$ memory savings relative to the FP16 baseline with only 0.01 perplexity degradation, and 12.5$\times$ memory savings with only $0.1$ perplexity degradation. XQuant exploits the rapidly increasing compute capabilities of hardware platforms to eliminate the memory bottleneck, while surpassing state-of-the-art KV cache quantization methods and achieving near-FP16 accuracy across a wide range of models.
Authors: Pengbo Shen, Yaqing Wang, Ni Mu, Yao Luan, Runpeng Xie, Senhao Yang, Lexiang Wang, Hao Hu, Shuang Xu, Yiqin Yang, Bo Xu
Abstract: Evaluating large language models (LLMs) in complex decision-making is essential for advancing AI's ability for strategic planning and real-time adaptation. However, existing benchmarks for tasks like StarCraft II fail to capture the game's full complexity, such as its complete game context, diverse action spaces, and all playable races. To address this gap, we present SC2Arena, a benchmark that fully supports all playable races, low-level action spaces, and optimizes text-based observations to tackle spatial reasoning challenges. Complementing this, we introduce StarEvolve, a hierarchical framework that integrates strategic planning with tactical execution, featuring iterative self-correction and continuous improvement via fine-tuning on high-quality gameplay data. Its key components include a Planner-Executor-Verifier structure to break down gameplay, and a scoring system for selecting high-quality training samples. Comprehensive analysis using SC2Arena provides valuable insights into developing generalist agents that were not possible with previous benchmarks. Experimental results also demonstrate that our proposed StarEvolve achieves superior performance in strategic planning. Our code, environment, and algorithms are publicly available.
Authors: Tianxiao Cao, Kyohei Atarashi, Hisashi Kashima
Abstract: Sharpness-Aware Minimization (SAM) has been proven to be an effective optimization technique for improving generalization in overparameterized models. While prior works have explored the implicit regularization of SAM in simple two-core scale-invariant settings, its behavior in more general tensorized or scale-invariant models remains underexplored. In this work, we leverage scale-invariance to analyze the norm dynamics of SAM in general tensorized models. We introduce the notion of \emph{Norm Deviation} as a global measure of core norm imbalance, and derive its evolution under SAM using gradient flow analysis. We show that SAM's implicit control of Norm Deviation is governed by the covariance between core norms and their gradient magnitudes. Motivated by these findings, we propose a simple yet effective method, \emph{Deviation-Aware Scaling (DAS)}, which explicitly mimics this regularization behavior by scaling core norms in a data-adaptive manner. Our experiments across tensor completion, noisy training, model compression, and parameter-efficient fine-tuning confirm that DAS achieves competitive or improved performance over SAM, while offering reduced computational overhead.
Authors: Asiful Arefeen, Shovito Barua Soumma, Hassan Ghasemzadeh
Abstract: Counterfactual explanations provide human-understandable reasoning for AI-made decisions by describing minimal changes to input features that would alter a model's prediction. To be truly useful in practice, such explanations must be realistic and feasible -- they should respect both the underlying data distribution and user-defined feasibility constraints. Existing approaches often enforce inter-feature dependencies through rigid, hand-crafted constraints or domain-specific knowledge, which limits their generalizability and ability to capture complex, nonlinear relations inherent in data. Moreover, they rarely accommodate user-specified preferences and suggest explanations that are causally implausible or infeasible to act upon. We introduce RealAC, a domain-agnostic framework for generating realistic and actionable counterfactuals. RealAC automatically preserves complex inter-feature dependencies without relying on explicit domain knowledge -- by aligning the joint distributions of feature pairs between factual and counterfactual instances. The framework also allows end-users to ``freeze'' attributes they cannot or do not wish to change by suppressing change in frozen features during optimization. Evaluations on three synthetic and two real datasets demonstrate that RealAC balances realism with actionability. Our method outperforms state-of-the-art baselines and Large Language Model-based counterfactual generation techniques in causal edge score, dependency preservation score, and IM1 realism metric and offers a solution for causality-aware and user-centric counterfactual generation.
Authors: Prajit Sengupta, Islem Rekik
Abstract: Graph neural networks (GNNs) have achieved state-of-the-art results in computer vision and medical image classification tasks by capturing structural dependencies across data instances. However, their decision-making remains largely opaque, limiting their trustworthiness in high-stakes clinical applications where interpretability is essential. Existing explainability techniques for GNNs are typically post-hoc and global, offering limited insight into individual node decisions or local reasoning. We introduce X-Node, a self-explaining GNN framework in which each node generates its own explanation as part of the prediction process. For every node, we construct a structured context vector encoding interpretable cues such as degree, centrality, clustering, feature saliency, and label agreement within its local topology. A lightweight Reasoner module maps this context into a compact explanation vector, which serves three purposes: (1) reconstructing the node's latent embedding via a decoder to enforce faithfulness, (2) generating a natural language explanation using a pre-trained LLM (e.g., Grok or Gemini), and (3) guiding the GNN itself via a "text-injection" mechanism that feeds explanations back into the message-passing pipeline. We evaluate X-Node on two graph datasets derived from MedMNIST and MorphoMNIST, integrating it with GCN, GAT, and GIN backbones. Our results show that X-Node maintains competitive classification accuracy while producing faithful, per-node explanations. Repository: https://github.com/basiralab/X-Node.
Authors: Xinrui Li, Qilin Fan, Tianfu Wang, Kaiwen Wei, Ke Yu, Xu Zhang
Abstract: Federated graph learning (FGL) enables multiple clients to collaboratively train powerful graph neural networks without sharing their private, decentralized graph data. Inherited from generic federated learning, FGL is critically challenged by statistical heterogeneity, where non-IID data distributions across clients can severely impair model performance. A particularly destructive form of this is class imbalance, which causes the global model to become biased towards majority classes and fail at identifying rare but critical events. This issue is exacerbated in FGL, as nodes from a minority class are often surrounded by biased neighborhood information, hindering the learning of expressive embeddings. To grapple with this challenge, we propose GraphFedMIG, a novel FGL framework that reframes the problem as a federated generative data augmentation task. GraphFedMIG employs a hierarchical generative adversarial network where each client trains a local generator to synthesize high-fidelity feature representations. To provide tailored supervision, clients are grouped into clusters, each sharing a dedicated discriminator. Crucially, the framework designs a mutual information-guided mechanism to steer the evolution of these client generators. By calculating each client's unique informational value, this mechanism corrects the local generator parameters, ensuring that subsequent rounds of mutual information-guided generation are focused on producing high-value, minority-class features. We conduct extensive experiments on four real-world datasets, and the results demonstrate the superiority of the proposed GraphFedMIG compared with other baselines.
Authors: Lisa Haxel, Jaivardhan Kapoor, Ulf Ziemann, Jakob H. Macke
Abstract: Brain-computer interfaces (BCIs) suffer from accuracy degradation as neural signals drift over time and vary across users, requiring frequent recalibration that limits practical deployment. We introduce EDAPT, a task- and model-agnostic framework that eliminates calibration through continual model adaptation. EDAPT first trains a baseline decoder using data from multiple users, then continually personalizes this model via supervised finetuning as the neural patterns evolve during use. We tested EDAPT across nine datasets covering three BCI tasks, and found that it consistently improved accuracy over conventional, static methods. These improvements primarily stem from combining population-level pretraining and online continual finetuning, with unsupervised domain adaptation providing further gains on some datasets. EDAPT runs efficiently, updating models within 200 milliseconds on consumer-grade hardware. Finally, decoding accuracy scales with total data budget rather than its allocation between subjects and trials. EDAPT provides a practical pathway toward calibration-free BCIs, reducing a major barrier to BCI deployment.
Authors: Alexander Merkov, David Rohde, Alexandre Gilotte, Benjamin Heymann
Abstract: Unobserved confounding arises when an unmeasured feature influences both the treatment and the outcome, leading to biased causal effect estimates. This issue undermines observational studies in fields like economics, medicine, ecology or epidemiology. Recommender systems leveraging fully observed data seem not to be vulnerable to this problem. However many standard practices in recommender systems result in observed features being ignored, resulting in effectively the same problem. This paper will show that numerous common practices such as feature engineering, A/B testing and modularization can in fact introduce confounding into recommendation systems and hamper their performance. Several illustrations of the phenomena are provided, supported by simulation studies with practical suggestions about how practitioners may reduce or avoid the affects of confounding in real systems.
Authors: Panagiotis D. Grontas, Antonio Terpin, Efe C. Balta, Raffaello D'Andrea, John Lygeros
Abstract: We introduce an output layer for neural networks that ensures satisfaction of convex constraints. Our approach, $\Pi$net, leverages operator splitting for rapid and reliable projections in the forward pass, and the implicit function theorem for backpropagation. We deploy $\Pi$net as a feasible-by-design optimization proxy for parametric constrained optimization problems and obtain modest-accuracy solutions faster than traditional solvers when solving a single problem, and significantly faster for a batch of problems. We surpass state-of-the-art learning approaches in terms of training time, solution quality, and robustness to hyperparameter tuning, while maintaining similar inference times. Finally, we tackle multi-vehicle motion planning with non-convex trajectory preferences and provide $\Pi$net as a GPU-ready package implemented in JAX with effective tuning heuristics.
Authors: Jonas Ulmen, Ganesh Sundaram, Daniel G\"orges
Abstract: With the advent of Joint Embedding Predictive Architectures (JEPAs), which appear to be more capable than reconstruction-based methods, this paper introduces a novel technique for creating world models using continuous-time dynamic systems from arbitrary observation data. The proposed method integrates sequence embeddings with neural ordinary differential equations (neural ODEs). It employs loss functions that enforce contractive embeddings and Lipschitz constants in state transitions to construct a well-organized latent state space. The approach's effectiveness is demonstrated through the generation of structured latent state-space models for a simple pendulum system using only image data. This opens up a new technique for developing more general control algorithms and estimation techniques with broad applications in robotics.
Authors: Amir Mehrpanah, Matteo Gamba, Kevin Smith, Hossein Azizpour
Abstract: ReLU networks, while prevalent for visual data, have sharp transitions, sometimes relying on individual pixels for predictions, making vanilla gradient-based explanations noisy and difficult to interpret. Existing methods, such as GradCAM, smooth these explanations by producing surrogate models at the cost of faithfulness. We introduce a unifying spectral framework to systematically analyze and quantify smoothness, faithfulness, and their trade-off in explanations. Using this framework, we quantify and regularize the contribution of ReLU networks to high-frequency information, providing a principled approach to identifying this trade-off. Our analysis characterizes how surrogate-based smoothing distorts explanations, leading to an ``explanation gap'' that we formally define and measure for different post-hoc methods. Finally, we validate our theoretical findings across different design choices, datasets, and ablations.
Authors: Che-Yu Chou, Hung-Hsuan Chen
Abstract: Although one-hot encoding is commonly used for multiclass classification, it is not always the most effective encoding mechanism. Error Correcting Output Codes (ECOC) address multiclass classification by mapping each class to a unique codeword used as a label. Traditional ECOC methods rely on manually designed or randomly generated codebooks, which are labor-intensive and may yield suboptimal, dataset-agnostic results. This paper introduces three models for automated codebook learning based on contrastive learning, allowing codebooks to be learned directly and adaptively from data. Across four datasets, our proposed models demonstrate superior robustness to adversarial attacks compared to two baselines. The source is available at https://github.com/YuChou20/Automated-Codebook-Learning-with-Error-Correcting-Output-Code-Technique.
URLs: https://github.com/YuChou20/Automated-Codebook-Learning-with-Error-Correcting-Output-Code-Technique.
Authors: Jiulin Li, Ping Huang, Yexin Li, Shuo Chen, Juewen Hu, Ye Tian
Abstract: Real-world multimodal applications often require any-to-any capabilities, enabling both understanding and generation across modalities including text, image, audio, and video. However, integrating the strengths of autoregressive language models (LLMs) for reasoning and diffusion models for high-fidelity generation remains challenging. Existing approaches rely on rigid pipelines or tightly coupled architectures, limiting flexibility and scalability. We propose MAGUS (Multi-Agent Guided Unified Multimodal System), a modular framework that unifies multimodal understanding and generation via two decoupled phases: Cognition and Deliberation. MAGUS enables symbolic multi-agent collaboration within a shared textual workspace. In the Cognition phase, three role-conditioned multimodal LLM agents - Perceiver, Planner, and Reflector - engage in collaborative dialogue to perform structured understanding and planning. The Deliberation phase incorporates a Growth-Aware Search mechanism that orchestrates LLM-based reasoning and diffusion-based generation in a mutually reinforcing manner. MAGUS supports plug-and-play extensibility, scalable any-to-any modality conversion, and semantic alignment - all without the need for joint training. Experiments across multiple benchmarks, including image, video, and audio generation, as well as cross-modal instruction following, demonstrate that MAGUS outperforms strong baselines and state-of-the-art systems. Notably, on the MME benchmark, MAGUS surpasses the powerful closed-source model GPT-4o.
Authors: Dmitrii Dobrynin, Masoud Mohseni, John Paul Strachan
Abstract: Optimizing or sampling complex cost functions of combinatorial optimization problems is a longstanding challenge across disciplines and applications. When employing family of conventional algorithms based on Markov Chain Monte Carlo (MCMC) such as simulated annealing or parallel tempering, one assumes homogeneous (equilibrium) temperature profiles across input. This instance independent approach was shown to be ineffective for the hardest benchmarks near a computational phase transition when the so-called overlap-gap-property holds. In these regimes conventional MCMC struggles to unfreeze rigid variables, escape suboptimal basins of attraction, and sample high-quality and diverse solutions. In order to mitigate these challenges, Nonequilibrium Nonlocal Monte Carlo (NMC) algorithms were proposed that leverage inhomogeneous temperature profiles thereby accelerating exploration of the configuration space without compromising its exploitation. Here, we employ deep reinforcement learning (RL) to train the nonlocal transition policies of NMC which were previously designed phenomenologically. We demonstrate that the resulting solver can be trained solely by observing energy changes of the configuration space exploration as RL rewards and the local minimum energy landscape geometry as RL states. We further show that the trained policies improve upon the standard MCMC-based and nonlocal simulated annealing on hard uniform random and scale-free random 4-SAT benchmarks in terms of residual energy, time-to-solution, and diversity of solutions metrics.
Authors: Hao Luan, Yi Xian Goh, See-Kiong Ng, Chun Kai Ling
Abstract: Modifications to test-time sampling have emerged as an important extension to diffusion algorithms, with the goal of biasing the generative process to achieve a given objective without having to retrain the entire diffusion model. However, generating jointly correlated samples from multiple pre-trained diffusion models while simultaneously enforcing task-specific constraints without costly retraining has remained challenging. To this end, we propose Projected Coupled Diffusion (PCD), a novel test-time framework for constrained joint generation. PCD introduces a coupled guidance term into the generative dynamics to encourage coordination between diffusion models and incorporates a projection step at each diffusion step to enforce hard constraints. Empirically, we demonstrate the effectiveness of PCD in application scenarios of image-pair generation, object manipulation, and multi-robot motion planning. Our results show improved coupling effects and guaranteed constraint satisfaction without incurring excessive computational costs.
Authors: Brian Shing-Hei Wong, Joshua Mincheol Kim, Sin-Hang Fung, Qing Xiong, Kelvin Fu-Kiu Ao, Junkang Wei, Ran Wang, Dan Michelle Wang, Jingying Zhou, Bo Feng, Alfred Sze-Lok Cheng, Kevin Y. Yip, Stephen Kwok-Wing Tsui, Qin Cao
Abstract: Allergens, typically proteins capable of triggering adverse immune responses, represent a significant public health challenge. To accurately identify allergen proteins, we introduce Applm (Allergen Prediction with Protein Language Models), a computational framework that leverages the 100-billion parameter xTrimoPGLM protein language model. We show that Applm consistently outperforms seven state-of-the-art methods in a diverse set of tasks that closely resemble difficult real-world scenarios. These include identifying novel allergens that lack similar examples in the training set, differentiating between allergens and non-allergens among homologs with high sequence similarity, and assessing functional consequences of mutations that create few changes to the protein sequences. Our analysis confirms that xTrimoPGLM, originally trained on one trillion tokens to capture general protein sequence characteristics, is crucial for Applm's performance by detecting important differences among protein sequences. In addition to providing Applm as open-source software, we also provide our carefully curated benchmark datasets to facilitate future research.
Authors: Zetian Sun, Dongfang Li, Zhuoen Chen, Yuhuai Qin, Baotian Hu
Abstract: Reward sparsity in long-horizon reinforcement learning (RL) tasks remains a significant challenge, while existing outcome-based reward shaping struggles to define meaningful immediate rewards without introducing bias or requiring explicit task decomposition. Alternatively, verification-based reward shaping uses stepwise critics, but misalignment between immediate rewards and long-term objectives can lead to reward hacking and suboptimal policies. In this work, we address this problem in the context of software engineering (SWE) tasks, where multi-turn reasoning and rule-based verification are critical. We introduce the SWE-oriented RL Framework, a unified system supporting multi-turn interaction, docker-based execution, and customizable reward functions. Additionally, we propose Gated Reward Accumulation (G-RA), a novel method that accumulates immediate rewards only when high-level (long-term) rewards meet a predefined threshold, ensuring stable RL optimization. Experiments on SWE-bench Verified and kBench demonstrate that G-RA leads to an increase in completion rates (47.6\% \rightarrow 93.8\% and 22.0\% \rightarrow 86.0\%) and modification rates (19.6\% \rightarrow 23.8\% and 12.0\% \rightarrow 42.0\%), while avoiding policy degradation caused by reward misalignment. Our findings highlight the importance of balanced reward accumulation in long-horizon RL and provide a practical solution.
Authors: Jeroen Berrevoets, Julianna Piskorz, Robert Davis, Harry Amad, Jim Weatherall, Mihaela van der Schaar
Abstract: Estimating treatment effects (TE) from observational data is a critical yet complex task in many fields, from healthcare and economics to public policy. While recent advances in machine learning and causal inference have produced powerful estimation techniques, their adoption remains limited due to the need for deep expertise in causal assumptions, adjustment strategies, and model selection. In this paper, we introduce CATE-B, an open-source co-pilot system that uses large language models (LLMs) within an agentic framework to guide users through the end-to-end process of treatment effect estimation. CATE-B assists in (i) constructing a structural causal model via causal discovery and LLM-based edge orientation, (ii) identifying robust adjustment sets through a novel Minimal Uncertainty Adjustment Set criterion, and (iii) selecting appropriate regression methods tailored to the causal structure and dataset characteristics. To encourage reproducibility and evaluation, we release a suite of benchmark tasks spanning diverse domains and causal complexities. By combining causal inference with intelligent, interactive assistance, CATE-B lowers the barrier to rigorous causal analysis and lays the foundation for a new class of benchmarks in automated treatment effect estimation.
Authors: Furkan Pala, Islem Rekik
Abstract: Deep learning models often struggle to maintain generalizability in medical imaging, particularly under domain-fracture scenarios where distribution shifts arise from varying imaging techniques, acquisition protocols, patient populations, demographics, and equipment. In practice, each hospital may need to train distinct models - differing in learning task, width, and depth - to match local data. For example, one hospital may use Euclidean architectures such as MLPs and CNNs for tabular or grid-like image data, while another may require non-Euclidean architectures such as graph neural networks (GNNs) for irregular data like brain connectomes. How to train such heterogeneous models coherently across datasets, while enhancing each model's generalizability, remains an open problem. We propose unified learning, a new paradigm that encodes each model into a graph representation, enabling unification in a shared graph learning space. A GNN then guides optimization of these unified models. By decoupling parameters of individual models and controlling them through a unified GNN (uGNN), our method supports parameter sharing and knowledge transfer across varying architectures (MLPs, CNNs, GNNs) and distributions, improving generalizability. Evaluations on MorphoMNIST and two MedMNIST benchmarks - PneumoniaMNIST and BreastMNIST - show that unified learning boosts performance when models are trained on unique distributions and tested on mixed ones, demonstrating strong robustness to unseen data with large distribution shifts. Code and benchmarks: https://github.com/basiralab/uGNN
Authors: Xuanhao Mu, G\"okhan Demirel, Yuzhe Zhang, Jianlei Liu, Thorsten Schlachter, Veit Hagenmeyer
Abstract: To bridge the temporal granularity gap in energy network design and operation based on Energy System Models, resampling of time series is required. While conventional upsampling methods are computationally efficient, they often result in significant information loss or increased noise. Advanced models such as time series generation models, Super-Resolution models and imputation models show potential, but also face fundamental challenges. The goal of time series generative models is to learn the distribution of the original data to generate high-resolution series with similar statistical characteristics. This is not entirely consistent with the definition of upsampling. Time series Super-Resolution models or imputation models can degrade the accuracy of upsampling because the input low-resolution time series are sparse and may have insufficient context. Moreover, such models usually rely on supervised learning paradigms. This presents a fundamental application paradox: their training requires the high-resolution time series that is intrinsically absent in upsampling application scenarios. To address the mentioned upsampling issue, this paper introduces a new method utilizing Generative Adversarial Transformers (GATs), which can be trained without access to any ground-truth high-resolution data. Compared with conventional interpolation methods, the introduced method can reduce the root mean square error (RMSE) of upsampling tasks by 9%, and the accuracy of a model predictive control (MPC) application scenario is improved by 13%.
Authors: Yunfeng Zhao, Yixin Liu, Shiyuan Li, Qingfeng Chen, Yu Zheng, Shirui Pan
Abstract: Graph Anomaly Detection (GAD) aims to identify nodes that deviate from the majority within a graph, playing a crucial role in applications such as social networks and e-commerce. Despite the current advancements in deep learning-based GAD, existing approaches often suffer from high deployment costs and poor scalability due to their complex and resource-intensive training processes. Surprisingly, our empirical findings suggest that the training phase of deep GAD methods, commonly perceived as crucial, may actually contribute less to anomaly detection performance than expected. Inspired by this, we propose FreeGAD, a novel training-free yet effective GAD method. Specifically, it leverages an affinity-gated residual encoder to generate anomaly-aware representations. Meanwhile, FreeGAD identifies anchor nodes as pseudo-normal and anomalous guides, followed by calculating anomaly scores through anchor-guided statistical deviations. Extensive experiments demonstrate that FreeGAD achieves superior anomaly detection performance, efficiency, and scalability on multiple benchmark datasets from diverse domains, without any training or iterative optimization.
Authors: Amir Mehrpanah, Erik Englesson, Hossein Azizpour
Abstract: Understanding the behavior of deep networks is crucial to increase our confidence in their results. Despite an extensive body of work for explaining their predictions, researchers have faced reliability issues, which can be attributed to insufficient formalism. In our research, we adopt novel probabilistic and spectral perspectives to formally analyze explanation methods. Our study reveals a pervasive spectral bias stemming from the use of gradient, and sheds light on some common design choices that have been discovered experimentally, in particular, the use of squared gradient and input perturbation. We further characterize how the choice of perturbation hyperparameters in explanation methods, such as SmoothGrad, can lead to inconsistent explanations and introduce two remedies based on our proposed formalism: (i) a mechanism to determine a standard perturbation scale, and (ii) an aggregation method which we call SpectralLens. Finally, we substantiate our theoretical results through quantitative evaluations.
Authors: Tanveer Khan, Antonis Michalas
Abstract: Split Learning (SL) is a collaborative learning approach that improves privacy by keeping data on the client-side while sharing only the intermediate output with a server. However, the distributed nature of SL introduces new security challenges, necessitating a comprehensive exploration of potential attacks. This paper systematically reviews various attacks on SL, classifying them based on factors such as the attacker's role, the type of privacy risks, when data leaks occur, and where vulnerabilities exist. We also analyze existing defense methods, including cryptographic methods, data modification approaches, distributed techniques, and hybrid solutions. Our findings reveal security gaps, highlighting the effectiveness and limitations of existing defenses. By identifying open challenges and future directions, this work provides valuable information to improve SL privacy issues and guide further research.
Authors: Davide Guidobene, Lorenzo Benedetti, Diego Arapovic
Abstract: Multi-Objective Reinforcement Learning (MORL) is a generalization of traditional Reinforcement Learning (RL) that aims to optimize multiple, often conflicting objectives simultaneously rather than focusing on a single reward. This approach is crucial in complex decision-making scenarios where agents must balance trade-offs between various goals, such as maximizing performance while minimizing costs. We consider the problem of MORL where the objectives are combined using a non-linear scalarization function. Just like in standard RL, policy gradient methods (PGMs) are amongst the most effective for handling large and continuous state-action spaces in MORL. However, existing PGMs for MORL suffer from high sample inefficiency, requiring large amounts of data to be effective. Previous attempts to solve this problem rely on overly strict assumptions, losing PGMs' benefits in scalability to large state-action spaces. In this work, we address the issue of sample efficiency by implementing variance-reduction techniques to reduce the sample complexity of policy gradients while maintaining general assumptions.
Authors: Lucas Cardoso, Vitor Santos, Jos\'e Ribeiro Filho, Ricardo Prud\^encio, Regiane Kawasaki, Ronnie Alves
Abstract: Robust validation of Machine Learning (ML) models is essential, but traditional data partitioning approaches often ignore the intrinsic quality of each instance. This study proposes the use of Item Response Theory (IRT) parameters to characterize and guide the partitioning of datasets in the model validation stage. The impact of IRT-informed partitioning strategies on the performance of several ML models in four tabular datasets was evaluated. The results obtained demonstrate that IRT reveals an inherent heterogeneity of the instances and highlights the existence of informative subgroups of instances within the same dataset. Based on IRT, balanced partitions were created that consistently help to better understand the tradeoff between bias and variance of the models. In addition, the guessing parameter proved to be a determining factor: training with high-guessing instances can significantly impair model performance and resulted in cases with accuracy below 50%, while other partitions reached more than 70% in the same dataset.
Authors: Patrick Soga, Zhenyu Lei, Yinhan He, Camille Bilodeau, Jundong Li
Abstract: Predicting changes in binding free energy ($\Delta\Delta G$) is a vital task in protein engineering and protein-protein interaction (PPI) engineering for drug discovery. Previous works have observed a high correlation between $\Delta\Delta G$ and entropy, using probabilities of biologically important objects such as side chain angles and residue identities to estimate $\Delta\Delta G$. However, estimating the full conformational distribution of a protein complex is generally considered intractable. In this work, we propose a new approach to $\Delta\Delta G$ prediction that avoids this issue by instead leveraging energy-based models for estimating the probability of a complex's conformation. Specifically, we novelly decompose $\Delta\Delta G$ into a sequence-based component estimated by an inverse folding model and a structure-based component estimated by an energy model. This decomposition is made tractable by assuming equilibrium between the bound and unbound states, allowing us to simplify the estimation of degeneracies associated with each state. Unlike previous deep learning-based methods, our method incorporates an energy-based physical inductive bias by connecting the often-used sequence log-odds ratio-based approach to $\Delta\Delta G$ prediction with a new $\Delta\Delta E$ term grounded in statistical mechanics. We demonstrate superiority over existing state-of-the-art structure and sequence-based deep learning methods in $\Delta\Delta G$ prediction and antibody optimization against SARS-CoV-2.
Authors: Yihua Wang, Qi Jia, Cong Xu, Feiyu Chen, Yuhan Liu, Haotian Zhang, Liang Jin, Lu Liu, Zhichun Wang
Abstract: Multimodal sarcasm detection is a complex task that requires distinguishing subtle complementary signals across modalities while filtering out irrelevant information. Many advanced methods rely on learning shortcuts from datasets rather than extracting intended sarcasm-related features. However, our experiments show that shortcut learning impairs the model's generalization in real-world scenarios. Furthermore, we reveal the weaknesses of current modality fusion strategies for multimodal sarcasm detection through systematic experiments, highlighting the necessity of focusing on effective modality fusion for complex emotion recognition. To address these challenges, we construct MUStARD++$^{R}$ by removing shortcut signals from MUStARD++. Then, a Multimodal Conditional Information Bottleneck (MCIB) model is introduced to enable efficient multimodal fusion for sarcasm detection. Experimental results show that the MCIB achieves the best performance without relying on shortcut learning.
Authors: Chenkai Guo, Yikai Zhu, Jing Yangum, Renxiang Guan, Por Lip Yee, Guangdun Peng, Dayu Hu
Abstract: By incorporating spatial location information, spatial-transcriptomics clustering yields more comprehensive insights into cell subpopulation identification. Despite recent progress, existing methods have at least two limitations: (i) topological learning typically considers only representations of individual cells or their interaction graphs; however, spatial transcriptomic profiles are often noisy, making these approaches vulnerable to low-quality topological signals, and (ii) insufficient modeling of spatial neighborhood information leads to low-quality spatial embeddings. To address these limitations, we propose SPHENIC, a novel Spatial Persistent Homology Enhanced Neighborhood Integrative Clustering method. Specifically, SPHENIC incorporates invariant topological features into the clustering network to achieve stable representation learning. Additionally, to construct high-quality spatial embeddings that reflect the true cellular distribution, we design the Spatial Constraint and Distribution Optimization Module (SCDOM). This module increases the similarity between a cell's embedding and those of its spatial neighbors, decreases similarity with non-neighboring cells, and thereby produces clustering-friendly spatial embeddings. Extensive experiments on 14 benchmark spatial transcriptomic slices demonstrate that SPHENIC achieves superior performance on the spatial clustering task, outperforming existing state-of-the-art methods by 3.31%-6.54% over the best alternative.
Authors: Debvrat Varshney, Vibhas Vats, Bhartendu Pandey, Christa Brelsford, Philipe Dias
Abstract: Land cover, both present and future, has a significant effect on several important Earth system processes. For example, impervious surfaces heat up and speed up surface water runoff and reduce groundwater infiltration, with concomitant effects on regional hydrology and flood risk. While regional Earth System models have increasing skill at forecasting hydrologic and atmospheric processes at high resolution in future climate scenarios, our ability to forecast land-use and land-cover change (LULC), a critical input to risk and consequences assessment for these scenarios, has lagged behind. In this paper, we propose a new paradigm exploiting Generative AI (GenAI) for land cover change forecasting by framing LULC forecasting as a data synthesis problem conditioned on historical and auxiliary data-sources. We discuss desirable properties of generative models that fundament our research premise, and demonstrate the feasibility of our methodology through experiments on imperviousness forecasting using historical data covering the entire conterminous United States. Specifically, we train a diffusion model for decadal forecasting of imperviousness and compare its performance to a baseline that assumes no change at all. Evaluation across 12 metropolitan areas for a year held-out during training indicate that for average resolutions $\geq 0.7\times0.7km^2$ our model yields MAE lower than such a baseline. This finding corroborates that such a generative model can capture spatiotemporal patterns from historical data that are significant for projecting future change. Finally, we discuss future research to incorporate auxiliary information on physical properties about the Earth, as well as supporting simulation of different scenarios by means of driver variables.
Authors: Reijo Jaakkola, Tomi Janhunen, Antti Kuusisto, Magdalena Ortiz, Matias Selin, Mantas \v{S}imkus
Abstract: We present a novel approach for graph classification based on tabularizing graph data via variants of the Weisfeiler-Leman algorithm and then applying methods for tabular data. We investigate a comprehensive class of Weisfeiler-Leman variants obtained by modifying the underlying logical framework and establish a precise theoretical characterization of their expressive power. We then test two selected variants on twelve benchmark datasets that span a range of different domains. The experiments demonstrate that our approach matches the accuracy of state-of-the-art graph neural networks and graph kernels while being more time or memory efficient, depending on the dataset. We also briefly discuss directly extracting interpretable modal logic formulas from graph datasets.
Authors: Yuchen Zhu, Wei Guo, Jaemoo Choi, Guan-Horng Liu, Yongxin Chen, Molei Tao
Abstract: We study the problem of learning a neural sampler to generate samples from discrete state spaces where the target probability mass function $\pi\propto\mathrm{e}^{-U}$ is known up to a normalizing constant, which is an important task in fields such as statistical physics, machine learning, combinatorial optimization, etc. To better address this challenging task when the state space has a large cardinality and the distribution is multi-modal, we propose $\textbf{M}$asked $\textbf{D}$iffusion $\textbf{N}$eural $\textbf{S}$ampler ($\textbf{MDNS}$), a novel framework for training discrete neural samplers by aligning two path measures through a family of learning objectives, theoretically grounded in the stochastic optimal control of the continuous-time Markov chains. We validate the efficiency and scalability of MDNS through extensive experiments on various distributions with distinct statistical properties, where MDNS learns to accurately sample from the target distributions despite the extremely high problem dimensions and outperforms other learning-based baselines by a large margin. A comprehensive study of ablations and extensions is also provided to demonstrate the efficacy and potential of the proposed framework.
Authors: Tianlong Yu, Lihong Liu, Ziyi Zhou, Fudu Xing, Kailong Wang, Yang Yang
Abstract: The exploitation of 1 day or n day vulnerabilities poses severe threats to networked devices due to massive deployment scales and delayed patching (average Mean Time To Patch exceeds 60 days). Existing defenses, including host based patching and network based filtering, are inadequate due to limited scalability across diverse devices, compatibility issues especially with embedded or legacy systems, and error prone deployment process (manual patch validation). To address these issues, we introduce REFN (Reinforcement Learning From Network), a novel framework that trains Large Language Models (LLMs) to autonomously generate network filters to prevent 1 day or n day exploitations. REFN ensures scalability by uniquely employs Reinforcement Learning (RL) driven by online network rewards instead of traditional Human Feedback (RLHF). REFN guarantees compatibility via unified deployment on edge security gateways (Amazon Eero). REFN provides robustness via online validation using real network traffic. Crucially, REFN addresses three core challenges in training LLMs for exploit prevention: 1) expanding current LLMs limited vulnerability fixing expertise via Agentic RAG based Knowledge Distillation, 2) bridging current LLMs language to network gaps through an RL From VNF Pipeline that translates language context (vulnerability description) into network enforcement, 3) addressing the LLM hallucination and non determinism via the Online Agentic Validation that penalizes erroneous outputs. Evaluated across 22 families of 1 day or n day exploits, REFN demonstrates effectiveness (21.1 percent higher accuracy than alternatives), efficiency (Mean Time To Patch of 3.65 hours) and scalability (easily scale to 10K devices). REFN serves as an initial step toward training LLMs to rapidly prevent massive scale 1 day or n day exploitations.
Authors: Murat Temiz, Vemund Bakken
Abstract: This study proposes an antenna simulation framework powered by graphics processing units (GPUs) based on an open-source electromagnetic (EM) simulation software (gprMax) for machine learning applications of antenna design and optimization. Furthermore, it compares the simulation results with those obtained through commercial EM software. The proposed software framework for machine learning and surrogate model applications will produce antenna data sets consisting of a large number of antenna simulation results using GPUs. Although machine learning methods can attain the optimum solutions for many problems, they are known to be data-hungry and require a great deal of samples for the training stage of the algorithms. However, producing a sufficient number of training samples in EM applications within a limited time is challenging due to the high computational complexity of EM simulations. Therefore, GPUs are utilized in this study to simulate a large number of antennas with predefined or random antenna shape parameters to produce data sets. Moreover, this study also compares various machine learning and deep learning models in terms of antenna parameter estimation performance. This study demonstrates that an entry-level GPU substantially outperforms a high-end CPU in terms of computational performance, while a high-end gaming GPU can achieve around 18 times more computational performance compared to a high-end CPU. Moreover, it is shown that the open-source EM simulation software can deliver similar results to those obtained via commercial software in the simulation of microstrip antennas when the spatial resolution of the simulations is sufficiently fine.
Authors: Kejia Fan, Jianheng Tang, Zhirui Yang, Feijiang Han, Jiaxu Li, Run He, Yajiang Huang, Anfeng Liu, Houbing Herbert Song, Yunhuai Liu, Huiping Zhuang
Abstract: Personalized Federated Learning (PFL) has presented a significant challenge to deliver personalized models to individual clients through collaborative training. Existing PFL methods are often vulnerable to non-IID data, which severely hinders collective generalization and then compromises the subsequent personalization efforts. In this paper, to address this non-IID issue in PFL, we propose an Analytic Personalized Federated Learning (APFL) approach via dual-stream least squares. In our APFL, we use a foundation model as a frozen backbone for feature extraction. Subsequent to the feature extractor, we develop dual-stream analytic models to achieve both collective generalization and individual personalization. Specifically, our APFL incorporates a shared primary stream for global generalization across all clients, and a dedicated refinement stream for local personalization of each individual client. The analytical solutions of our APFL enable its ideal property of heterogeneity invariance, theoretically meaning that each personalized model remains identical regardless of how heterogeneous the data are distributed across all other clients. Empirical results across various datasets also validate the superiority of our APFL over state-of-the-art baselines, with advantages of at least 1.10%-15.45% in accuracy.
Authors: Zhipeng Chen, Xiaobo Qin, Youbin Wu, Yue Ling, Qinghao Ye, Wayne Xin Zhao, Guang Shi
Abstract: Reinforcement learning with verifiable rewards (RLVR), which typically adopts Pass@1 as the reward, has faced the issues in balancing exploration and exploitation, causing policies to prefer conservative actions, converging to a local optimum. Identifying an appropriate reward metric is therefore crucial. Regarding the prior work, although Pass@k has been used in evaluation, its connection to LLM exploration ability in RLVR remains largely overlooked. To investigate this, we first use Pass@k as the reward to train the policy model (i.e., $\textbf{Pass@k Training}$), and observe the improvement on its exploration ability. Next, we derive an analytical solution for the advantage of Pass@k Training, leading to an efficient and effective process. Building on this, our analysis reveals that exploration and exploitation are not inherently conflicting objectives, while they can mutually enhance each other. Moreover, Pass@k Training with analytical derivation essentially involves directly designing the advantage function. Inspired by this, we preliminarily explore the advantage design for RLVR, showing promising results and highlighting a potential future direction.
Authors: Nicolas Lapautre, Maria Marchenko, Carlos Miguel Pati\~no, Xin Zhou
Abstract: Unlocking the potential of transformers on datasets of large physical systems depends on overcoming the quadratic scaling of the attention mechanism. This work explores combining the Erwin architecture with the Native Sparse Attention (NSA) mechanism to improve the efficiency and receptive field of transformer models for large-scale physical systems, addressing the challenge of quadratic attention complexity. We adapt the NSA mechanism for non-sequential data, implement the Erwin NSA model, and evaluate it on three datasets from the physical sciences -- cosmology simulations, molecular dynamics, and air pressure modeling -- achieving performance that matches or exceeds that of the original Erwin model. Additionally, we reproduce the experimental results from the Erwin paper to validate their implementation.
Authors: Dong Xu, Zhangfan Yang, Jenna Xinyi Yao, Shuangbao Song, Zexuan Zhu, Junkai Ji
Abstract: Three-dimensional generative models increasingly drive structure-based drug discovery, yet it remains constrained by the scarce publicly available protein-ligand complexes. Under such data scarcity, almost all existing pipelines struggle to learn transferable geometric priors and consequently overfit to training-set biases. As such, we present IBEX, an Information-Bottleneck-EXplored coarse-to-fine pipeline to tackle the chronic shortage of protein-ligand complex data in structure-based drug design. Specifically, we use PAC-Bayesian information-bottleneck theory to quantify the information density of each sample. This analysis reveals how different masking strategies affect generalization and indicates that, compared with conventional de novo generation, the constrained Scaffold Hopping task endows the model with greater effective capacity and improved transfer performance. IBEX retains the original TargetDiff architecture and hyperparameters for training to generate molecules compatible with the binding pocket; it then applies an L-BFGS optimization step to finely refine each conformation by optimizing five physics-based terms and adjusting six translational and rotational degrees of freedom in under one second. With only these modifications, IBEX raises the zero-shot docking success rate on CBGBench CrossDocked2020-based from 53% to 64%, improves the mean Vina score from $-7.41 kcal mol^{-1}$ to $-8.07 kcal mol^{-1}$, and achieves the best median Vina energy in 57 of 100 pockets versus 3 for the original TargetDiff. IBEX also increases the QED by 25%, achieves state-of-the-art validity and diversity, and markedly reduces extrapolation error.
Authors: Shouju Wang, Yuchen Song, Sheng'en Li, Dongmian Zou
Abstract: Graph anomaly detection (GAD) has become an increasingly important task across various domains. With the rapid development of graph neural networks (GNNs), GAD methods have achieved significant performance improvements. However, fairness considerations in GAD remain largely underexplored. Indeed, GNN-based GAD models can inherit and amplify biases present in training data, potentially leading to unfair outcomes. While existing efforts have focused on developing fair GNNs, most approaches target node classification tasks, where models often rely on simple layer architectures rather than autoencoder-based structures, which are the most widely used architecturs for anomaly detection. To address fairness in autoencoder-based GAD models, we propose \textbf{D}is\textbf{E}ntangled \textbf{C}ounterfactual \textbf{A}dversarial \textbf{F}air (DECAF)-GAD, a framework that alleviates bias while preserving GAD performance. Specifically, we introduce a structural causal model (SCM) to disentangle sensitive attributes from learned representations. Based on this causal framework, we formulate a specialized autoencoder architecture along with a fairness-guided loss function. Through extensive experiments on both synthetic and real-world datasets, we demonstrate that DECAF-GAD not only achieves competitive anomaly detection performance but also significantly enhances fairness metrics compared to baseline GAD methods. Our code is available at https://github.com/Tlhey/decaf_code.
Authors: Yu-Heng Hung, Ping-Chun Hsieh, Kai Wang
Abstract: Online restless multi-armed bandits (RMABs) typically assume that each arm follows a stationary Markov Decision Process (MDP) with fixed state transitions and rewards. However, in real-world applications like healthcare and recommendation systems, these assumptions often break due to non-stationary dynamics, posing significant challenges for traditional RMAB algorithms. In this work, we specifically consider $N$-armd RMAB with non-stationary transition constrained by bounded variation budgets $B$. Our proposed \rmab\; algorithm integrates sliding window reinforcement learning (RL) with an upper confidence bound (UCB) mechanism to simultaneously learn transition dynamics and their variations. We further establish that \rmab\; achieves $\widetilde{\mathcal{O}}(N^2 B^{\frac{1}{4}} T^{\frac{3}{4}})$ regret bound by leveraging a relaxed definition of regret, providing a foundational theoretical framework for non-stationary RMAB problems for the first time.
Authors: Thore Wietzke, Knut Graichen
Abstract: Gaussian Processes (GPs) are widely used for regression and system identification due to their flexibility and ability to quantify uncertainty. However, their computational complexity limits their applicability to small datasets. Moreover in a streaming scenario, more and more datapoints accumulate which is intractable even for Sparse GPs. Online GPs aim to alleviate this problem by e.g. defining a maximum budget of datapoints and removing redundant datapoints. This work provides a unified comparison of several reduction criteria, analyzing both their computational complexity and reduction behavior. The criteria are evaluated on benchmark functions and real-world datasets, including dynamic system identification tasks. Additionally, acceptance criteria are proposed to further filter out redundant datapoints. This work yields practical guidelines for choosing a suitable criterion for an online GP algorithm.
Authors: Parsa Omidi, Xingshuai Huang, Axel Laborieux, Bahareh Nikpour, Tianyu Shi, Armaghan Eshaghi
Abstract: Memory is fundamental to intelligence, enabling learning, reasoning, and adaptability across biological and artificial systems. While Transformer architectures excel at sequence modeling, they face critical limitations in long-range context retention, continual learning, and knowledge integration. This review presents a unified framework bridging neuroscience principles, including dynamic multi-timescale memory, selective attention, and consolidation, with engineering advances in Memory-Augmented Transformers. We organize recent progress through three taxonomic dimensions: functional objectives (context extension, reasoning, knowledge integration, adaptation), memory representations (parameter-encoded, state-based, explicit, hybrid), and integration mechanisms (attention fusion, gated control, associative retrieval). Our analysis of core memory operations (reading, writing, forgetting, and capacity management) reveals a shift from static caches toward adaptive, test-time learning systems. We identify persistent challenges in scalability and interference, alongside emerging solutions including hierarchical buffering and surprise-gated updates. This synthesis provides a roadmap toward cognitively-inspired, lifelong-learning Transformer architectures.
Authors: Robin Staab, Nikola Jovanovi\'c, Kimberly Mai, Prakhar Ganesh, Martin Vechev, Ferdinando Fioretto, Matthew Jagielski
Abstract: Data minimization (DM) describes the principle of collecting only the data strictly necessary for a given task. It is a foundational principle across major data protection regulations like GDPR and CPRA. Violations of this principle have substantial real-world consequences, with regulatory actions resulting in fines reaching hundreds of millions of dollars. Notably, the relevance of data minimization is particularly pronounced in machine learning (ML) applications, which typically rely on large datasets, resulting in an emerging research area known as Data Minimization in Machine Learning (DMML). At the same time, existing work on other ML privacy and security topics often addresses concerns relevant to DMML without explicitly acknowledging the connection. This disconnect leads to confusion among practitioners, complicating their efforts to implement DM principles and interpret the terminology, metrics, and evaluation criteria used across different research communities. To address this gap, our work introduces a comprehensive framework for DMML, including a unified data pipeline, adversaries, and points of minimization. This framework allows us to systematically review the literature on data minimization and \emph{DM-adjacent} methodologies, for the first time presenting a structured overview designed to help practitioners and researchers effectively apply DM principles. Our work facilitates a unified DM-centric understanding and broader adoption of data minimization strategies in AI/ML.
Authors: Ari Karchmer, Seth Neel, Martin Pawelczyk
Abstract: Data attribution methods aim to answer useful counterfactual questions like "what would a ML model's prediction be if it were trained on a different dataset?" However, estimation of data attribution models through techniques like empirical influence or "datamodeling" remains very computationally expensive. This causes a critical trust issue: if only a few computationally rich parties can obtain data attributions, how can resource-constrained parties trust that the provided attributions are indeed "good," especially when they are used for important downstream applications (e.g., data pricing)? In this paper, we address this trust issue by proposing an interactive verification paradigm for data attribution. An untrusted and computationally powerful Prover learns data attributions, and then engages in an interactive proof with a resource-constrained Verifier. Our main result is a protocol that provides formal completeness, soundness, and efficiency guarantees in the sense of Probably-Approximately-Correct (PAC) verification. Specifically, if both Prover and Verifier follow the protocol, the Verifier accepts data attributions that are {\epsilon}-close to the optimal data attributions (in terms of the Mean Squared Error) with probability 1-{\delta}. Conversely, if the Prover arbitrarily deviates from the protocol, even with infinite compute, then this is detected (or it still yields data attributions to the Verifier) except with probability {\delta}. Importantly, our protocol ensures the Verifier's workload, measured by the number of independent model retrainings it must perform, scales only as O(1/{\epsilon}); i.e., independently of the dataset size. At a technical level, our results apply to efficiently verifying any linear function over the boolean hypercube computed by the Prover, making them broadly applicable to various attribution tasks.
Authors: Haydn Thomas Jones, Natalie Maus, Josh Magnus Ludan, Maggie Ziyu Huan, Jiaming Liang, Marcelo Der Torossian Torres, Jiatao Liang, Zachary Ives, Yoseph Barash, Cesar de la Fuente-Nunez, Jacob R. Gardner, Mark Yatskar
Abstract: AI-driven discovery can greatly reduce design time and enhance new therapeutics' effectiveness. Models using simulators explore broad design spaces but risk violating implicit constraints due to a lack of experimental priors. For example, in a new analysis we performed on a diverse set of models on the GuacaMol benchmark using supervised classifiers, over 60\% of molecules proposed had high probability of being mutagenic. In this work, we introduce \ourdataset, a dataset of priors for design problems extracted from literature describing compounds used in lab settings. It is constructed with LLM pipelines for discovering therapeutic entities in relevant paragraphs and summarizing information in concise fair-use facts. \ourdataset~ consists of 32.3 million pairs of natural language facts, and appropriate entity representations (i.e. SMILES or refseq IDs). To demonstrate the potential of the data, we train LLM, CLIP, and LLava architectures to reason jointly about text and design targets and evaluate on tasks from the Therapeutic Data Commons (TDC). \ourdataset~is highly effective for creating models with strong priors: in supervised prediction problems that use our data as pretraining, our best models with 15M learnable parameters outperform larger 2B TxGemma on both regression and classification TDC tasks, and perform comparably to 9B models on average. Models built with \ourdataset~can be used as constraints while optimizing for novel molecules in GuacaMol, resulting in proposals that are safer and nearly as effective. We release our dataset at \href{https://huggingface.co/datasets/medexanon/Medex}{huggingface.co/datasets/medexanon/Medex}, and will provide expanded versions as available literature grows.
Authors: Alexandre Brown, Glen Berseth
Abstract: Visual reinforcement learning (RL) is challenging due to the need to learn both perception and actions from high-dimensional inputs and noisy rewards. Although large perception models exist, integrating them effectively into RL for visual generalization and improved sample efficiency remains unclear. We propose SegDAC, a Segmentation-Driven Actor-Critic method. SegDAC uses Segment Anything (SAM) for object-centric decomposition and YOLO-World to ground segments semantically via text prompts. It includes a novel transformer-based architecture that supports a dynamic number of segments at each time step and effectively learns which segments to focus on using online RL, without using human labels. By evaluating SegDAC over a challenging visual generalization benchmark using Maniskill3, which covers diverse manipulation tasks under strong visual perturbations, we demonstrate that SegDAC achieves significantly better visual generalization, doubling prior performance on the hardest setting and matching or surpassing prior methods in sample efficiency across all evaluated tasks.
Authors: Lovedeep Gondara, Gregory Arbour, Raymond Ng, Jonathan Simkin, Shebnum Devji
Abstract: Automating data extraction from clinical documents offers significant potential to improve efficiency in healthcare settings, yet deploying Natural Language Processing (NLP) solutions presents practical challenges. Drawing upon our experience implementing various NLP models for information extraction and classification tasks at the British Columbia Cancer Registry (BCCR), this paper shares key lessons learned throughout the project lifecycle. We emphasize the critical importance of defining problems based on clear business objectives rather than solely technical accuracy, adopting an iterative approach to development, and fostering deep interdisciplinary collaboration and co-design involving domain experts, end-users, and ML specialists from inception. Further insights highlight the need for pragmatic model selection (including hybrid approaches and simpler methods where appropriate), rigorous attention to data quality (representativeness, drift, annotation), robust error mitigation strategies involving human-in-the-loop validation and ongoing audits, and building organizational AI literacy. These practical considerations, generalizable beyond cancer registries, provide guidance for healthcare organizations seeking to successfully implement AI/NLP solutions to enhance data management processes and ultimately improve patient care and public health outcomes.
Authors: Zheng Jie Wong, Bingquan Shen
Abstract: Currently, Automatic Speech Recognition (ASR) models are deployed in an extensive range of applications. However, recent studies have demonstrated the possibility of adversarial attack on these models which could potentially suppress or disrupt model output. We investigate and verify the robustness of these attacks and explore if it is possible to increase their imperceptibility. We additionally find that by relaxing the optimisation objective from complete suppression to partial suppression, we can further decrease the imperceptibility of the attack. We also explore possible defences against these attacks and show a low-pass filter defence could potentially serve as an effective defence.
Authors: Rui Zhou, Haohui Ma, Tianle Xin, Lixin Zou, Qiuyue Hu, Hongxi Cheng, Mingzhi Lin, Jingjing Guo, Sheng Wang, Guoqing Zhang, Yanjie Wei, Liangzhen Zheng
Abstract: The rapid expansion of enzyme kinetics literature has outpaced the curation capabilities of major biochemical databases, creating a substantial barrier to AI-driven modeling and knowledge discovery. We present zERExtractor, an automated and extensible platform for comprehensive extraction of enzyme-catalyzed reaction and activity data from scientific literature. zERExtractor features a unified, modular architecture that supports plug-and-play integration of state-of-the-art models, including large language models (LLMs), as interchangeable components, enabling continuous system evolution alongside advances in AI. Our pipeline combines domain-adapted deep learning, advanced OCR, semantic entity recognition, and prompt-driven LLM modules, together with human expert corrections, to extract kinetic parameters (e.g., kcat, Km), enzyme sequences, substrate SMILES, experimental conditions, and molecular diagrams from heterogeneous document formats. Through active learning strategies integrating AI-assisted annotation, expert validation, and iterative refinement, the system adapts rapidly to new data sources. We also release a large benchmark dataset comprising over 1,000 annotated tables and 5,000 biological fields from 270 P450-related enzymology publications. Benchmarking demonstrates that zERExtractor consistently outperforms existing baselines in table recognition (Acc 89.9%), molecular image interpretation (up to 99.1%), and relation extraction (accuracy 94.2%). zERExtractor bridges the longstanding data gap in enzyme kinetics with a flexible, plugin-ready framework and high-fidelity extraction, laying the groundwork for future AI-powered enzyme modeling and biochemical knowledge discovery.
Authors: Yuzhuo Xiao, Zeyu Han, Yuhan Wang, Huaizu Jiang
Abstract: The rapid spread of multimodal misinformation on social media calls for more effective and robust detection methods. Recent advances leveraging multimodal large language models (MLLMs) have shown the potential in addressing this challenge. However, it remains unclear exactly where the bottleneck of existing approaches lies (evidence retrieval v.s. reasoning), hindering the further advances in this field. On the dataset side, existing benchmarks either contain outdated events, leading to evaluation bias due to discrepancies with contemporary social media scenarios as MLLMs can simply memorize these events, or artificially synthetic, failing to reflect real-world misinformation patterns. Additionally, it lacks comprehensive analyses of MLLM-based model design strategies. To address these issues, we introduce XFacta, a contemporary, real-world dataset that is better suited for evaluating MLLM-based detectors. We systematically evaluate various MLLM-based misinformation detection strategies, assessing models across different architectures and scales, as well as benchmarking against existing detection methods. Building on these analyses, we further enable a semi-automatic detection-in-the-loop framework that continuously updates XFacta with new content to maintain its contemporary relevance. Our analysis provides valuable insights and practices for advancing the field of multimodal misinformation detection. The code and data have been released.
Authors: Chenhao Xue, Yuanzhe Jin, Adrian Carrasco-Revilla, Joyraj Chakraborty, Min Chen
Abstract: When developing text classification models for real world applications, one major challenge is the difficulty to collect sufficient data for all text classes. In this work, we address this challenge by utilizing large language models (LLMs) to generate synthetic data and using such data to improve the performance of the models without waiting for more real data to be collected and labelled. As an LLM generates different synthetic data in response to different input examples, we formulate an automated workflow, which searches for input examples that lead to more ``effective'' synthetic data for improving the model concerned. We study three search strategies with an extensive set of experiments, and use experiment results to inform an ensemble algorithm that selects a search strategy according to the characteristics of a class. Our further experiments demonstrate that this ensemble approach is more effective than each individual strategy in our automated workflow for improving classification models using LLMs.
Authors: Andr\'es Carvallo, Denis Parra, Peter Brusilovsky, Hernan Valdivieso, Gabriel Rada, Ivania Donoso, Vladimir Araujo
Abstract: The attention mechanism is a core component of the Transformer architecture. Beyond improving performance, attention has been proposed as a mechanism for explainability via attention weights, which are associated with input features (e.g., tokens in a document). In this context, larger attention weights may imply more relevant features for the model's prediction. In evidence-based medicine, such explanations could support physicians' understanding and interaction with AI systems used to categorize biomedical literature. However, there is still no consensus on whether attention weights provide helpful explanations. Moreover, little research has explored how visualizing attention affects its usefulness as an explanation aid. To bridge this gap, we conducted a user study to evaluate whether attention-based explanations support users in biomedical document classification and whether there is a preferred way to visualize them. The study involved medical experts from various disciplines who classified articles based on study design (e.g., systematic reviews, broad synthesis, randomized and non-randomized trials). Our findings show that the Transformer model (XLNet) classified documents accurately; however, the attention weights were not perceived as particularly helpful for explaining the predictions. However, this perception varied significantly depending on how attention was visualized. Contrary to Munzner's principle of visual effectiveness, which favors precise encodings like bar length, users preferred more intuitive formats, such as text brightness or background color. While our results do not confirm the overall utility of attention weights for explanation, they suggest that their perceived helpfulness is influenced by how they are visually presented.
Authors: Antonio Leandro Martins Candido, Jose Everardo Bessa Maia
Abstract: The automatic curation of discussion forums in online courses requires constant updates, making frequent retraining of Large Language Models (LLMs) a resource-intensive process. To circumvent the need for costly fine-tuning, this paper proposes and evaluates the use of Bayesian fusion. The approach combines the multidimensional classification scores of a pre-trained generic LLM with those of a classifier trained on local data. The performance comparison demonstrated that the proposed fusion improves the results compared to each classifier individually, and is competitive with the LLM fine-tuning approach
Authors: Rodrigo Tertulino
Abstract: Federated Learning (FL) presents a groundbreaking approach for collaborative health research, allowing model training on decentralized data while safeguarding patient privacy. FL offers formal security guarantees when combined with Differential Privacy (DP). The integration of these technologies, however, introduces a significant trade-off between privacy and clinical utility, a challenge further complicated by the severe class imbalance often present in medical datasets. The research presented herein addresses these interconnected issues through a systematic, multi-stage analysis. An FL framework was implemented for cardiovascular risk prediction, where initial experiments showed that standard methods struggled with imbalanced data, resulting in a recall of zero. To overcome such a limitation, we first integrated the hybrid Synthetic Minority Over-sampling Technique with Tomek Links (SMOTETomek) at the client level, successfully developing a clinically useful model. Subsequently, the framework was optimized for non-IID data using a tuned FedProx algorithm. Our final results reveal a clear, non-linear trade-off between the privacy budget (epsilon) and model recall, with the optimized FedProx consistently out-performing standard FedAvg. An optimal operational region was identified on the privacy-utility frontier, where strong privacy guarantees (with epsilon 9.0) can be achieved while maintaining high clinical utility (recall greater than 77%). Ultimately, our study provides a practical methodological blueprint for creating effective, secure, and accurate diagnostic tools that can be applied to real-world, heterogeneous healthcare data.
Authors: Silvia Garc\'ia-M\'endez, Francisco de Arriba-P\'erez
Abstract: Among the many challenges mothers undergo after childbirth, postpartum depression (PPD) is a severe condition that significantly impacts their mental and physical well-being. Consequently, the rapid detection of ppd and their associated risk factors is critical for in-time assessment and intervention through specialized prevention procedures. Accordingly, this work addresses the need to help practitioners make decisions with the latest technological advancements to enable real-time screening and treatment recommendations. Mainly, our work contributes to an intelligent PPD screening system that combines Natural Language Processing, Machine Learning (ML), and Large Language Models (LLMs) towards an affordable, real-time, and non-invasive free speech analysis. Moreover, it addresses the black box problem since the predictions are described to the end users thanks to the combination of LLMs with interpretable ml models (i.e., tree-based algorithms) using feature importance and natural language. The results obtained are 90 % on ppd detection for all evaluation metrics, outperforming the competing solutions in the literature. Ultimately, our solution contributes to the rapid detection of PPD and their associated risk factors, critical for in-time and proper assessment and intervention.
Authors: Kai Zhao, Yanjun Zhao, Jiaming Song, Shien He, Lusheng Zhang, Qiang Zhang, Tianjiao Li
Abstract: Large language models (LLMs) empowered by chain-of-thought reasoning have achieved impressive accuracy on complex tasks but suffer from excessive inference costs and latency when applied uniformly to all problems. We propose SABER (Switchable and Balanced Training for Efficient LLM Reasoning), a reinforcement learning framework that endows LLMs with user-controllable, token-budgeted reasoning. SABER first profiles each training example's base-model thinking token usage and assigns it to one of the predefined budget tiers. During fine-tuning, the model is guided by system prompts and length-aware rewards to respect its assigned budget. In parallel, we incorporate no-think examples to ensure the model remains reliable even when explicit reasoning is turned off. SABER further supports four discrete inference modes - NoThink, FastThink, CoreThink, and DeepThink, enabling flexible trade-offs between latency and reasoning depth. Extensive evaluations on math reasoning (MATH, GSM8K), code generation (MBPP), and logical reasoning (LiveBench-Reasoning) demonstrate that SABER achieves high accuracy under tight budgets, graceful degradation, and effective cross-scale and cross-domain generalization. In particular, SABER-FastThink cuts reasoning length by 65.4% and yields a 3.6% accuracy gain compared with the base model on the MATH benchmark.
Authors: Xiao Fu, Hossein A. Rahmani, Bin Wu, Jerome Ramos, Emine Yilmaz, Aldo Lipani
Abstract: Personalised text generation is essential for user-centric information systems, yet most evaluation methods overlook the individuality of users. We introduce \textbf{PREF}, a \textbf{P}ersonalised \textbf{R}eference-free \textbf{E}valuation \textbf{F}ramework that jointly measures general output quality and user-specific alignment without requiring gold personalised references. PREF operates in a three-step pipeline: (1) a coverage stage uses a large language model (LLM) to generate a comprehensive, query-specific guideline covering universal criteria such as factuality, coherence, and completeness; (2) a preference stage re-ranks and selectively augments these factors using the target user's profile, stated or inferred preferences, and context, producing a personalised evaluation rubric; and (3) a scoring stage applies an LLM judge to rate candidate answers against this rubric, ensuring baseline adequacy while capturing subjective priorities. This separation of coverage from preference improves robustness, transparency, and reusability, and allows smaller models to approximate the personalised quality of larger ones. Experiments on the PrefEval benchmark, including implicit preference-following tasks, show that PREF achieves higher accuracy, better calibration, and closer alignment with human judgments than strong baselines. By enabling scalable, interpretable, and user-aligned evaluation, PREF lays the groundwork for more reliable assessment and development of personalised language generation systems.
Authors: Juvenal Bassa, Vidya Manian, Sudhir Malik, Arghya Chattopadhyay
Abstract: Jet classification in high-energy particle physics is important for understanding fundamental interactions and probing phenomena beyond the Standard Model. Jets originate from the fragmentation and hadronization of quarks and gluons, and pose a challenge for identification due to their complex, multidimensional structure. Traditional classification methods often fall short in capturing these intricacies, necessitating advanced machine learning approaches. In this paper, we employ two neural networks simultaneously as an ensemble to tag various jet types. We convert the jet data to two-dimensional histograms instead of representing them as points in a higher-dimensional space. Specifically, this ensemble approach, hereafter referred to as Ensemble Model, is used to tag jets into classes from the JetNet dataset, corresponding to: Top Quarks, Light Quarks (up or down), and W and Z bosons. For the jet classes mentioned above, we show that the Ensemble Model can be used for both binary and multi-categorical classification. This ensemble approach learns jet features by leveraging the strengths of each constituent network achieving superior performance compared to either individual network.
Authors: Varsha Sen, Biswash Basnet
Abstract: False Data Injection Attacks (FDIAs) pose a significant threat to smart grid infrastructures, particularly Home Area Networks (HANs), where real-time monitoring and control are highly adopted. Owing to the comparatively less stringent security controls and widespread availability of HANs, attackers view them as an attractive entry point to manipulate aggregated demand patterns, which can ultimately propagate and disrupt broader grid operations. These attacks undermine the integrity of smart meter data, enabling malicious actors to manipulate consumption values without activating conventional alarms, thereby creating serious vulnerabilities across both residential and utility-scale infrastructures. This paper presents a machine learning-based framework for both the detection and classification of FDIAs using residential energy data. A real-time detection is provided by the lightweight Artificial Neural Network (ANN), which works by using the most vital features of energy consumption, cost, and time context. For the classification of different attack types, a Bidirectional LSTM is trained to recognize normal, trapezoidal, and sigmoid attack shapes through learning sequential dependencies in the data. A synthetic time-series dataset was generated to emulate realistic household behaviour. Experimental results demonstrate that the proposed models are effective in identifying and classifying FDIAs, offering a scalable solution for enhancing grid resilience at the edge. This work contributes toward building intelligent, data-driven defence mechanisms that strengthen smart grid cybersecurity from residential endpoints.
Authors: Dong Zhao, Yadong Wang, Xiang Chen, Chenxi Wang, Hongliang Dai, Chuanxing Geng, Shengzhong Zhang, Shaoyuan Li, Sheng-Jun Huang
Abstract: Large Language Models (LLMs) show remarkable potential for few-shot information extraction (IE), yet their performance is highly sensitive to the choice of in-context examples. Conventional selection strategies often fail to provide informative guidance, as they overlook a key source of model fallibility: confusion stemming not just from semantic content, but also from the generation of well-structured formats required by IE tasks. To address this, we introduce Active Prompting for Information Extraction (APIE), a novel active prompting framework guided by a principle we term introspective confusion. Our method empowers an LLM to assess its own confusion through a dual-component uncertainty metric that uniquely quantifies both Format Uncertainty (difficulty in generating correct syntax) and Content Uncertainty (inconsistency in extracted semantics). By ranking unlabeled data with this comprehensive score, our framework actively selects the most challenging and informative samples to serve as few-shot exemplars. Extensive experiments on four benchmarks show that our approach consistently outperforms strong baselines, yielding significant improvements in both extraction accuracy and robustness. Our work highlights the critical importance of a fine-grained, dual-level view of model uncertainty when it comes to building effective and reliable structured generation systems.
Authors: Alokesh Manna, Sujit K. Ghosh
Abstract: We develop a Bayesian framework for variable selection in linear regression with autocorrelated errors, accommodating lagged covariates and autoregressive structures. This setting occurs in time series applications where responses depend on contemporaneous or past explanatory variables and persistent stochastic shocks, including financial modeling, hydrological forecasting, and meteorological applications requiring temporal dependency capture. Our methodology uses hierarchical Bayesian models with spike-and-slab priors to simultaneously select relevant covariates and lagged error terms. We propose an efficient two-stage MCMC algorithm separating sampling of variable inclusion indicators and model parameters to address high-dimensional computational challenges. Theoretical analysis establishes posterior selection consistency under mild conditions, even when candidate predictors grow exponentially with sample size, common in modern time series with many potential lagged variables. Through simulations and real applications (groundwater depth prediction, S&P 500 log returns modeling), we demonstrate substantial gains in variable selection accuracy and predictive performance. Compared to existing methods, our framework achieves lower MSPE, improved true model component identification, and greater robustness with autocorrelated noise, underscoring practical utility for model interpretation and forecasting in autoregressive settings.
Authors: Xia Chen
Abstract: The computational capabilities of a neural network are widely assumed to be determined by its static architecture. Here we challenge this view by establishing that a fixed neural structure can operate in fundamentally different computational modes, driven not by its structure but by the temporal dynamics of its input signals. We term this principle 'Dynamical Alignment'. Applying this principle offers a novel resolution to the long-standing paradox of why brain-inspired spiking neural networks (SNNs) underperform. By encoding static input into controllable dynamical trajectories, we uncover a bimodal optimization landscape with a critical phase transition governed by phase space volume dynamics. A 'dissipative' mode, driven by contracting dynamics, achieves superior energy efficiency through sparse temporal codes. In contrast, an 'expansive' mode, driven by expanding dynamics, unlocks the representational power required for SNNs to match or even exceed their artificial neural network counterparts on diverse tasks, including classification, reinforcement learning, and cognitive integration. We find this computational advantage emerges from a timescale alignment between input dynamics and neuronal integration. This principle, in turn, offers a unified, computable perspective on long-observed dualities in neuroscience, from stability-plasticity dilemma to segregation-integration dynamic. It demonstrates that computation in both biological and artificial systems can be dynamically sculpted by 'software' on fixed 'hardware', pointing toward a potential paradigm shift for AI research: away from designing complex static architectures and toward mastering adaptive, dynamic computation principles.
Authors: Ruofan Lu, Yintong Huo, Meng Zhang, Yichen Li, Michael R. Lyu
Abstract: The rapid advancement of large language models (LLMs) has led to the widespread adoption of AI-powered coding assistants integrated into a development environment. On one hand, low-latency code completion offers completion suggestions but is fundamentally constrained to the cursor's current position. On the other hand, chat-based editing can perform complex modifications, yet forces developers to stop their work, describe the intent in natural language, which causes a context-switch away from the code. This creates a suboptimal user experience, as neither paradigm proactively predicts the developer's next edit in a sequence of related edits. To bridge this gap and provide the seamless code edit suggestion, we introduce the task of Next Edit Prediction, a novel task designed to infer developer intent from recent interaction history to predict both the location and content of the subsequent edit. Specifically, we curate a high-quality supervised fine-tuning dataset and an evaluation benchmark for the Next Edit Prediction task. Then, we conduct supervised fine-tuning on a series of models and performed a comprehensive evaluation of both the fine-tuned models and other baseline models, yielding several novel findings. This work lays the foundation for a new interaction paradigm that proactively collaborate with developers by anticipating their following action, rather than merely reacting to explicit instructions.
Authors: Oriane Sim\'eoni, Huy V. Vo, Maximilian Seitzer, Federico Baldassarre, Maxime Oquab, Cijo Jose, Vasil Khalidov, Marc Szafraniec, Seungeun Yi, Micha\"el Ramamonjisoa, Francisco Massa, Daniel Haziza, Luca Wehrstedt, Jianyuan Wang, Timoth\'ee Darcet, Th\'eo Moutakanni, Leonel Sentana, Claire Roberts, Andrea Vedaldi, Jamie Tolan, John Brandt, Camille Couprie, Julien Mairal, Herv\'e J\'egou, Patrick Labatut, Piotr Bojanowski
Abstract: Self-supervised learning holds the promise of eliminating the need for manual data annotation, enabling models to scale effortlessly to massive datasets and larger architectures. By not being tailored to specific tasks or domains, this training paradigm has the potential to learn visual representations from diverse sources, ranging from natural to aerial images -- using a single algorithm. This technical report introduces DINOv3, a major milestone toward realizing this vision by leveraging simple yet effective strategies. First, we leverage the benefit of scaling both dataset and model size by careful data preparation, design, and optimization. Second, we introduce a new method called Gram anchoring, which effectively addresses the known yet unsolved issue of dense feature maps degrading during long training schedules. Finally, we apply post-hoc strategies that further enhance our models' flexibility with respect to resolution, model size, and alignment with text. As a result, we present a versatile vision foundation model that outperforms the specialized state of the art across a broad range of settings, without fine-tuning. DINOv3 produces high-quality dense features that achieve outstanding performance on various vision tasks, significantly surpassing previous self- and weakly-supervised foundation models. We also share the DINOv3 suite of vision models, designed to advance the state of the art on a wide spectrum of tasks and data by providing scalable solutions for diverse resource constraints and deployment scenarios.
Authors: Nguyen Manh Son, Pham Huu Vang, Nguyen Thi Dung, Nguyen Manh Ha. Ta Thi Thao, Tran Thi Thu Thuy, Phan Minh Giang
Abstract: Cancer is recognized as a complex group of diseases, contributing to the highest global mortality rates, with increasing prevalence and a trend toward affecting younger populations. It is characterized by uncontrolled proliferation of abnormal cells, invasion of adjacent tissues, and metastasis to distant organs. Garcinia cowa, a traditional medicinal plant widely used in Southeast Asia, including Vietnam, is employed to treat fever, cough, indigestion, as a laxative, and for parasitic diseases. Numerous xanthone compounds isolated from this species exhibit a broad spectrum of biological activities, with some showing promise as anti cancer and antimalarial agents. Network pharmacology analysis successfully identified key bioactive compounds Rubraxanthone, Garcinone D, Norcowanin, Cowanol, and Cowaxanthone alongside their primary protein targets (TNF, CTNNB1, SRC, NFKB1, and MTOR), providing critical insights into the molecular mechanisms underlying their anti-cancer effects. The Graph Attention Network algorithm demonstrated superior predictive performance, achieving an R2 of 0.98 and an RMSE of 0.02 after data augmentation, highlighting its accuracy in predicting pIC50 values for xanthone based compounds. Additionally, molecular docking revealed MTOR as a potential target for inducing cytotoxicity in HeLa cancer cells from Garcinia cowa.
Authors: Chiara Zugarini, Cristina Sgattoni, Luca Sgheri
Abstract: Cloud detection is essential for atmospheric retrievals, climate studies, and weather forecasting. We analyze infrared radiances from the Infrared Atmospheric Sounding Interferometer (IASI) onboard Meteorological Operational (MetOp) satellites to classify scenes as clear or cloudy. We apply the Support Vector Machine (SVM) approach, based on kernel methods for non-separable data. In this study, the method is implemented for Cloud Identification (CISVM) to classify the test set using radiances or brightness temperatures, with dimensionality reduction through Principal Component Analysis (PCA) and cloud-sensitive channel selection to focus on the most informative features. Our best configuration achieves 88.30 percent agreement with reference labels and shows strong consistency with cloud masks from the Moderate Resolution Imaging Spectroradiometer (MODIS), with the largest discrepancies in polar regions due to sensor differences. These results demonstrate that CISVM is a robust, flexible, and efficient method for automated cloud classification from infrared radiances, suitable for operational retrievals and future missions such as Far infrared Outgoing Radiation Understanding and Monitoring (FORUM), the ninth European Space Agency Earth Explorer Mission.
Authors: Jyotishka Datta, Nicholas G. Polson
Abstract: Prediction-powered inference (PPI) is a recent framework for valid statistical inference with partially labeled data, combining model-based predictions on a large unlabeled set with bias correction from a smaller labeled subset. We show that PPI can be extended to handle informative labeling by replacing its unweighted bias-correction term with an inverse probability weighted (IPW) version, using the classical Horvitz--Thompson or H\'ajek forms. This connection unites design-based survey sampling ideas with modern prediction-assisted inference, yielding estimators that remain valid when labeling probabilities vary across units. We consider the common setting where the inclusion probabilities are not known but estimated from a correctly specified model. In simulations, the performance of IPW-adjusted PPI with estimated propensities closely matches the known-probability case, retaining both nominal coverage and the variance-reduction benefits of PPI.
Authors: Timon Merk, Saeed Salehi, Richard M. Koehler, Qiming Cui, Maria Olaru, Amelia Hahn, Nicole R. Provenza, Simon Little, Reza Abbasi-Asl, Phil A. Starr, Wolf-Julian Neumann
Abstract: Neural decoding of pathological and physiological states can enable patient-individualized closed-loop neuromodulation therapy. Recent advances in pre-trained large-scale foundation models offer the potential for generalized state estimation without patient-individual training. Here we present a foundation model trained on chronic longitudinal deep brain stimulation recordings spanning over 24 days. Adhering to long time-scale symptom fluctuations, we highlight the extended context window of 30 minutes. We present an optimized pre-training loss function for neural electrophysiological data that corrects for the frequency bias of common masked auto-encoder loss functions due to the 1-over-f power law. We show in a downstream task the decoding of Parkinson's disease symptoms with leave-one-subject-out cross-validation without patient-individual training.
Authors: Jozef Skakala
Abstract: Shelf seas are important for carbon sequestration and carbon cycle, but available in situ, or satellite data for carbon pools in the shelf sea environment are often sparse, or highly uncertain. Alternative can be provided by reanalyses, but these are often expensive to run. We propose to use an ensemble of neural networks (NN) to learn from a coupled physics-biogeochemistry model the relationship between the directly observable variables and carbon pools. We demonstrate for North-West European Shelf (NWES) sea environment, that when the NN trained on a model free run simulation is applied to the NWES reanalysis, it is capable to reproduce the reanalysis outputs for carbon pools. Moreover, unlike the existing NWES reanalysis, the NN ensemble is also capable to provide uncertainty information for the pools. We focus on explainability of the results and demonstrate potential use of the NNs for future climate what-if scenarios. We suggest that model-informed machine learning presents a viable alternative to expensive reanalyses and could complement observational data, wherever they are missing and/or highly uncertain.
Authors: Abdullah Hashmat, Muhammad Arham Mirza, Agha Ali Raza
Abstract: With the widespread adoption of Large Language Models (LLMs) across various applications, it is empirical to ensure their fairness across all user communities. However, most LLMs are trained and evaluated on Western centric data, with little attention paid to low-resource languages and regional contexts. To address this gap, we introduce PakBBQ, a culturally and regionally adapted extension of the original Bias Benchmark for Question Answering (BBQ) dataset. PakBBQ comprises over 214 templates, 17180 QA pairs across 8 categories in both English and Urdu, covering eight bias dimensions including age, disability, appearance, gender, socio-economic status, religious, regional affiliation, and language formality that are relevant in Pakistan. We evaluate multiple multilingual LLMs under both ambiguous and explicitly disambiguated contexts, as well as negative versus non negative question framings. Our experiments reveal (i) an average accuracy gain of 12\% with disambiguation, (ii) consistently stronger counter bias behaviors in Urdu than in English, and (iii) marked framing effects that reduce stereotypical responses when questions are posed negatively. These findings highlight the importance of contextualized benchmarks and simple prompt engineering strategies for bias mitigation in low resource settings.
Authors: Igor Halperin
Abstract: The proliferation of Large Language Models (LLMs) is challenged by hallucinations, critical failure modes where models generate non-factual, nonsensical or unfaithful text. This paper introduces Semantic Divergence Metrics (SDM), a novel lightweight framework for detecting Faithfulness Hallucinations -- events of severe deviations of LLMs responses from input contexts. We focus on a specific implementation of these LLM errors, {confabulations, defined as responses that are arbitrary and semantically misaligned with the user's query. Existing methods like Semantic Entropy test for arbitrariness by measuring the diversity of answers to a single, fixed prompt. Our SDM framework improves upon this by being more prompt-aware: we test for a deeper form of arbitrariness by measuring response consistency not only across multiple answers but also across multiple, semantically-equivalent paraphrases of the original prompt. Methodologically, our approach uses joint clustering on sentence embeddings to create a shared topic space for prompts and answers. A heatmap of topic co-occurances between prompts and responses can be viewed as a quantified two-dimensional visualization of the user-machine dialogue. We then compute a suite of information-theoretic metrics to measure the semantic divergence between prompts and responses. Our practical score, $\mathcal{S}_H$, combines the Jensen-Shannon divergence and Wasserstein distance to quantify this divergence, with a high score indicating a Faithfulness hallucination. Furthermore, we identify the KL divergence KL(Answer $||$ Prompt) as a powerful indicator of \textbf{Semantic Exploration}, a key signal for distinguishing different generative behaviors. These metrics are further combined into the Semantic Box, a diagnostic framework for classifying LLM response types, including the dangerous, confident confabulation.
Authors: Kennon Stewart
Abstract: Machine unlearning work assumes a static, i.i.d training environment that doesn't truly exist. Modern ML pipelines need to learn, unlearn, and predict continuously on production streams of data. We translate the notion of the batch unlearning scenario to the online setting using notions of regret, sample complexity, and deletion capacity. We further tighten regret bounds to a logarithmic $\mathcal{O}(\ln{T})$, a first for a machine unlearning algorithm. And we swap out an expensive Hessian inversion with online variant of L-BFGS optimization, removing a memory footprint that scales linearly with time. Such changes extend the lifespan of an ML model before expensive retraining, making for a more efficient unlearning process.
Authors: Dixon Domfeh, Saeid Safarveisi
Abstract: Traditional models for pricing catastrophe (CAT) bonds struggle to capture the complex, relational data inherent in these instruments. This paper introduces CATNet, a novel framework that applies a geometric deep learning architecture, the Relational Graph Convolutional Network (R-GCN), to model the CAT bond primary market as a graph, leveraging its underlying network structure for spread prediction. Our analysis reveals that the CAT bond market exhibits the characteristics of a scale-free network, a structure dominated by a few highly connected and influential hubs. CATNet demonstrates high predictive performance, significantly outperforming a strong Random Forest benchmark. The inclusion of topological centrality measures as features provides a further, significant boost in accuracy. Interpretability analysis confirms that these network features are not mere statistical artifacts; they are quantitative proxies for long-held industry intuition regarding issuer reputation, underwriter influence, and peril concentration. This research provides evidence that network connectivity is a key determinant of price, offering a new paradigm for risk assessment and proving that graph-based models can deliver both state-of-the-art accuracy and deeper, quantifiable market insights.
Authors: Ethan Gordon, Nishank Kuppa, Rigved Tummala, Sriram Anasuri
Abstract: This project explores emoji prediction from short text sequences using four deep learning architectures: a feed-forward network, CNN, transformer, and BERT. Using the TweetEval dataset, we address class imbalance through focal loss and regularization techniques. Results show BERT achieves the highest overall performance due to its pre-training advantage, while CNN demonstrates superior efficacy on rare emoji classes. This research shows the importance of architecture selection and hyperparameter tuning for sentiment-aware emoji prediction, contributing to improved human-computer interaction.
Authors: Marco Bondaschi, Michael Gastpar
Abstract: We derive a conditional version of the classical regret-capacity theorem. This result can be used in universal prediction to find lower bounds on the minimal batch regret, which is a recently introduced generalization of the average regret, when batches of training data are available to the predictor. As an example, we apply this result to the class of binary memoryless sources. Finally, we generalize the theorem to R\'enyi information measures, revealing a deep connection between the conditional R\'enyi divergence and the conditional Sibson's mutual information.
Authors: Abhijit Sinha, Harishankar Kumar, Mohit Joshi, Hemant Kumar Kathania, Shrikanth Narayanan, Sudarsana Reddy Kadiri
Abstract: Children's speech presents challenges for age and gender classification due to high variability in pitch, articulation, and developmental traits. While self-supervised learning (SSL) models perform well on adult speech tasks, their ability to encode speaker traits in children remains underexplored. This paper presents a detailed layer-wise analysis of four Wav2Vec2 variants using the PFSTAR and CMU Kids datasets. Results show that early layers (1-7) capture speaker-specific cues more effectively than deeper layers, which increasingly focus on linguistic information. Applying PCA further improves classification, reducing redundancy and highlighting the most informative components. The Wav2Vec2-large-lv60 model achieves 97.14% (age) and 98.20% (gender) on CMU Kids; base-100h and large-lv60 models reach 86.05% and 95.00% on PFSTAR. These results reveal how speaker traits are structured across SSL model depth and support more targeted, adaptive strategies for child-aware speech interfaces.
Authors: Chenliang Zhang, Lin Wang, Yuanyuan Lu, Yusheng Qi, Kexin Wang, Peixu Hou, Wenshi Chen
Abstract: This paper describes the solutions of the Dianping-Trust-Safety team for the META CRAG-MM challenge. The challenge requires building a comprehensive retrieval-augmented generation system capable for multi-modal multi-turn question answering. The competition consists of three tasks: (1) answering questions using structured data retrieved from an image-based mock knowledge graph, (2) synthesizing information from both knowledge graphs and web search results, and (3) handling multi-turn conversations that require context understanding and information aggregation from multiple sources. For Task 1, our solution is based on the vision large language model, enhanced by supervised fine-tuning with knowledge distilled from GPT-4.1. We further applied curriculum learning strategies to guide reinforcement learning, resulting in improved answer accuracy and reduced hallucination. For Task 2 and Task 3, we additionally leveraged web search APIs to incorporate external knowledge, enabling the system to better handle complex queries and multi-turn conversations. Our approach achieved 1st place in Task 1 with a significant lead of 52.38\%, and 3rd place in Task 3, demonstrating the effectiveness of the integration of curriculum learning with reinforcement learning in our training pipeline.
Authors: Andrew Bai, Justin Cui, Ruochen Wang, Cho-Jui Hsieh
Abstract: Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performance of a given benchmark. We first extract the concepts/skills from the benchmark, determine whether the benchmark predominantly benefits from similar concepts or skills, and finally select instructions with the most matching concepts/skills. Experiments on 10+ benchmarks validate the effectiveness of our targeted data selection method, showing +0.9\% over the best existing baseline averaged over all benchmarks and +1.5\% on the skill-focused subset. Our findings underscore the importance of recognizing the inherent trade-off within instruction selection, which requires balancing the acquisition of conceptual knowledge against visual skill.
Authors: Tianjun Yuan, Jiaxiang Geng, Pengchao Han, Xianhao Chen, Bing Luo
Abstract: Fine-tuning foundation models is critical for superior performance on personalized downstream tasks, compared to using pre-trained models. Collaborative learning can leverage local clients' datasets for fine-tuning, but limited client data and heterogeneous data distributions hinder effective collaboration. To address the challenge, we propose a flexible personalized federated learning paradigm that enables clients to engage in collaborative learning while maintaining personalized objectives. Given the limited and heterogeneous computational resources available on clients, we introduce \textbf{flexible personalized split federated learning (FlexP-SFL)}. Based on split learning, FlexP-SFL allows each client to train a portion of the model locally while offloading the rest to a server, according to resource constraints. Additionally, we propose an alignment strategy to improve personalized model performance on global data. Experimental results show that FlexP-SFL outperforms baseline models in personalized fine-tuning efficiency and final accuracy.
Authors: Michael Weiss, Robert Rosenbach, Christian Eggenberger
Abstract: Ranking product recommendations to optimize for a high click-through rate (CTR) or for high conversion, such as add-to-cart rate (ACR) and Order-Submit-Rate (OSR, view-to-purchase conversion) are standard practices in e-commerce. Optimizing for CTR appears like a straightforward choice: Training data (i.e., click data) are simple to collect and often available in large quantities. Additionally, CTR is used far beyond e-commerce, making it a generalist, easily implemented option. ACR and OSR, on the other hand, are more directly linked to a shop's business goals, such as the Gross Merchandise Value (GMV). In this paper, we compare the effects of using either of these objectives using an online A/B test. Among our key findings, we demonstrate that in our shops, optimizing for OSR produces a GMV uplift more than five times larger than when optimizing for CTR, without sacrificing new product discovery. Our results also provide insights into the different feature importances for each of the objectives.
Authors: Juyuan Wang, Rongchen Zhao, Wei Wei, Yufeng Wang, Mo Yu, Jie Zhou, Jin Xu, Liyan Xu
Abstract: Narrative comprehension on long stories and novels has been a challenging domain attributed to their intricate plotlines and entangled, often evolving relations among characters and entities. Given the LLM's diminished reasoning over extended context and high computational cost, retrieval-based approaches remain a pivotal role in practice. However, traditional RAG methods can fall short due to their stateless, single-step retrieval process, which often overlooks the dynamic nature of capturing interconnected relations within long-range context. In this work, we propose ComoRAG, holding the principle that narrative reasoning is not a one-shot process, but a dynamic, evolving interplay between new evidence acquisition and past knowledge consolidation, analogous to human cognition when reasoning with memory-related signals in the brain. Specifically, when encountering a reasoning impasse, ComoRAG undergoes iterative reasoning cycles while interacting with a dynamic memory workspace. In each cycle, it generates probing queries to devise new exploratory paths, then integrates the retrieved evidence of new aspects into a global memory pool, thereby supporting the emergence of a coherent context for the query resolution. Across four challenging long-context narrative benchmarks (200K+ tokens), ComoRAG outperforms strong RAG baselines with consistent relative gains up to 11% compared to the strongest baseline. Further analysis reveals that ComoRAG is particularly advantageous for complex queries requiring global comprehension, offering a principled, cognitively motivated paradigm for retrieval-based long context comprehension towards stateful reasoning. Our code is publicly released at https://github.com/EternityJune25/ComoRAG
Authors: Yan Ting Chok, Soyon Park, Seungheun Baek, Hajung Kim, Junhyun Lee, Jaewoo Kang
Abstract: Medication recommendation is a crucial task for assisting physicians in making timely decisions from longitudinal patient medical records. However, real-world EHR data present significant challenges due to the presence of rarely observed medical entities and incomplete records that may not fully capture the clinical ground truth. While data-driven models trained on longitudinal Electronic Health Records often achieve strong empirical performance, they struggle to generalize under missing or novel conditions, largely due to their reliance on observed co-occurrence patterns. To address these issues, we propose Hierarchical Ontology and Network Refinement for Robust Medication Recommendation (HiRef), a unified framework that combines two complementary structures: (i) the hierarchical semantics encoded in curated medical ontologies, and (ii) refined co-occurrence patterns derived from real-world EHRs. We embed ontology entities in hyperbolic space, which naturally captures tree-like relationships and enables knowledge transfer through shared ancestors, thereby improving generalizability to unseen codes. To further improve robustness, we introduce a prior-guided sparse regularization scheme that refines the EHR co-occurrence graph by suppressing spurious edges while preserving clinically meaningful associations. Our model achieves strong performance on EHR benchmarks (MIMIC-III and MIMIC-IV) and maintains high accuracy under simulated unseen-code settings. Extensive experiments with comprehensive ablation studies demonstrate HiRef's resilience to unseen medical codes, supported by in-depth analyses of the learned sparsified graph structure and medical code embeddings.
Authors: Runqi Qiao, Qiuna Tan, Peiqing Yang, Yanzi Wang, Xiaowan Wang, Enhui Wan, Sitong Zhou, Guanting Dong, Yuchen Zeng, Yida Xu, Jie Wang, Chong Sun, Chen Li, Honggang Zhang
Abstract: Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities across various tasks, but still struggle with complex mathematical reasoning. Existing research primarily focuses on dataset construction and method optimization, often overlooking two critical aspects: comprehensive knowledge-driven design and model-centric data space modeling. In this paper, we introduce We-Math 2.0, a unified system that integrates a structured mathematical knowledge system, model-centric data space modeling, and a reinforcement learning (RL)-based training paradigm to comprehensively enhance the mathematical reasoning abilities of MLLMs. The key contributions of We-Math 2.0 are fourfold: (1) MathBook Knowledge System: We construct a five-level hierarchical system encompassing 491 knowledge points and 1,819 fundamental principles. (2) MathBook-Standard & Pro: We develop MathBook-Standard, a dataset that ensures broad conceptual coverage and flexibility through dual expansion. Additionally, we define a three-dimensional difficulty space and generate 7 progressive variants per problem to build MathBook-Pro, a challenging dataset for robust training. (3) MathBook-RL: We propose a two-stage RL framework comprising: (i) Cold-Start Fine-tuning, which aligns the model with knowledge-oriented chain-of-thought reasoning; and (ii) Progressive Alignment RL, leveraging average-reward learning and dynamic data scheduling to achieve progressive alignment across difficulty levels. (4) MathBookEval: We introduce a comprehensive benchmark covering all 491 knowledge points with diverse reasoning step distributions. Experimental results show that MathBook-RL performs competitively with existing baselines on four widely-used benchmarks and achieves strong results on MathBookEval, suggesting promising generalization in mathematical reasoning.
Authors: Iksoon Jeong, Kyung-Joong Kim, Kang-Hun Ahn
Abstract: Speech enhancement using artificial neural networks aims to remove noise from noisy speech signals while preserving the speech content. However, speech enhancement networks often introduce distortions to the speech signal, referred to as artifacts, which can degrade audio quality. In this work, we propose a post-processing neural network designed to mitigate artifacts introduced by speech enhancement models. Inspired by the analogy of making a `Putt' after an `Approach' in golf, we name our model PuttNet. We demonstrate that alternating between a speech enhancement model and the proposed Putt model leads to improved speech quality, as measured by perceptual quality scores (PESQ), objective intelligibility (STOI), and background noise intrusiveness (CBAK) scores. Furthermore, we illustrate with graphical analysis why this alternating Approach outperforms repeated application of either model alone.
Authors: Dhruv Dosi, Rohit Meena, Param Rajpura, Yogesh Kumar Meena
Abstract: Legacy floor plans, often preserved only as scanned documents, remain essential resources for architecture, urban planning, and facility management in the construction industry. However, the lack of machine-readable floor plans render large-scale interpretation both time-consuming and error-prone. Automated symbol spotting offers a scalable solution by enabling the identification of service key symbols directly from floor plans, supporting workflows such as cost estimation, infrastructure maintenance, and regulatory compliance. This work introduces a labelled Digitised Electrical Layout Plans (DELP) dataset comprising 45 scanned electrical layout plans annotated with 2,450 instances across 34 distinct service key classes. A systematic evaluation framework is proposed using pretrained object detection models for DELP dataset. Among the models benchmarked, YOLOv8 achieves the highest performance with a mean Average Precision (mAP) of 82.5\%. Using YOLOv8, we develop SkeySpot, a lightweight, open-source toolkit for real-time detection, classification, and quantification of electrical symbols. SkeySpot produces structured, standardised outputs that can be scaled up for interoperable building information workflows, ultimately enabling compatibility across downstream applications and regulatory platforms. By lowering dependency on proprietary CAD systems and reducing manual annotation effort, this approach makes the digitisation of electrical layouts more accessible to small and medium-sized enterprises (SMEs) in the construction industry, while supporting broader goals of standardisation, interoperability, and sustainability in the built environment.
Authors: Hanna Herasimchyk, Robin Labryga, Tomislav Prusina
Abstract: We present a multi-head vision transformer approach for multi-label plant species prediction in vegetation plot images, addressing the PlantCLEF 2025 challenge. The task involves training models on single-species plant images while testing on multi-species quadrat images, creating a drastic domain shift. Our methodology leverages a pre-trained DINOv2 Vision Transformer Base (ViT-B/14) backbone with multiple classification heads for species, genus, and family prediction, utilizing taxonomic hierarchies. Key contributions include multi-scale tiling to capture plants at different scales, dynamic threshold optimization based on mean prediction length, and ensemble strategies through bagging and Hydra model architectures. The approach incorporates various inference techniques including image cropping to remove non-plant artifacts, top-n filtering for prediction constraints, and logit thresholding strategies. Experiments were conducted on approximately 1.4 million training images covering 7,806 plant species. Results demonstrate strong performance, making our submission 3rd best on the private leaderboard. Our code is available at https://github.com/geranium12/plant-clef-2025/tree/v1.0.0.
URLs: https://github.com/geranium12/plant-clef-2025/tree/v1.0.0.
Authors: Wei Tian, Jieming Shi, Man Lung Yiu
Abstract: Real-world trajectories are often sparse with low-sampling rates (i.e., long intervals between consecutive GPS points) and misaligned with road networks, yet many applications demand high-quality data for optimal performance. To improve data quality with sparse trajectories as input, we systematically study two related research problems: trajectory recovery on road network, which aims to infer missing points to recover high-sampling trajectories, and map matching, which aims to map GPS points to road segments to determine underlying routes. In this paper, we present efficient methods TRMMA and MMA for accurate trajectory recovery and map matching, respectively, where MMA serves as the first step of TRMMA. In MMA, we carefully formulate a classification task to map a GPS point from sparse trajectories to a road segment over a small candidate segment set, rather than the entire road network. We develop techniques in MMA to generate effective embeddings that capture the patterns of GPS data, directional information, and road segments, to accurately align sparse trajectories to routes. For trajectory recovery, TRMMA focuses on the segments in the route returned by MMA to infer missing points with position ratios on road segments, producing high-sampling trajectories efficiently by avoiding evaluation of all road segments. Specifically, in TRMMA, we design a dual-transformer encoding process to cohesively capture latent patterns in trajectories and routes, and an effective decoding technique to sequentially predict the position ratios and road segments of missing points. We conduct extensive experiments to compare TRMMA and MMA with numerous existing methods for trajectory recovery and map matching, respectively, on 4 large real-world datasets. TRMMA and MMA consistently achieve the best result quality, often by a significant margin.
Authors: Bella Specktor-Fadida, Malte Hoffmann
Abstract: Deep learning segmentation relies heavily on labeled data, but manual labeling is laborious and time-consuming, especially for volumetric images such as brain magnetic resonance imaging (MRI). While recent domain-randomization techniques alleviate the dependency on labeled data by synthesizing diverse training images from label maps, they offer limited anatomical variability when very few label maps are available. Semi-supervised self-training addresses label scarcity by iteratively incorporating model predictions into the training set, enabling networks to learn from unlabeled data. In this work, we combine domain randomization with self-training to train three-dimensional skull-stripping networks using as little as a single labeled example. First, we automatically bin voxel intensities, yielding labels we use to synthesize images for training an initial skull-stripping model. Second, we train a convolutional autoencoder (AE) on the labeled example and use its reconstruction error to assess the quality of brain masks predicted for unlabeled data. Third, we select the top-ranking pseudo-labels to fine-tune the network, achieving skull-stripping performance on out-of-distribution data that approaches models trained with more labeled images. We compare AE-based ranking to consistency-based ranking under test-time augmentation, finding that the AE approach yields a stronger correlation with segmentation accuracy. Our results highlight the potential of combining domain randomization and AE-based quality control to enable effective semi-supervised segmentation from extremely limited labeled data. This strategy may ease the labeling burden that slows progress in studies involving new anatomical structures or emerging imaging techniques.
Authors: Yushi Feng, Junye Du, Yingying Hong, Qifan Wang, Lequan Yu
Abstract: Existing tool-augmented agentic systems are limited in the real world by (i) black-box reasoning steps that undermine trust of decision-making and pose safety risks, (ii) poor multimodal integration, which is inherently critical for healthcare tasks, and (iii) rigid and computationally inefficient agentic pipelines. We introduce PASS (Probabilistic Agentic Supernet Sampling), the first multimodal framework to address these challenges in the context of Chest X-Ray (CXR) reasoning. PASS adaptively samples agentic workflows over a multi-tool graph, yielding decision paths annotated with interpretable probabilities. Given the complex CXR reasoning task with multimodal medical data, PASS leverages its learned task-conditioned distribution over the agentic supernet. Thus, it adaptively selects the most suitable tool at each supernet layer, offering probability-annotated trajectories for post-hoc audits and directly enhancing medical AI safety. PASS also continuously compresses salient findings into an evolving personalized memory, while dynamically deciding whether to deepen its reasoning path or invoke an early exit for efficiency. To optimize a Pareto frontier balancing performance and cost, we design a novel three-stage training procedure, including expert knowledge warm-up, contrastive path-ranking, and cost-aware reinforcement learning. To facilitate rigorous evaluation, we introduce CAB-E, a comprehensive benchmark for multi-step, safety-critical, free-form CXR reasoning. Experiments across various benchmarks validate that PASS significantly outperforms strong baselines in multiple metrics (e.g., accuracy, AUC, LLM-J.) while balancing computational costs, pushing a new paradigm shift towards interpretable, adaptive, and multimodal medical agentic systems.
Authors: Andrea Urgolo, Monika Stipsitz, Helios Sanchis-Alepuz
Abstract: Monitoring the degradation state of Insulated Gate Bipolar Transistor (IGBT) modules is essential for ensuring the reliability and longevity of power electronic systems, especially in safety-critical and high-performance applications. However, direct measurement of key degradation indicators - such as junction temperature, solder fatigue or delamination - remains challenging due to the physical inaccessibility of internal components and the harsh environment. In this context, machine learning-based virtual sensing offers a promising alternative by bridging the gap from feasible sensor placement to the relevant but inaccessible locations. This paper explores the feasibility of estimating the degradation state of solder layers, and the corresponding full temperature maps based on a limited number of physical sensors. Based on synthetic data of a specific degradation mode, we obtain a high accuracy in the estimation of the degraded solder area (1.17% mean absolute error), and are able to reproduce the surface temperature of the IGBT with a maximum relative error of 4.56% (corresponding to an average relative error of 0.37%).
Authors: Michael Poppel, David Bucher, Maximilian Zorn, Nico Kraus, Jonas Stein, Claudia Linnhoff-Popien
Abstract: To leverage the potential computational speedup of quantum computing (QC), research in quantum machine learning (QML) has gained increasing prominence. Angle encoding techniques in QML models have been shown to generate truncated Fourier series, offering asymptotically universal function approximation capabilities. By selecting efficient feature maps (FMs) within quantum circuits, one can leverage the exponential growth of Fourier frequencies for improved approximation. In multi-dimensional settings, additional input dimensions induce further exponential scaling via mixed frequencies. In practice, however, quantum models frequently fail at regression tasks. Through two white-box experiments, we show that such failures can occur even when the relevant frequencies are present, due to an insufficient number of trainable parameters. In order to mitigate the double-exponential parameter growth resulting from double-exponentially growing frequencies, we propose frequency selection and dimensional separation as techniques to constrain the number of parameters, thereby improving trainability. By restricting the QML model to essential frequencies and permitting mixed frequencies only among feature dimensions with known interdependence, we expand the set of tractable problems on current hardware. We demonstrate the reduced parameter requirements by fitting two white-box functions with known frequency spectrum and dimensional interdependencies that could not be fitted with the default methods. The reduced parameter requirements permit us to perform training on a noisy quantum simulator and to demonstrate inference on real quantum hardware.
Authors: Haoran Sun, Daoqi Liu, Hongyu Zhou, Maokun Li, Shenheng Xu, Fan Yang
Abstract: Inverse scattering problems are critical in electromagnetic imaging and medical diagnostics but are challenged by their nonlinearity and diverse measurement scenarios. This paper proposes a physics-informed deep contrast source inversion framework (DeepCSI) for fast and accurate medium reconstruction across various measurement conditions. Inspired by contrast source inversion (CSI) and neural operator methods, a residual multilayer perceptron (ResMLP) is employed to model current distributions in the region of interest under different transmitter excitations, effectively linearizing the nonlinear inverse scattering problem and significantly reducing the computational cost of traditional full-waveform inversion. By modeling medium parameters as learnable tensors and utilizing a hybrid loss function that integrates state equation loss, data equation loss, and total variation regularization, DeepCSI establishes a fully differentiable framework for joint optimization of network parameters and medium properties. Compared with conventional methods, DeepCSI offers advantages in terms of simplicity and universal modeling capabilities for diverse measurement scenarios, including phase-less and multi-frequency observation. Simulations and experiments demonstrate that DeepCSI achieves high-precision, robust reconstruction under full-data, phaseless data, and multifrequency conditions, outperforming traditional CSI methods and providing an efficient and universal solution for complex inverse scattering problems.
Authors: Andrea Gargano, Jasin Machkour, Mimma Nardelli, Enzo Pasquale Scilingo, Michael Muma
Abstract: In Affective Computing, a key challenge lies in reliably linking subjective emotional experiences with objective physiological markers. This preliminary study addresses the issue of reproducibility by identifying physiological features from cardiovascular and electrodermal signals that are associated with continuous self-reports of arousal levels. Using the Continuously Annotated Signal of Emotion dataset, we analyzed 164 features extracted from cardiac and electrodermal signals of 30 participants exposed to short emotion-evoking videos. Feature selection was performed using the Terminating-Random Experiments (T-Rex) method, which performs variable selection systematically controlling a user-defined target False Discovery Rate. Remarkably, among all candidate features, only two electrodermal-derived features exhibited reproducible and statistically significant associations with arousal, achieving a 100\% confirmation rate. These results highlight the necessity of rigorous reproducibility assessments in physiological features selection, an aspect often overlooked in Affective Computing. Our approach is particularly promising for applications in safety-critical environments requiring trustworthy and reliable white box models, such as mental disorder recognition and human-robot interaction systems.
Authors: Amine Tellache, Abdelaziz Amara Korba, Amdjed Mokhtari, Horea Moldovan, Yacine Ghamri-Doudane
Abstract: Effective incident response (IR) is critical for mitigating cyber threats, yet security teams are overwhelmed by alert fatigue, high false-positive rates, and the vast volume of unstructured Cyber Threat Intelligence (CTI) documents. While CTI holds immense potential for enriching security operations, its extensive and fragmented nature makes manual analysis time-consuming and resource-intensive. To bridge this gap, we introduce a novel Retrieval-Augmented Generation (RAG)-based framework that leverages Large Language Models (LLMs) to automate and enhance IR by integrating dynamically retrieved CTI. Our approach introduces a hybrid retrieval mechanism that combines NLP-based similarity searches within a CTI vector database with standardized queries to external CTI platforms, facilitating context-aware enrichment of security alerts. The augmented intelligence is then leveraged by an LLM-powered response generation module, which formulates precise, actionable, and contextually relevant incident mitigation strategies. We propose a dual evaluation paradigm, wherein automated assessment using an auxiliary LLM is systematically cross-validated by cybersecurity experts. Empirical validation on real-world and simulated alerts demonstrates that our approach enhances the accuracy, contextualization, and efficiency of IR, alleviating analyst workload and reducing response latency. This work underscores the potential of LLM-driven CTI fusion in advancing autonomous security operations and establishing a foundation for intelligent, adaptive cybersecurity frameworks.
Authors: Fabian Kresse, Georgios Pilikos, Mario Azcueta, Nicolas Floury
Abstract: Synthetic Aperture Radar (SAR) data enables large-scale surveillance of maritime vessels. However, near-real-time monitoring is currently constrained by the need to downlink all raw data, perform image focusing, and subsequently analyze it on the ground. On-board processing to generate higher-level products could reduce the data volume that needs to be downlinked, alleviating bandwidth constraints and minimizing latency. However, traditional image focusing and processing algorithms face challenges due to the satellite's limited memory, processing power, and computational resources. This work proposes and evaluates neural networks designed for real-time inference on unfocused SAR data acquired in Stripmap and Interferometric Wide (IW) modes captured with Sentinel-1. Our results demonstrate the feasibility of using one of our models for on-board processing and deployment on an FPGA. Additionally, by investigating a binary classification task between ships and windmills, we demonstrate that target classification is possible.
Authors: Wei Shan Lee, I Hang Kwok, Kam Ian Leong, Chi Kiu Althina Chau, Kei Chon Sio
Abstract: Accurate prediction of electronic band structures in two-dimensional materials remains a fundamental challenge, with existing methods struggling to balance computational efficiency and physical accuracy. We present the Symmetry-Constrained Multi-Scale Physics-Informed Neural Network (SCMS-PINN) v35, which directly learns graphene band structures while rigorously enforcing crystallographic symmetries through a multi-head architecture. Our approach introduces three specialized ResNet-6 pathways -- K-head for Dirac physics, M-head for saddle points, and General head for smooth interpolation -- operating on 31 physics-informed features extracted from k-points. Progressive Dirac constraint scheduling systematically increases the weight parameter from 5.0 to 25.0, enabling hierarchical learning from global topology to local critical physics. Training on 10,000 k-points over 300 epochs achieves 99.99\% reduction in training loss (34.597 to 0.003) with validation loss of 0.0085. The model predicts Dirac point gaps within 30.3 $\mu$eV of theoretical zero and achieves average errors of 53.9 meV (valence) and 40.5 meV (conduction) across the Brillouin zone. All twelve C$_{6v}$ operations are enforced through systematic averaging, guaranteeing exact symmetry preservation. This framework establishes a foundation for extending physics-informed learning to broader two-dimensional materials for accelerated discovery.
Authors: Luyao Tang, Kunze Huang, Chaoqi Chen, Yuxuan Yuan, Chenxin Li, Xiaotong Tu, Xinghao Ding, Yue Huang
Abstract: Human perceptual systems excel at inducing and recognizing objects across both known and novel categories, a capability far beyond current machine learning frameworks. While generalized category discovery (GCD) aims to bridge this gap, existing methods predominantly focus on optimizing objective functions. We present an orthogonal solution, inspired by the human cognitive process for novel object understanding: decomposing objects into visual primitives and establishing cross-knowledge comparisons. We propose ConGCD, which establishes primitive-oriented representations through high-level semantic reconstruction, binding intra-class shared attributes via deconstruction. Mirroring human preference diversity in visual processing, where distinct individuals leverage dominant or contextual cues, we implement dominant and contextual consensus units to capture class-discriminative patterns and inherent distributional invariants, respectively. A consensus scheduler dynamically optimizes activation pathways, with final predictions emerging through multiplex consensus integration. Extensive evaluations across coarse- and fine-grained benchmarks demonstrate ConGCD's effectiveness as a consensus-aware paradigm. Code is available at github.com/lytang63/ConGCD.
Authors: Sayan Nag, K J Joseph, Koustava Goswami, Vlad I Morariu, Balaji Vasan Srinivasan
Abstract: Evaluating graphic designs involves assessing it from multiple facets like alignment, composition, aesthetics and color choices. Evaluating designs in a holistic way involves aggregating feedback from individual expert reviewers. Towards this, we propose an Agentic Design Review System (AgenticDRS), where multiple agents collaboratively analyze a design, orchestrated by a meta-agent. A novel in-context exemplar selection approach based on graph matching and a unique prompt expansion method plays central role towards making each agent design aware. Towards evaluating this framework, we propose DRS-BENCH benchmark. Thorough experimental evaluation against state-of-the-art baselines adapted to the problem setup, backed-up with critical ablation experiments brings out the efficacy of Agentic-DRS in evaluating graphic designs and generating actionable feedback. We hope that this work will attract attention to this pragmatic, yet under-explored research direction.
Authors: Adam E. Essex (Loughborough University, England), Natalia B. Janson (Loughborough University, England), Rachel A. Norris (Loughborough University, England), Alexander G. Balanov (Loughborough University, England)
Abstract: Despite explosive expansion of artificial intelligence based on artificial neural networks (ANNs), these are employed as "black boxes'', as it is unclear how, during learning, they form memories or develop unwanted features, including spurious memories and catastrophic forgetting. Much research is available on isolated aspects of learning ANNs, but due to their high dimensionality and non-linearity, their comprehensive analysis remains a challenge. In ANNs, knowledge is thought to reside in connection weights or in attractor basins, but these two paradigms are not linked explicitly. Here we comprehensively analyse mechanisms of memory formation in an 81-neuron Hopfield network undergoing Hebbian learning by revealing bifurcations leading to formation and destruction of attractors and their basin boundaries. We show that, by affecting evolution of connection weights, the applied stimuli induce a pitchfork and then a cascade of saddle-node bifurcations creating new attractors with their basins that can code true or spurious memories, and an abrupt disappearance of old memories (catastrophic forgetting). With successful learning, new categories are represented by the basins of newly born point attractors, and their boundaries by the stable manifolds of new saddles. With this, memorisation and forgetting represent two manifestations of the same mechanism. Our strategy to analyse high-dimensional learning ANNs is universal and applicable to recurrent ANNs of any form. The demonstrated mechanisms of memory formation and of catastrophic forgetting shed light on the operation of a wider class of recurrent ANNs and could aid the development of approaches to mitigate their flaws.
Authors: Youping Gu, Xiaolong Li, Yuhao Hu, Bohan Zhuang
Abstract: Diffusion transformers currently lead the field in high-quality video generation, but their slow iterative denoising process and prohibitive quadratic attention costs for long sequences create significant inference bottlenecks. While both step distillation and sparse attention mechanisms have shown promise as independent acceleration strategies, effectively combining these approaches presents critical challenges -- training-free integration yields suboptimal results, while separately training sparse attention after step distillation requires prohibitively expensive high-quality video data. To overcome these limitations, we propose BLADE, an innovative data-free joint training framework that introduces: (1) an Adaptive Block-Sparse Attention (ASA) mechanism for dynamically generating content-aware sparsity masks to focus computation on salient spatiotemporal features, and (2) a sparsity-aware step distillation paradigm built upon Trajectory Distribution Matching (TDM) that directly incorporates sparsity into the distillation process rather than treating it as a separate compression step, with fast convergence. We validate BLADE on text-to-video models like CogVideoX-5B and Wan2.1-1.3B. Our framework demonstrates remarkable efficiency gains across different scales. On Wan2.1-1.3B, BLADE achieves a 14.10x end-to-end inference acceleration over a 50-step baseline. Moreover, on models such as CogVideoX-5B with short video sequence lengths, our framework delivers a robust 8.89x speedup. Crucially, the acceleration is accompanied by a consistent quality improvement. On the VBench-2.0 benchmark, BLADE boosts the score of CogVideoX-5B to 0.569 (from 0.534) and Wan2.1-1.3B to 0.570 (from 0.563), results that are further corroborated by superior ratings in human evaluations. Our code and model weights are publicly available at: http://ziplab.co/BLADE-Homepage/.
Authors: Xuexin Xu, Siyu Wang, Radhika Joshi, Rihan Hai, Mohammad H. Ansari
Abstract: We present a native three-qubit entangling gate that exploits engineered interactions to realize control-control-target and control-target-target operations in a single coherent step. Unlike conventional decompositions into multiple two-qubit gates, our hybrid optimization approach selectively amplifies desired interactions while suppressing unwanted couplings, yielding robust performance across the computational subspace and beyond. The new gate can be classified as a cross-resonance gate. We show it can be utilized in several ways, for example, in GHZ triplet state preparation, Toffoli-class logic demonstrations with many-body interactions, and in implementing a controlled-ZZ gate. The latter maps the parity of two data qubits directly onto a measurement qubit, enabling faster and higher-fidelity stabilizer measurements in surface-code quantum error correction. In all these examples, we show that the three-qubit gate performance remains robust across Hilbert space sizes, as confirmed by testing under increasing total excitation numbers. This work lays the foundation for co-designing circuit architectures and control protocols that leverage native multiqubit interactions as core elements of next-generation superconducting quantum processors.
Authors: Anand Kumar, Harminder Pal Monga, Tapasi Brahma, Satyam Kalra, Navas Sherif
Abstract: Plant diseases are a major threat to food security globally. It is important to develop early detection systems which can accurately detect. The advancement in computer vision techniques has the potential to solve this challenge. We have developed a mobile-friendly solution which can accurately classify 101 plant diseases across 33 crops. We built a comprehensive dataset by combining different datasets, Plant Doc, PlantVillage, and PlantWild, all of which are for the same purpose. We evaluated performance across several lightweight architectures - MobileNetV2, MobileNetV3, MobileNetV3-Large, and EfficientNet-B0, B1 - specifically chosen for their efficiency on resource-constrained devices. The results were promising, with EfficientNet-B1 delivering our best performance at 94.7% classification accuracy. This architecture struck an optimal balance between accuracy and computational efficiency, making it well-suited for real-world deployment on mobile devices.
Authors: Alexander Plaschzug, Amit Reza, Ludmila Carone, Sebastian Gernjak, Christiane Helling
Abstract: With the development of ever-improving telescopes capable of observing exoplanet atmospheres in greater detail and number, there is a growing demand for enhanced 3D climate models to support and help interpret observational data from space missions like CHEOPS, TESS, JWST, PLATO, and Ariel. However, the computationally intensive and time-consuming nature of general circulation models (GCMs) poses significant challenges in simulating a wide range of exoplanetary atmospheres. This study aims to determine whether machine learning (ML) algorithms can be used to predict the 3D temperature and wind structure of arbitrary tidally-locked gaseous exoplanets in a range of planetary parameters. A new 3D GCM grid with 60 inflated hot Jupiters orbiting A, F, G, K, and M-type host stars modelled with Exorad has been introduced. A dense neural network (DNN) and a decision tree algorithm (XGBoost) are trained on this grid to predict local gas temperatures along with horizontal and vertical winds. To ensure the reliability and quality of the ML model predictions, WASP-121 b, HATS-42 b, NGTS-17 b, WASP-23 b, and NGTS-1 b-like planets, which are all targets for PLATO observation, are selected and modelled with ExoRad and the two ML methods as test cases. The DNN predictions for the gas temperatures are to such a degree that the calculated spectra agree within 32 ppm for all but one planet, for which only one single HCN feature reaches a 100 ppm difference. The developed ML emulators can reliably predict the complete 3D temperature field of an inflated warm to ultra-hot tidally locked Jupiter around A to M-type host stars. It provides a fast tool to complement and extend traditional GCM grids for exoplanet ensemble studies. The quality of the predictions is such that no or minimal effects on the gas phase chemistry, hence on the cloud formation and transmission spectra, are to be expected.
Authors: Jim Dilkes, Vahid Yazdanpanah, Sebastian Stein
Abstract: Large Language Models (LLMs) show potential as sequential decision-making agents, but their application is often limited due to a reliance on large, computationally expensive models. This creates a need to improve smaller models, yet existing post-training methods are designed for single-turn interactions and cannot handle credit assignment in multi-step agentic tasks. To address this, we introduce Multi-Step Group-Relative Policy Optimization (MS-GRPO), a new algorithm for post-training LLM agents, grounded in formal Text-Mediated Stochastic Game (TSMG) and Language-Agent Policy (LAP) frameworks. For credit assignment, MS-GRPO attributes the entire cumulative episode reward to each individual episode step. We supplement this algorithm with a novel absolute-advantage-weighted episode sampling strategy that we show improves training performance. We evaluate our approach by post-training a 3-billion parameter model on Snake and Frozen Lake. Our experiments demonstrate that the method is effective in improving decision-making performance: our post-trained 3B parameter model outperforms a 72B parameter baseline by 50% on the Frozen Lake task. This work demonstrates that targeted post-training is a practical and efficient alternative to relying on model scale for creating sequential decision-making agents using LLMs.
Authors: Viktor Zaverkin, Matheus Ferraz, Francesco Alesiani, Mathias Niepert
Abstract: Universal machine-learned potentials promise transferable accuracy across compositional and vibrational degrees of freedom, yet their application to biomolecular simulations remains underexplored. This work systematically evaluates equivariant message-passing architectures trained on the SPICE-v2 dataset with and without explicit long-range dispersion and electrostatics. We assess the impact of model size, training data composition, and electrostatic treatment across in- and out-of-distribution benchmark datasets, as well as molecular simulations of bulk liquid water, aqueous NaCl solutions, and biomolecules, including alanine tripeptide, the mini-protein Trp-cage, and Crambin. While larger models improve accuracy on benchmark datasets, this trend does not consistently extend to properties obtained from simulations. Predicted properties also depend on the composition of the training dataset. Long-range electrostatics show no systematic impact across systems. However, for Trp-cage, their inclusion yields increased conformational variability. Our results suggest that imbalanced datasets and immature evaluation practices currently challenge the applicability of universal machine-learned potentials to biomolecular simulations.
Authors: Ze Liu, Xianquan Wang, Shuochen Liu, Jie Ma, Huibo Xu, Yupeng Han, Zhe Yang, Kai Zhang, Longfei Li, Jun Zhou
Abstract: Recommender systems heavily rely on implicit feedback, which is inherently noisy due to false positives and negatives, severely degrading recommendation accuracy. Existing denoising strategies often overlook entity-aware modeling, suffer from high computational overhead, or demand excessive hyperparameter tuning, limiting their real-world applicability. We propose CrossDenoise, a novel and lightweight framework that addresses these challenges by disentangling noise estimation into user-, item-, and interaction-specific factors. Leveraging empirical observations that show significant heterogeneity in user and item noise propensities, CrossDenoise computes entity reputation factors (user/item reliability) via a rank-based linear mapping of average training losses. These are fused with interaction-level weights derived from an empirical cumulative distribution function (ECDF) of individual losses. This design is model-agnostic, computationally efficient, and requires only two intuitive hyperparameters. Extensive experiments on ML-1M, Yelp, and Amazon-book datasets, across GMF, NeuMF, and CDAE backbones, demonstrate that CrossDenoise consistently and significantly outperforms state-of-the-art baselines. For instance, it achieves up to 27.01% NDCG@50 gain on Yelp with NeuMF, while incurring negligible computational and memory overhead. Our analysis confirms that CrossDenoise effectively separates clean from noisy samples and remains robust under varied hyperparameter settings. It offers a practical and scalable solution for denoising implicit feedback.
Authors: Tianyi Li, Mingda Chen, Bowei Guo, Zhiqiang Shen
Abstract: Diffusion Language Models (DLMs) are rapidly emerging as a powerful and promising alternative to the dominant autoregressive (AR) paradigm. By generating tokens in parallel through an iterative denoising process, DLMs possess inherent advantages in reducing inference latency and capturing bidirectional context, thereby enabling fine-grained control over the generation process. While achieving a several-fold speed-up, recent advancements have allowed DLMs to show performance comparable to their autoregressive counterparts, making them a compelling choice for various natural language processing tasks. In this survey, we provide a holistic overview of the current DLM landscape. We trace its evolution and relationship with other paradigms, such as autoregressive and masked language models, and cover both foundational principles and state-of-the-art models. Our work offers an up-to-date, comprehensive taxonomy and an in-depth analysis of current techniques, from pre-training strategies to advanced post-training methods. Another contribution of this survey is a thorough review of DLM inference strategies and optimizations, including improvements in decoding parallelism, caching mechanisms, and generation quality. We also highlight the latest approaches to multimodal extensions of DLMs and delineate their applications across various practical scenarios. Furthermore, our discussion addresses the limitations and challenges of DLMs, including efficiency, long-sequence handling, and infrastructure requirements, while outlining future research directions to sustain progress in this rapidly evolving field. Project GitHub is available at https://github.com/VILA-Lab/Awesome-DLMs.
Authors: Johanna D\"ungler, Amartya Sanyal
Abstract: Given $n$ i.i.d. random matrices $A_i \in \mathbb{R}^{d \times d}$ that share a common expectation $\Sigma$, the objective of Differentially Private Stochastic PCA is to identify a subspace of dimension $k$ that captures the largest variance directions of $\Sigma$, while preserving differential privacy (DP) of each individual $A_i$. Existing methods either (i) require the sample size $n$ to scale super-linearly with dimension $d$, even under Gaussian assumptions on the $A_i$, or (ii) introduce excessive noise for DP even when the intrinsic randomness within $A_i$ is small. Liu et al. (2022a) addressed these issues for sub-Gaussian data but only for estimating the top eigenvector ($k=1$) using their algorithm DP-PCA. We propose the first algorithm capable of estimating the top $k$ eigenvectors for arbitrary $k \leq d$, whilst overcoming both limitations above. For $k=1$ our algorithm matches the utility guarantees of DP-PCA, achieving near-optimal statistical error even when $n = \tilde{\!O}(d)$. We further provide a lower bound for general $k > 1$, matching our upper bound up to a factor of $k$, and experimentally demonstrate the advantages of our algorithm over comparable baselines.
Authors: Brooke R. Weborg, Gursel Serpen
Abstract: This paper examines Echo State Network, a reservoir computer, performance using four different benchmark problems, then proposes heuristics or rules of thumb for configuring the architecture, as well as the selection of parameters and their values, which are applicable to problems within the same domain, to help serve to fill the experience gap needed by those entering this field of study. The influence of various parameter selections and their value adjustments, as well as architectural changes made to an Echo State Network, a powerful recurrent neural network configured as a reservoir computer, can be challenging to fully comprehend without experience in the field, and even some hyperparameter optimization algorithms may have difficulty adjusting parameter values without proper manual selections made first. Therefore, it is imperative to understand the effects of parameters and their value selection on Echo State Network architecture performance for a successful build. Thus, to address the requirement for an extensive background in Echo State Network architecture, as well as examine how Echo State Network performance is affected with respect to variations in architecture, design, and parameter selection and values, a series of benchmark tasks representing different problem domains, including time series prediction, pattern generation, chaotic system prediction, and time series classification, were modeled and experimented on to show the impact on the performance of Echo State Network.
Authors: Dongyue Li, Hongyang R. Zhang
Abstract: A widely used algorithm for transfer learning is fine-tuning, where a pre-trained model is fine-tuned on a target task with a small amount of labeled data. When the capacity of the pre-trained model is significantly larger than the size of the target dataset, fine-tuning is prone to overfitting and memorizing the training labels. Hence, a crucial question is to regularize fine-tuning and ensure its robustness against noise. To address this question, we begin by analyzing the generalization properties of fine-tuning. We present a PAC-Bayes generalization bound that depends on the distance traveled in each layer during fine-tuning and the noise stability of the fine-tuned model. We empirically measure these quantities. Based on the analysis, we propose regularized self-labeling -- the interpolation between regularization and self-labeling methods, including (i) layer-wise regularization to constrain the distance traveled in each layer; (ii) self-label-correction and label-reweighting to correct mislabeled data points (that the model is confident) and reweight less confident data points. We validate our approach on an extensive collection of image and text datasets using multiple pre-trained model architectures. Our approach improves baseline methods by 1.76% (on average) for seven image classification tasks and 0.75% for a few-shot classification task. When the target data set includes noisy labels, our approach outperforms baseline methods by an average of 3.56% in two noisy settings.
Authors: Jung-hun Kim, Milan Vojnovic
Abstract: We consider the problem of scheduling in multi-class, parallel-server queuing systems with uncertain rewards from job-server assignments. In this scenario, jobs incur holding costs while awaiting completion, and job-server assignments yield observable stochastic rewards with unknown mean values. The mean rewards for job-server assignments are assumed to follow a bilinear model with respect to features that characterize jobs and servers. Our objective is to minimize regret by maximizing the cumulative reward of job-server assignments over a time horizon, while keeping the total job holding cost bounded to ensure the stability of the queueing system. This problem is motivated by applications requiring resource allocation in network systems. In this problem, it is essential to control the tradeoff between reward maximization and fair allocation for the stability of the underlying queuing system (i.e., maximizing network throughput). To address this problem, we propose a scheduling algorithm based on a weighted proportional fair criteria augmented with marginal costs for reward maximization, incorporating a bandit algorithm tailored for bilinear rewards. Our algorithm achieves a sub-linear regret bound and a sub-linear mean holding cost (and queue length bound) of $\tilde{O}(\sqrt{T})$, respectively, with respect to the time horizon $T$, thus guaranteeing queuing system stability. Additionally, we establish stability conditions for distributed iterative algorithms for computing allocations, which are relevant to large-scale system applications. Finally, we demonstrate the efficiency of our algorithm through numerical experiments.
Authors: Emanuele Sansone, Robin Manhaeve
Abstract: Self-supervised learning excels at learning representations from large amounts of data. At the same time, generative models offer the complementary property of learning information about the underlying data generation process. In this study, we aim at establishing a principled connection between these two paradigms and highlight the benefits of their complementarity. In particular, we perform an analysis of self-supervised learning objectives, elucidating the underlying probabilistic graphical models and presenting a standardized methodology for their derivation from first principles. The analysis suggests a natural means of integrating self-supervised learning with likelihood-based generative models. We instantiate this concept within the realm of cluster-based self-supervised learning and energy models, introducing a lower bound proven to reliably penalize the most important failure modes and unlocking full unification. Our theoretical findings are substantiated through experiments on synthetic and real-world data, including SVHN, CIFAR10, and CIFAR100, demonstrating that our objective function allows to jointly train a backbone network in a discriminative and generative fashion, consequently outperforming existing self-supervised learning strategies in terms of clustering, generation and out-of-distribution detection performance by a wide margin. We also demonstrate that the solution can be integrated into a neuro-symbolic framework to tackle a simple yet non-trivial instantiation of the symbol grounding problem. The code is publicly available at https://github.com/emsansone/GEDI.
Authors: Mohammad Amaz Uddin, Md Mahiuddin, Iqbal H. Sarker
Abstract: Phishing email is a serious cyber threat that tries to deceive users by sending false emails with the intention of stealing confidential information or causing financial harm. Attackers, often posing as trustworthy entities, exploit technological advancements and sophistication to make detection and prevention of phishing more challenging. Despite extensive academic research, phishing detection remains an ongoing and formidable challenge in the cybersecurity landscape. Large Language Models (LLMs) and Masked Language Models (MLMs) possess immense potential to offer innovative solutions to address long-standing challenges. In this research paper, we present an optimized, fine-tuned transformer-based DistilBERT model designed for the detection of phishing emails. In the detection process, we work with a phishing email dataset and utilize the preprocessing techniques to clean and solve the imbalance class issues. Through our experiments, we found that our model effectively achieves high accuracy, demonstrating its capability to perform well. Finally, we demonstrate our fine-tuned model using Explainable-AI (XAI) techniques such as Local Interpretable Model-Agnostic Explanations (LIME) and Transformer Interpret to explain how our model makes predictions in the context of text classification for phishing emails.
Authors: Savelii Chezhegov, Yaroslav Klyukin, Andrei Semenov, Aleksandr Beznosikov, Alexander Gasnikov, Samuel Horv\'ath, Martin Tak\'a\v{c}, Eduard Gorbunov
Abstract: Methods with adaptive stepsizes, such as AdaGrad and Adam, are essential for training modern Deep Learning models, especially Large Language Models. Typically, the noise in the stochastic gradients is heavy-tailed for the later ones. Gradient clipping provably helps to achieve good high-probability convergence for such noises. However, despite the similarity between AdaGrad/Adam and Clip-SGD, the current understanding of the high-probability convergence of AdaGrad/Adam-type methods is limited in this case. In this work, we prove that AdaGrad/Adam (and their delayed version) can have provably bad high-probability convergence if the noise is heavy-tailed. We also show that gradient clipping fixes this issue, i.e., we derive new high-probability convergence bounds with polylogarithmic dependence on the confidence level for AdaGrad-Norm and Adam-Norm with clipping and with/without delay for smooth convex/non-convex stochastic optimization with heavy-tailed noise. We extend our results to the case of AdaGrad/Adam with delayed stepsizes. Our empirical evaluations highlight the superiority of clipped versions of AdaGrad/Adam in handling the heavy-tailed noise.
Authors: Xueping Gong, Wei You, Jiheng Zhang
Abstract: We study contextual dynamic pricing, where a decision maker posts personalized prices based on observable contexts and receives binary purchase feedback indicating whether the customer's valuation exceeds the price. Each valuation is modeled as an unknown latent function of the context, corrupted by independent and identically distributed market noise from an unknown distribution. Relying only on Lipschitz continuity of the noise distribution and bounded valuations, we propose a minimax-optimal algorithm. To accommodate the unknown distribution, our method discretizes the relevant noise range to form a finite set of candidate prices, then applies layered data partitioning to obtain confidence bounds substantially tighter than those derived via the elliptical-potential lemma. A key advantage is that estimation bias in the valuation function cancels when comparing upper confidence bounds, eliminating the need to know the Lipschitz constant. The framework extends beyond linear models to general function classes through offline regression oracles. Our regret analysis depends solely on the oracle's estimation error, typically governed by the statistical complexity of the class. These techniques yield a regret upper bound matching the minimax lower bound up to logarithmic factors. Furthermore, we refine these guarantees under additional structures -- e.g., linear valuation models, second-order smoothness, sparsity, and known noise distribution or observable valuations -- and compare our bounds and assumptions with prior dynamic-pricing methods. Finally, numerical experiments corroborate the theory and show clear improvements over benchmark methods.
Authors: Lakshmi Jayalal, Gokularam Muthukrishnan, Sheetal Kalyani
Abstract: The performance of the standard Online Robust Principal Component Analysis (OR-PCA) technique depends on the optimum tuning of the explicit regularizers and this tuning is dataset sensitive. We aim to remove the dependency on these tuning parameters by using implicit regularization. We propose to use the implicit regularization effect of various modified gradient descents to make OR-PCA tuning free. Our method incorporates three different versions of modified gradient descent that separately but naturally encourage sparsity and low-rank structures in the data. The proposed method performs comparable or better than the tuned OR-PCA for both simulated and real-world datasets. Tuning-free ORPCA makes it more scalable for large datasets since we do not require dataset-dependent parameter tuning.
Authors: Sourav Chatterjee, Timothy Sudijono
Abstract: We show that feedforward neural networks with ReLU activation generalize on low complexity data, suitably defined. Given i.i.d.~data generated from a simple programming language, the minimum description length (MDL) feedforward neural network which interpolates the data generalizes with high probability. We define this simple programming language, along with a notion of description length of such networks. We provide several examples on basic computational tasks, such as checking primality of a natural number. For primality testing, our theorem shows the following and more. Suppose that we draw an i.i.d.~sample of $n$ numbers uniformly at random from $1$ to $N$. For each number $x_i$, let $y_i = 1$ if $x_i$ is a prime and $0$ if it is not. Then, the interpolating MDL network accurately answers, with error probability $1- O((\ln N)/n)$, whether a newly drawn number between $1$ and $N$ is a prime or not. Note that the network is not designed to detect primes; minimum description learning discovers a network which does so. Extensions to noisy data are also discussed, suggesting that MDL neural network interpolators can demonstrate tempered overfitting.
Authors: Zhenglin Wan, Xingrui Yu, David Mark Bossens, Yueming Lyu, Qing Guo, Flint Xiaofeng Fan, Yew Soon Ong, Ivor Tsang
Abstract: Imitation learning (IL) has shown promise in various applications (e.g. robot locomotion) but is often limited to learning a single expert policy, constraining behavior diversity and robustness in unpredictable real-world scenarios. To address this, we introduce Quality Diversity Inverse Reinforcement Learning (QD-IRL), a novel framework that integrates quality-diversity optimization with IRL methods, enabling agents to learn diverse behaviors from limited demonstrations. This work introduces Extrinsic Behavioral Curiosity (EBC), which allows agents to receive additional curiosity rewards from an external critic based on how novel the behaviors are with respect to a large behavioral archive. To validate the effectiveness of EBC in exploring diverse locomotion behaviors, we evaluate our method on multiple robot locomotion tasks. EBC improves the performance of QD-IRL instances with GAIL, VAIL, and DiffAIL across all included environments by up to 185%, 42%, and 150%, even surpassing expert performance by 20% in Humanoid. Furthermore, we demonstrate that EBC is applicable to Gradient-Arborescence-based Quality Diversity Reinforcement Learning (QD-RL) algorithms, where it substantially improves performance and provides a generic technique for learning behavioral-diverse policies. The source code of this work is provided at https://github.com/vanzll/EBC.
Authors: Daohan Su, Xunkai Li, Zhenjun Li, Yinping Liao, Rong-Hua Li, Guoren Wang
Abstract: Recently, graph neural network (GNN) has emerged as a powerful representation learning tool for graph-structured data. However, most approaches are tailored for undirected graphs, neglecting the abundant information in the edges of directed graphs (digraphs). In fact, digraphs are widely applied in the real world and confirmed to address heterophily challenges. Despite recent advancements, existing spatial- and spectral-based DiGNNs have limitations due to their complex learning mechanisms and reliance on high-quality topology, resulting in low efficiency and unstable performance. To address these issues, we propose Directed Random Walk (DiRW), a plug-and-play strategy for most spatial-based DiGNNs and also an innovative model which offers a new digraph learning paradigm. Specifically, it utilizes a direction-aware path sampler optimized from the perspectives of walk probability, length, and number in a weight-free manner by considering node profiles and topologies. Building upon this, DiRW incorporates a node-wise learnable path aggregator for generalized node representations. Extensive experiments on 9 datasets demonstrate that DiRW: (1) enhances most spatial-based methods as a plug-and-play strategy; (2) achieves SOTA performance as a new digraph learning paradigm. The source code and data are available at https://github.com/dhsiuu/DiRW.
Authors: Qianru Zhang, Xinyi Gao, Haixin Wang, Dong Huang, Siu-Ming Yiu, Hongzhi Yin
Abstract: Spatial-temporal graph representations play a crucial role in urban sensing applications, including traffic analysis, human mobility behavior modeling, and citywide crime prediction. However, a key challenge lies in the noisy and sparse nature of spatial-temporal data, which limits existing neural networks' ability to learn meaningful region representations in the spatial-temporal graph. To overcome these limitations, we propose HGAurban, a novel heterogeneous spatial-temporal graph masked autoencoder that leverages generative self-supervised learning for robust urban data representation. Our framework introduces a spatial-temporal heterogeneous graph encoder that extracts region-wise dependencies from multi-source data, enabling comprehensive modeling of diverse spatial relationships. Within our self-supervised learning paradigm, we implement a masked autoencoder that jointly processes node features and graph structure. This approach automatically learns heterogeneous spatial-temporal patterns across regions, significantly improving the representation of dynamic temporal correlations. Comprehensive experiments across multiple spatiotemporal mining tasks demonstrate that our framework outperforms state-of-the-art methods and robustly handles real-world urban data challenges, including noise and sparsity in both spatial and temporal dimensions.
Authors: Runzhen Xue, Hao Wu, Mingyu Yan, Ziheng Xiao, Guangyu Sun, Xiaochun Ye, Dongrui Fan
Abstract: Design Space Exploration (DSE) is essential to modern CPU design, yet current frameworks struggle to scale and generalize in high-dimensional architectural spaces. As the dimensionality of design spaces continues to grow, existing DSE frameworks face three fundamental challenges: (1) reduced accuracy and poor scalability of surrogate models in large design spaces; (2) inefficient acquisition guided by hand-crafted heuristics or exhaustive search; (3) limited interpretability, making it hard to pinpoint architectural bottlenecks. In this work, we present \textbf{AttentionDSE}, the first end-to-end DSE framework that \emph{natively integrates} performance prediction and design guidance through an attention-based neural architecture. Unlike traditional DSE workflows that separate surrogate modeling from acquisition and rely heavily on hand-crafted heuristics, AttentionDSE establishes a unified, learning-driven optimization loop, in which attention weights serve a dual role: enabling accurate performance estimation and simultaneously exposing the performance bottleneck. This paradigm shift elevates attention from a passive representation mechanism to an active, interpretable driver of design decision-making. Key innovations include: (1) a \textbf{Perception-Driven Attention} mechanism that exploits architectural hierarchy and locality, scaling attention complexity from $\mathcal{O}(n^2)$ to $\mathcal{O}(n)$ via sliding windows; (2) an \textbf{Attention-aware Bottleneck Analysis} that automatically surfaces critical parameters for targeted optimization, eliminating the need for domain-specific heuristics. Evaluated on high-dimensional CPU design space using the SPEC CPU2017 benchmark suite, AttentionDSE achieves up to \textbf{3.9\% higher Pareto Hypervolume} and over \textbf{80\% reduction in exploration time} compared to state-of-the-art baselines.
Authors: Zhi Wen Soi, Chenrui Fan, Aditya Shankar, Abele M\u{a}lan, Lydia Y. Chen
Abstract: Distributed time series data presents a challenge for federated learning, as clients often possess different feature sets and have misaligned time steps. Existing federated time series models are limited by the assumption of perfect temporal or feature alignment across clients. In this paper, we propose FedTDD, a novel federated time series diffusion model that jointly learns a synthesizer across clients. At the core of FedTDD is a novel data distillation and aggregation framework that reconciles the differences between clients by imputing the misaligned timesteps and features. In contrast to traditional federated learning, FedTDD learns the correlation across clients' time series through the exchange of local synthetic outputs instead of model parameters. A coordinator iteratively improves a global distiller network by leveraging shared knowledge from clients through the exchange of synthetic data. As the distiller becomes more refined over time, it subsequently enhances the quality of the clients' local feature estimates, allowing each client to then improve its local imputations for missing data using the latest, more accurate distiller. Experimental results on five datasets demonstrate FedTDD's effectiveness compared to centralized training, and the effectiveness of sharing synthetic outputs to transfer knowledge of local time series. Notably, FedTDD achieves 79.4% and 62.8% improvement over local training in Context-FID and Correlational scores.
Authors: Philip Zmushko, Aleksandr Beznosikov, Martin Tak\'a\v{c}, Samuel Horv\'ath
Abstract: With the increase in the number of parameters in large language models, the process of pre-training and fine-tuning increasingly demands larger volumes of GPU memory. A significant portion of this memory is typically consumed by the optimizer state. To overcome this challenge, recent approaches such as low-rank adaptation (LoRA (Hu et al., 2021)), low-rank gradient projection (GaLore (Zhao et al., 2024)), and blockwise optimization (BAdam (Luo et al., 2024)) have been proposed. However, in all these algorithms, the $\textit{effective rank of the weight updates remains low-rank}$, which can lead to a substantial loss of information from the gradient. This loss can be critically important, especially during the pre-training stage. In this paper, we introduce $\texttt{FRUGAL}$ ($\textbf{F}$ull-$\textbf{R}$ank $\textbf{U}$pdates with $\textbf{G}$r$\textbf{A}$dient sp$\textbf{L}$itting), a new memory-efficient optimization framework. $\texttt{FRUGAL}$ leverages gradient splitting to perform low-dimensional updates using advanced algorithms (such as Adam), while updates along the remaining directions are executed via state-free methods like SGD or signSGD (Bernstein et al., 2018). Our framework can be integrated with various low-rank update selection techniques, including GaLore and BAdam. We provide theoretical convergence guarantees for our framework when using SGDM for low-dimensional updates and SGD for state-free updates. Additionally, our method consistently outperforms concurrent approaches across various fixed memory budgets, achieving state-of-the-art results in pre-training and fine-tuning tasks while balancing memory efficiency and performance metrics.
Authors: Zaikang Lin, Sei Chang, Aaron Zweig, Minseo Kang, Elham Azizi, David A. Knowles
Abstract: Modern high-throughput biological datasets with thousands of perturbations provide the opportunity for large-scale discovery of causal graphs that represent the regulatory interactions between genes. Differentiable causal graphical models have been proposed to infer a gene regulatory network (GRN) from large scale interventional datasets, capturing the causal gene regulatory relationships from genetic perturbations. However, existing models are limited in their expressivity and scalability while failing to address the dynamic nature of biological processes such as cellular differentiation. We propose PerturbODE, a novel framework that incorporates biologically informative neural ordinary differential equations (neural ODEs) to model cell state trajectories under perturbations and derive the causal GRN from the neural ODE's parameters. We demonstrate PerturbODE's efficacy in trajectory prediction and GRN inference across simulated and real over-expression datasets.
Authors: Yanxia Deng, Aozhong Zhang, Selcuk Gurses, Naigang Wang, Zi Yang, Penghang Yin
Abstract: Fine-tuning large language models (LLMs) using low-rank adaptation (LoRA) has become a highly efficient approach for downstream tasks, particularly in scenarios with limited computational resources. However, applying LoRA techniques to quantized LLMs poses unique challenges due to the reduced representational precision of quantized weights. In this paper, we introduce CLoQ (Calibrated LoRA initialization for Quantized LLMs), a simplistic initialization strategy designed to overcome these challenges. Our approach focuses on minimizing the layer-wise discrepancy between the original LLM and its quantized counterpart with LoRA components during initialization. By leveraging a small calibration dataset, CLoQ quantizes a pre-trained LLM and determines the optimal LoRA components for each layer, ensuring a strong foundation for subsequent fine-tuning. A key contribution of this work is a novel theoretical result that enables the accurate and closed-form construction of these optimal LoRA components. We validate the efficacy of CLoQ across multiple tasks such as language generation, arithmetic reasoning, and commonsense reasoning, demonstrating that it consistently outperforms existing LoRA fine-tuning methods for quantized LLMs, especially at ultra low-bit widths.
Authors: Isha Puri, Shivchander Sudalairaj, Guangxuan Xu, Kai Xu, Akash Srivastava
Abstract: Large language models (LLMs) have achieved significant performance gains via scaling up model sizes and/or data. However, recent evidence suggests diminishing returns from such approaches, motivating scaling the computation spent at inference time. Existing inference-time scaling methods, usually with reward models, cast the task as a search problem, which tends to be vulnerable to reward hacking as a consequence of approximation errors in reward models. In this paper, we instead cast inference-time scaling as a probabilistic inference task and leverage sampling-based techniques to explore the typical set of the state distribution of a state-space model with an approximate likelihood, rather than optimize for its mode directly. We propose a novel inference-time scaling approach by adapting particle-based Monte Carlo methods to this task. Our empirical evaluation demonstrates that our methods have a 4-16x better scaling rate over our deterministic search counterparts on various challenging mathematical reasoning tasks. Using our approach, we show that Qwen2.5-Math-1.5B-Instruct can surpass GPT-4o accuracy in only 4 rollouts, while Qwen2.5-Math-7B-Instruct scales to o1 level accuracy in only 32 rollouts. Our work not only presents an effective method to inference-time scaling, but also connects the rich literature in probabilistic inference with inference-time scaling of LLMs to develop more robust algorithms in future work. Code, videos, and further information available at https://probabilistic-inference-scaling.github.io.
Authors: Chenlu Ding, Jiancan Wu, Yancheng Yuan, Cunchun Li, Xiang Wang, Dingxian Wang, Frank Yang, Andrew Rabinovich
Abstract: In online advertising under the cost-per-conversion (CPA) model, accurate conversion rate (CVR) prediction is crucial. A major challenge is delayed feedback, where conversions may occur long after user interactions, leading to incomplete recent data and biased model training. Existing solutions partially mitigate this issue but often rely on auxiliary models, making them computationally inefficient and less adaptive to user interest shifts. We propose IF-DFM, an \underline{I}nfluence \underline{F}unction-empowered for \underline{D}elayed \underline{F}eedback \underline{M}odeling which estimates the impact of newly arrived and delayed conversions on model parameters, enabling efficient updates without full retraining. By reformulating the inverse Hessian-vector product as an optimization problem, IF-DFM achieves a favorable trade-off between scalability and effectiveness. Experiments on benchmark datasets show that IF-DFM outperforms prior methods in both accuracy and adaptability.
Authors: Su Hyeong Lee, Manzil Zaheer, Tian Li
Abstract: Distributed optimization has become the default training paradigm in modern machine learning due to the growing scale of models and datasets. To mitigate communication overhead, local updates are often applied before global aggregation, resulting in a nested optimization approach with inner and outer steps. However, heavy-tailed stochastic gradient noise remains a significant challenge, particularly in attention-based models, hindering effective training. In this work, we propose TailOPT, an efficient framework designed to address heavy-tailed noise by leveraging adaptive optimization or clipping techniques. We establish convergence guarantees for the TailOPT framework under heavy-tailed noise with potentially unbounded gradient variance and local updates. Among its variants, we highlight a memory and communication efficient instantiation which we call $Bi^2Clip$, which performs coordinate-wise clipping at both the inner and outer optimizers, achieving adaptive-like performance (e.g., Adam) without the cost of maintaining or transmitting additional gradient statistics. Empirically, TailOPT, including $Bi^2Clip$, demonstrates superior performance on several language tasks and models, outperforming state-of-the-art methods.
Authors: Hoony Kang, Wolfgang Losert
Abstract: The brain rapidly adapts to new contexts and learns from limited data, a coveted characteristic that artificial intelligence (AI) algorithms struggle to mimic. Inspired by the mechanical oscillatory rhythms of neural cells, we developed a learning paradigm utilizing link strength oscillations, where learning is associated with the coordination of these oscillations. Link oscillations can rapidly change coordination, allowing the network to sense and adapt to subtle contextual changes without supervision. The network becomes a generalist AI architecture, capable of predicting dynamics of multiple contexts including unseen ones. These results make our paradigm a powerful starting point for novel models of cognition. Because our paradigm is agnostic to specifics of the neural network, our study opens doors for introducing rapid adaptive learning into leading AI models.
Authors: Haoyu Lei, Kaiwen Zhou, Yinchuan Li, Zhitang Chen, Farzan Farnia
Abstract: Diffusion-based Neural Combinatorial Optimization (NCO) has demonstrated effectiveness in solving NP-complete (NPC) problems by learning discrete diffusion models for solution generation, eliminating hand-crafted domain knowledge. Despite their success, existing NCO methods face significant challenges in both cross-scale and cross-problem generalization, and high training costs compared to traditional solvers. While recent studies on diffusion models have introduced training-free guidance approaches that leverage pre-defined guidance functions for conditional generation, such methodologies have not been extensively explored in combinatorial optimization. To bridge this gap, we propose a training-free inference time adaptation framework (DIFU-Ada) that enables both the zero-shot cross-problem transfer and cross-scale generalization capabilities of diffusion-based NCO solvers without requiring additional training. We provide theoretical analysis that helps understanding the cross-problem transfer capability. Our experimental results demonstrate that a diffusion solver, trained exclusively on the Traveling Salesman Problem (TSP), can achieve competitive zero-shot transfer performance across different problem scales on TSP variants, such as Prize Collecting TSP (PCTSP) and the Orienteering Problem (OP), through inference time adaptation.
Authors: Ohad Einav, Nir Rosenfeld
Abstract: Machine learning models play a key role for service providers looking to gain market share in consumer markets. However, traditional learning approaches do not take into account the existence of additional providers, who compete with each other for consumers. Our work aims to study learning in this market setting, as it affects providers, consumers, and the market itself. We begin by analyzing such markets through the lens of the learning objective, and show that accuracy cannot be the only consideration. We then propose a method for classification under competition, so that a learner can maximize market share in the presence of competitors. We show that our approach benefits the providers as well as the consumers, and find that the timing of market entry and model updates can be crucial. We display the effectiveness of our approach across a range of domains, from simple distributions to noisy datasets, and show that the market as a whole remains stable by converging quickly to an equilibrium.
Authors: Yonatan Sommer, Ivri Hikri, Lotan Amit, Nir Rosenfeld
Abstract: When learning is used to inform decisions about humans, such as for loans, hiring, or admissions, this can incentivize users to strategically modify their features, at a cost, to obtain positive predictions. The common assumption is that the function governing costs is exogenous, fixed, and predetermined. We challenge this assumption, and assert that costs can emerge as a result of deploying a classifier. Our idea is simple: when users seek positive predictions, this creates demand for important features; and if features are available for purchase, then a market will form, and competition will give rise to prices. We extend the strategic classification framework to support this notion, and study learning in a setting where a classifier can induce a market for features. We present an analysis of the learning task, devise an algorithm for computing market prices, propose a differentiable learning framework, and conduct experiments to explore our novel setting and approach.
Authors: Sahar Admoni, Assaf Hallak, Yftah Ziser, Omer Ben-Porat, Ofra Amir
Abstract: Policies generated by Reinforcement Learning (RL) algorithms are difficult to explain to users, as they emerge from the interaction of complex reward structures and neural network representations. Consequently, analyzing and predicting agent behavior can be challenging, undermining user trust in real-world applications. To facilitate user understanding, current methods for global policy summarization typically rely on videos that demonstrate agent behavior in a subset of world states. However, users can only watch a limited number of demonstrations, constraining their understanding. Moreover, these methods place the burden of interpretation on users by presenting raw behaviors rather than synthesizing them into coherent patterns. To resolve these issues, we introduce SySLLM (Synthesized Summary using Large Language Models), advocating for a new paradigm of abstractive-textual policy explanations. By leveraging Large Language Models (LLMs)-which possess extensive world knowledge and pattern synthesis capabilities-SySLLM generates textual summaries that provide structured and comprehensible explanations of agent policies. SySLLM demonstrates that LLMs can interpret spatio-temporally structured descriptions of state-action trajectories from an RL agent and generate valuable policy insights in a zero-shot setting, without any prior knowledge or fine-tuning. Our evaluation shows that SySLLM captures key insights, such as goal preferences and exploration strategies, that were also identified by human experts. Furthermore, in a large-scale user study (with 200 participants), SySLLM summaries were preferred over demonstration-based summaries (HIGHLIGHTS) by a clear majority (75.5%) of participants.
Authors: Suhas G Hegde, Shilpy Kaur, Aruna Tiwari
Abstract: Popular PEFT methods reduce trainable parameter count for fine-tuning by parameterizing new low-rank or sparse trainable weights in parallel to the frozen pre-trained weights $W$. However, these weights are trained from scratch, and there exists a performance gap between these methods and full fine-tuning, especially in low-budget settings. We introduce VectorFit, a new way of parameterization that efficiently utilizes the existing knowledge embedded in $W$ by adaptively training their singular vectors and biases. We show that utilizing the structural and transformational properties of $W$ in this way can lead to high-rank incremental weight matrices $\Delta W$, comparable to that of full fine-tuning. VectorFit delivers superior results with 9$\boldsymbol\times$ fewer trainable parameters than the leading PEFT methods. Through comprehensive experiments across 19 datasets covering a wide range of language and vision tasks such as natural language understanding and generation, question answering, image classification, and image generation, we demonstrate that VectorFit surpasses baselines in terms of performance as a function of parameter-efficiency.
Authors: Chandra Raskoti, Iftekharul Islam, Xuan Wang, Weizi Li
Abstract: Accurate vehicle trajectory prediction is critical for safe and efficient autonomous driving, especially in mixed traffic environments when both human-driven and autonomous vehicles co-exist. However, uncertainties introduced by inherent driving behaviors -- such as acceleration, deceleration, and left and right maneuvers -- pose significant challenges for reliable trajectory prediction. We introduce a Maneuver-Intention-Aware Transformer (MIAT) architecture, which integrates a maneuver intention awareness control mechanism with spatiotemporal interaction modeling to enhance long-horizon trajectory predictions. We systematically investigate the impact of varying awareness of maneuver intention on both short- and long-horizon trajectory predictions. Evaluated on the real-world NGSIM dataset and benchmarked against various transformer- and LSTM-based methods, our approach achieves an improvement of up to 4.7% in short-horizon predictions and a 1.6% in long-horizon predictions compared to other intention-aware benchmark methods. Moreover, by leveraging intention awareness control mechanism, MIAT realizes an 11.1% performance boost in long-horizon predictions, with a modest drop in short-horizon performance. The source code and datasets are available at https://github.com/cpraskoti/MIAT.
Authors: Zekai Chen, Xunkai Li, Yinlin Zhu, Rong-Hua Li, Guoren Wang
Abstract: As a new distributed graph learning paradigm, Federated Graph Learning (FGL) facilitates collaborative model training across local systems while preserving data privacy. We review existing FGL approaches and categorize their optimization mechanisms into: (1) Server-Client (S-C), where clients upload local model parameters for server-side aggregation and global updates; (2) Client-Client (C-C), which allows direct exchange of information between clients and customizing their local training process. We reveal that C-C shows superior potential due to its refined communication structure. However, existing C-C methods broadcast redundant node representations, incurring high communication costs and privacy risks at the node level. To this end, we propose FedC4, which combines graph Condensation with C-C Collaboration optimization. Specifically, FedC4 employs graph condensation technique to refine the knowledge of each client's graph into a few synthetic embeddings instead of transmitting node-level knowledge. Moreover, FedC4 introduces three novel modules that allow the source client to send distinct node representations tailored to the target client's graph properties. Experiments on eight public real-world datasets show that FedC4 outperforms state-of-the-art baselines in both task performance and communication cost. Our code is now available on https://github.com/Ereshkigal1/FedC4.
Authors: Luca-Andrei Fechete, Mohamed Sana, Fadhel Ayed, Nicola Piovesan, Wenjie Li, Antonio De Domenico, Tareq Si Salem
Abstract: Conventional time-series forecasting methods typically aim to minimize overall prediction error, without accounting for the varying importance of different forecast ranges in downstream applications. We propose a training methodology that enables forecasting models to adapt their focus to application-specific regions of interest at inference time, without retraining. The approach partitions the prediction space into fine-grained segments during training, which are dynamically reweighted and aggregated to emphasize the target range specified by the application. Unlike prior methods that predefine these ranges, our framework supports flexible, on-demand adjustments. Experiments on standard benchmarks and a newly collected wireless communication dataset demonstrate that our method not only improves forecast accuracy within regions of interest but also yields measurable gains in downstream task performance. These results highlight the potential for closer integration between predictive modeling and decision-making in real-world systems.
Authors: Shubham Vaishnav, Praveen Kumar Donta, Sindri Magn\'usson
Abstract: Internet of Things (IoT) systems increasingly operate in environments where devices must respond in real time while managing fluctuating resource constraints, including energy and bandwidth. Yet, current approaches often fall short in addressing scenarios where operational constraints evolve over time. To address these limitations, we propose a novel Budgeted Multi-Armed Bandit framework tailored for IoT applications with dynamic operational limits. Our model introduces a decaying violation budget, which permits limited constraint violations early in the learning process and gradually enforces stricter compliance over time. We present the Budgeted Upper Confidence Bound (UCB) algorithm, which adaptively balances performance optimization and compliance with time-varying constraints. We provide theoretical guarantees showing that Budgeted UCB achieves sublinear regret and logarithmic constraint violations over the learning horizon. Extensive simulations in a wireless communication setting show that our approach achieves faster adaptation and better constraint satisfaction than standard online learning methods. These results highlight the framework's potential for building adaptive, resource-aware IoT systems.
Authors: Euntae Choi, Sumin Song, Woosang Lim, Sungjoo Yoo
Abstract: Large Language Models (LLMs) face deployment challenges due to high computational costs, and while Post-Training Quantization (PTQ) offers a solution, existing rotation-based methods struggle at very low bit-widths like 2-bit. We introduce a novel, training-free approach to construct an improved rotation matrix, addressing the limitations of current methods. The key contributions include leveraging the Walsh-Hadamard transform with sequency ordering, which clusters similar frequency components to reduce quantization error compared to standard Hadamard matrices, significantly improving performance. Furthermore, we propose a Grouped Sequency-arranged Rotation (GSR) using block-diagonal matrices with smaller Walsh blocks, effectively isolating outlier impacts and achieving performance comparable to optimization-based methods without requiring any training. Our method demonstrates robust performance on reasoning tasks and Perplexity (PPL) score on WikiText-2. Our method also enhances results even when applied over existing learned rotation techniques.
Authors: Nathan Justin, Qingshi Sun, Andr\'es G\'omez, Phebe Vayanos
Abstract: In the last few decades, Machine Learning (ML) has achieved significant success across domains ranging from healthcare, sustainability, and the social sciences, to criminal justice and finance. But its deployment in increasingly sophisticated, critical, and sensitive areas affecting individuals, the groups they belong to, and society as a whole raises critical concerns around fairness, transparency and robustness, among others. As the complexity and scale of ML systems and of the settings in which they are deployed grow, so does the need for responsible ML methods that address these challenges while providing guaranteed performance in deployment. Mixed-integer optimization (MIO) offers a powerful framework for embedding responsible ML considerations directly into the learning process while maintaining performance. For example, it enables learning of inherently transparent models that can conveniently incorporate fairness or other domain specific constraints. This tutorial paper provides an accessible and comprehensive introduction to this topic discussing both theoretical and practical aspects. It outlines some of the core principles of responsible ML, their importance in applications, and the practical utility of MIO for building ML models that align with these principles. Through examples and mathematical formulations, it illustrates practical strategies and available tools for efficiently solving MIO problems for responsible ML. It concludes with a discussion on current limitations and open research questions, providing suggestions for future work.
Authors: Quang-Duy Tran, Bao Duong, Phuoc Nguyen, Thin Nguyen
Abstract: Telling apart the cause and effect between two random variables with purely observational data is a challenging problem that finds applications in various scientific disciplines. A key principle utilized in this task is the algorithmic Markov condition, which postulates that the joint distribution, when factorized according to the causal direction, yields a more succinct codelength compared to the anti-causal direction. Previous approaches approximate these codelengths by relying on simple functions or Gaussian processes (GPs) with easily evaluable complexity, compromising between model fitness and computational complexity. To address these limitations, we propose leveraging the variational Bayesian learning of neural networks as an interpretation of the codelengths. This allows the improvement of model fitness, while maintaining the succinctness of the codelengths, and the avoidance of the significant computational complexity of the GP-based approaches. Extensive experiments on both synthetic and real-world benchmarks in cause-effect identification demonstrate the effectiveness of our proposed method, showing promising performance enhancements on several datasets in comparison to most related methods.
Authors: Guangda Liu, Chengwei Li, Zhenyu Ning, Minyi Guo, Jieru Zhao
Abstract: Large language models (LLMs) have been widely deployed with rapidly expanding context windows to support increasingly demanding applications. However, long contexts pose significant deployment challenges, primarily due to the KV cache whose size grows proportionally with context length. While KV cache compression methods are proposed to address this issue, KV dropping methods incur considerable accuracy loss, and KV retrieval methods suffer from significant efficiency bottlenecks. We propose FreeKV, an algorithm-system co-optimization framework to enhance KV retrieval efficiency while preserving accuracy. On the algorithm side, FreeKV introduces speculative retrieval to shift the KV selection and recall processes out of the critical path, combined with fine-grained correction to ensure accuracy. On the system side, FreeKV employs hybrid KV layouts across CPU and GPU memory to eliminate fragmented data transfers, and leverages double-buffered streamed recall to further improve efficiency. Experiments demonstrate that FreeKV achieves near-lossless accuracy across various scenarios and models, delivering up to 13$\times$ speedup compared to SOTA KV retrieval methods.
Authors: Enric Ribera Borrell, Lorenz Richter, Christof Sch\"utte
Abstract: We extend the standard reinforcement learning framework to random time horizons. While the classical setting typically assumes finite and deterministic or infinite runtimes of trajectories, we argue that multiple real-world applications naturally exhibit random (potentially trajectory-dependent) stopping times. Since those stopping times typically depend on the policy, their randomness has an effect on policy gradient formulas, which we (mostly for the first time) derive rigorously in this work both for stochastic and deterministic policies. We present two complementary perspectives, trajectory or state-space based, and establish connections to optimal control theory. Our numerical experiments demonstrate that using the proposed formulas can significantly improve optimization convergence compared to traditional approaches.
Authors: Olya Mastikhina, Dhruv Sreenivas, Pablo Samuel Castro
Abstract: Actor-critic methods have been central to many of the recent advances in deep reinforcement learning. The most common approach is to use symmetric architectures, whereby both actor and critic have the same network topology and number of parameters. However, recent works have argued for the advantages of asymmetric setups, specifically with the use of smaller actors. We perform broad empirical investigations and analyses to better understand the implications of this and find that, in general, smaller actors result in performance degradation and overfit critics. Our analyses suggest poor data collection, due to value underestimation, as one of the main causes for this behavior, and further highlight the crucial role the critic can play in alleviating this pathology. We explore techniques to mitigate the observed value underestimation, which enables further research in asymmetric actor-critic methods.
Authors: Andrew P. Berg, Qian Zhang, Mia Y. Wang
Abstract: As unmanned aerial vehicles (UAVs) become increasingly prevalent in both consumer and defense applications, the need for reliable, modality-specific classification systems grows in urgency. This paper addresses the challenge of data scarcity in UAV audio classification by expanding on prior work through the integration of pre-trained deep learning models, parameter-efficient fine-tuning (PEFT) strategies, and targeted data augmentation techniques. Using a custom dataset of 3,100 UAV audio clips (15,500 seconds) spanning 31 distinct drone types, we evaluate the performance of transformer-based and convolutional neural network (CNN) architectures under various fine-tuning configurations. Experiments were conducted with five-fold cross-validation, assessing accuracy, training efficiency, and robustness. Results show that full fine-tuning of the EfficientNet-B0 model with three augmentations achieved the highest validation accuracy (95.95), outperforming both the custom CNN and transformer-based models like AST. These findings suggest that combining lightweight architectures with PEFT and well-chosen augmentations provides an effective strategy for UAV audio classification on limited datasets. Future work will extend this framework to multimodal UAV classification using visual and radar telemetry.
Authors: Ching Chang, Ming-Chih Lo, Wen-Chih Peng, Tien-Fu Chen
Abstract: Multivariate time series data, collected across various fields such as manufacturing and wearable technology, exhibit states at multiple levels of granularity, from coarse-grained system behaviors to fine-grained, detailed events. Effectively segmenting and integrating states across these different granularities is crucial for tasks like predictive maintenance and performance optimization. However, existing time series segmentation methods face two key challenges: (1) the inability to handle multiple levels of granularity within a unified model, and (2) limited adaptability to new, evolving patterns in dynamic environments. To address these challenges, we propose PromptTSS, a novel framework for time series segmentation with multi-granularity states. PromptTSS uses a unified model with a prompting mechanism that leverages label and boundary information to guide segmentation, capturing both coarse- and fine-grained patterns while adapting dynamically to unseen patterns. Experiments show PromptTSS improves accuracy by 24.49% in multi-granularity segmentation, 17.88% in single-granularity segmentation, and up to 599.24% in transfer learning, demonstrating its adaptability to hierarchical states and evolving time series dynamics. Our code is available at https://github.com/blacksnail789521/PromptTSS.
Authors: Daniel Zhengyu Huang, Jiaoyang Huang, Zhengjiang Lin
Abstract: Diffusion probabilistic models generate samples by learning to reverse a noise-injection process that transforms data into noise. A key development is the reformulation of the reverse sampling process as a deterministic probability flow ordinary differential equation (ODE), which allows for efficient sampling using high-order numerical solvers. Unlike traditional time integrator analysis, the accuracy of this sampling procedure depends not only on numerical integration errors but also on the approximation quality and regularity of the learned score function, as well as their interaction. In this work, we present a rigorous convergence analysis of deterministic samplers derived from probability flow ODEs for general forward processes with arbitrary variance schedules. Specifically, we develop and analyze $p$-th order (exponential) Runge-Kutta schemes, under the practical assumption that the first and second derivatives of the learned score function are bounded. We prove that the total variation distance between the generated and target distributions can be bounded as \begin{align*} O\bigl(d^{\frac{7}{4}}\varepsilon_{\text{score}}^{\frac{1}{2}} +d(dH_{\max})^p\bigr), \end{align*} where $\varepsilon^2_{\text{score}}$ denotes the $L^2$ error in the score function approximation, $d$ is the data dimension, and $H_{\max}$ represents the maximum solver step size. Numerical experiments on benchmark datasets further confirm that the derivatives of the learned score function are bounded in practice.
Authors: Ziyu Zheng, Yaming Yang, Ziyu Guan, Wei Zhao, Weigang Lu
Abstract: Masked Graph Auto-Encoder, a powerful graph self-supervised training paradigm, has recently shown superior performance in graph representation learning. Existing works typically rely on node contextual information to recover the masked information. However, they fail to generalize well to heterophilic graphs where connected nodes may be not similar, because they focus only on capturing the neighborhood information and ignoring the discrepancy information between different nodes, resulting in indistinguishable node representations. In this paper, to address this issue, we propose a Discrepancy-Aware Graph Mask Auto-Encoder (DGMAE). It obtains more distinguishable node representations by reconstructing the discrepancy information of neighboring nodes during the masking process. We conduct extensive experiments on 17 widely-used benchmark datasets. The results show that our DGMAE can effectively preserve the discrepancies of nodes in low-dimensional space. Moreover, DGMAE significantly outperforms state-of-the-art graph self-supervised learning methods on three graph analytic including tasks node classification, node clustering, and graph classification, demonstrating its remarkable superiority. The code of DGMAE is available at https://github.com/zhengziyu77/DGMAE.
Authors: Shan Shen, Shenglu Hua, Jiajun Zou, Jiawei Liu, Jianwang Zhai, Chuan Shi, Wenjian Yu
Abstract: Graph representation learning on Analog-Mixed Signal (AMS) circuits is crucial for various downstream tasks, e.g., parasitic estimation. However, the scarcity of design data, the unbalanced distribution of labels, and the inherent diversity of circuit implementations pose significant challenges to learning robust and transferable circuit representations. To address these limitations, we propose CircuitGCL, a novel graph contrastive learning framework that integrates representation scattering and label rebalancing to enhance transferability across heterogeneous circuit graphs. CircuitGCL employs a self-supervised strategy to learn topology-invariant node embeddings through hyperspherical representation scattering, eliminating dependency on large-scale data. Simultaneously, balanced mean squared error (BMSE) and balanced softmax cross-entropy (BSCE) losses are introduced to mitigate label distribution disparities between circuits, enabling robust and transferable parasitic estimation. Evaluated on parasitic capacitance estimation (edge-level task) and ground capacitance classification (node-level task) across TSMC 28nm AMS designs, CircuitGCL outperforms all state-of-the-art (SOTA) methods, with the $R^2$ improvement of $33.64\% \sim 44.20\%$ for edge regression and F1-score gain of $0.9\times \sim 2.1\times$ for node classification. Our code is available at https://github.com/ShenShan123/CircuitGCL.
Authors: Elisa Tsai, Haizhong Zheng, Atul Prakash
Abstract: High-quality training data is essential for building reliable and efficient machine learning systems. One-shot coreset selection addresses this by pruning the dataset while maintaining or even improving model performance, often relying on training-dynamics-based data difficulty scores. However, most existing methods implicitly assume class-wise homogeneity in data difficulty, overlooking variation in data difficulty across different classes. In this work, we challenge this assumption by showing that, in domains such as network intrusion detection and medical imaging, data difficulty often clusters by class. We formalize this as class-difficulty separability and introduce the Class Difficulty Separability Coefficient (CDSC) as a quantitative measure. We demonstrate that high CDSC values correlate with performance degradation in class-agnostic coreset methods, which tend to overrepresent easy majority classes while neglecting rare but informative ones. To address this, we introduce class-proportional variants of multiple sampling strategies. Evaluated on five diverse datasets spanning security and medical domains, our methods consistently achieve state-of-the-art performance. For instance, on CTU-13, at an extreme 99% pruning rate, a class-proportional variant of Coverage-centric Coreset Selection (CCS-CP) shows remarkable stability, with accuracy dropping only 2.58%, precision 0.49%, and recall 0.19%. In contrast, the class-agnostic CCS baseline, the next best method, suffers sharper declines of 7.59% in accuracy, 4.57% in precision, and 4.11% in recall. We further show that aggressive pruning enhances generalization in noisy, imbalanced, and large-scale datasets. Our results underscore that explicitly modeling class-difficulty separability leads to more effective, robust, and generalizable data pruning, particularly in high-stakes scenarios.
Authors: Scott Pesme, Giacomo Meanti, Michael Arbel, Julien Mairal
Abstract: Denoiser models have become powerful tools for inverse problems, enabling the use of pretrained networks to approximate the score of a smoothed prior distribution. These models are often used in heuristic iterative schemes aimed at solving Maximum a Posteriori (MAP) optimisation problems, where the proximal operator of the negative log-prior plays a central role. In practice, this operator is intractable, and practitioners plug in a pretrained denoiser as a surrogate-despite the lack of general theoretical justification for this substitution. In this work, we show that a simple algorithm, closely related to several used in practice, provably converges to the proximal operator under a log-concavity assumption on the prior $p$. We show that this algorithm can be interpreted as a gradient descent on smoothed proximal objectives. Our analysis thus provides a theoretical foundation for a class of empirically successful but previously heuristic methods.
Authors: Xuchuang Wang, Bo Sun, Hedyeh Beyhaghi, John C. S. Lui, Mohammad Hajiesmaili, Adam Wierman
Abstract: This paper introduces a novel multi-agent ski-rental problem that generalizes the classical ski-rental dilemma to a group setting where agents incur individual and shared costs. In our model, each agent can either rent at a fixed daily cost, or purchase a pass at an individual cost, with an additional third option of a discounted group pass available to all. We consider scenarios in which agents' active days differ, leading to dynamic states as agents drop out of the decision process. To address this problem from different perspectives, we define three distinct competitive ratios: overall, state-dependent, and individual rational. For each objective, we design and analyze optimal deterministic and randomized policies. Our deterministic policies employ state-aware threshold functions that adapt to the dynamic states, while our randomized policies sample and resample thresholds from tailored state-aware distributions. The analysis reveals that symmetric policies, in which all agents use the same threshold, outperform asymmetric ones. Our results provide competitive ratio upper and lower bounds and extend classical ski-rental insights to multi-agent settings, highlighting both theoretical and practical implications for group decision-making under uncertainty.
Authors: Lili Chen, Mihir Prabhudesai, Katerina Fragkiadaki, Hao Liu, Deepak Pathak
Abstract: Can large language models improve without external data -- by generating their own questions and answers? We hypothesize that a pre-trained language model can improve its reasoning skills given only a single prompt specifying the topic (e.g., algebra word problems) and asking the model to generate its own questions. To do this, we propose Self-Questioning Language Models (SQLM): an asymmetric self-play framework where a proposer is given the topic and generates a question for a solver, who tries to answer it. Both the proposer and solver are trained via reinforcement learning. The proposer receives a reward if the problem is not too easy or too difficult, and the solver receives a reward based on majority voting, a proxy for correctness in the absence of ground-truth answers. For coding, the proposer can instead generate unit tests which are used for verification. We study this asymmetric self-play framework on three benchmarks: three-digit multiplication, algebra problems from the OMEGA benchmark, and programming problems from Codeforces. By continually generating more interesting problems and attempting to solve them, language models can improve on downstream benchmarks without access to any curated training datasets.
Authors: Feiyu Wang, Guoan Wang, Yihao Zhang, Shengfan Wang, Weitao Li, Bokai Huang, Shimao Chen, Zihan Jiang, Rui Xu, Tong Yang
Abstract: Quantization-Aware Training (QAT) integrates quantization into the training loop, enabling LLMs to learn robust low-bit representations, and is widely recognized as one of the most promising research directions. All current QAT research focuses on minimizing quantization error on full-precision models, where the full-precision accuracy acts as an upper bound (accuracy ceiling). No existing method has even attempted to surpass this ceiling. To break this ceiling, we propose a new paradigm: raising the ceiling (full-precision model), and then still quantizing it efficiently into 2 bits. We propose Fairy$\pm i$, the first 2-bit quantization framework for complex-valued LLMs. Specifically, our method leverages the representational advantages of the complex domain to boost full-precision accuracy. We map weights to the fourth roots of unity $\{\pm1, \pm i\}$, forming a perfectly symmetric and information-theoretically optimal 2-bit representation. Importantly, each quantized weight has either a zero real or imaginary part, enabling multiplication-free inference using only additions and element swaps. Experimental results show that Fairy$\pm i$ outperforms the ceiling of existing 2-bit quantization approaches in terms of both PPL and downstream tasks, while maintaining strict storage and compute efficiency. This work opens a new direction for building highly accurate and practical LLMs under extremely low-bit constraints.
Authors: Linghao Zhu, Yiran Guan, Dingkang Liang, Jianzhong Ju, Zhenbo Luo, Bin Qin, Jian Luan, Yuliang Liu, Xiang Bai
Abstract: Reinforcement learning (RL) has emerged as an effective post-training paradigm for enhancing the reasoning capabilities of multimodal large language model (MLLM). However, current RL pipelines often suffer from training inefficiencies caused by two underexplored issues: Advantage Collapsing, where most advantages in a batch concentrate near zero, and Rollout Silencing, where the proportion of rollouts contributing non-zero gradients diminishes over time. These issues lead to suboptimal gradient updates and hinder long-term learning efficiency. To address these issues, we propose Shuffle-R1, a simple yet principled framework that improves RL fine-tuning efficiency by dynamically restructuring trajectory sampling and batch composition. It introduces (1) Pairwise Trajectory Sampling, which selects high-contrast trajectories with large advantages to improve gradient signal quality, and (2) Advantage-based Trajectory Shuffle, which increases exposure of valuable rollouts through informed batch reshuffling. Experiments across multiple reasoning benchmarks show that our framework consistently outperforms strong RL baselines with minimal overhead. These results highlight the importance of data-centric adaptations for more efficient RL training in MLLM.
Authors: Massimiliano Romiti
Abstract: The vast corpus of physics equations forms an implicit network of mathematical relationships that traditional analysis cannot fully explore. This work introduces a graph-based framework combining neural networks with symbolic analysis to systematically discover and validate mathematical patterns across physics domains. Starting from 659 equations, we performed rigorous semantic disambiguation to resolve notational polysemy affecting 213 equations, then focused on 400 advanced physics equations by excluding elementary mechanics to emphasize inter-branch connections of modern physics. This corpus was represented as a weighted knowledge graph where a Graph Attention Network achieved 97.4% AUC in link prediction, significantly outperforming classical baselines. The framework's primary value emerges from its dual capability: generating hypotheses and auditing knowledge. First, it functions as a hypothesis generator, producing hundreds of candidate cross-domain connections, from blackbody radiation coupled with Navier-Stokes equations to radioactive decay linked with electromagnetic induction. Second, through symbolic analysis of 30 equation clusters, it serves as a computational auditor that verified established theory consistencies, synthesized the Magnetic Reynolds Number from electromagnetic-fluid coupling, and revealed how even parsing errors could potentially point toward legitimate research like analog gravity. This proof-of-concept intentionally over-generates candidates to ensure comprehensive exploration of mathematical possibility space. Even tautologies and errors serve scientific purposes: redundancy identification and knowledge base quality assessment. The system transforms the intractable combinatorial space into a filtered stream of mathematical patterns for human interpretation.
Authors: Aoming Liang, Chi Cheng, Dashuai Chen, Boai Sun, Dixia Fan
Abstract: In the domain of scientific machine learning, designing effective reward functions remains a challenge in reinforcement learning (RL), particularly in environments where task goals are difficult to specify numerically. Reward functions in existing work are predominantly based on heuristics, manual engineering, or task-specific tuning. In this work, we introduce a semantically aligned reinforcement learning method where rewards are computed by aligning the current state with a target semantic instruction using a Sentence-Bidirectional Encoder Representations from Transformers (SBERT). Instead of relying on manually defined reward functions, the policy receives feedback based on the reward, which is a cosine similarity between the goal textual description and the statement description in the episode. We evaluated our approach in several environments and showed that semantic reward can guide learning to achieve competitive control behavior, even in the absence of hand-crafted reward functions. Our study demonstrates a correlation between the language embedding space and the conventional Euclidean space. This framework opens new horizons for aligning agent behavior with natural language goals and lays the groundwork for a more seamless integration of larger language models (LLMs) and fluid control applications.
Authors: Zichuan Liu, Jinyu Wang, Lei Song, Jiang Bian
Abstract: Recent advancements in post-training Large Language Models (LLMs), particularly through Reinforcement Learning (RL) and preference optimization methods, are key drivers for enhancing their reasoning capabilities. However, these methods are often plagued by low sample efficiency and a susceptibility to primacy bias, where overfitting to initial experiences degrades policy quality and damages the learning process. To address these challenges, we introduce LLM optimization with Reset Replay (LoRR), a general and powerful plugin designed to enhance sample efficiency in any preference-based optimization framework. LoRR core mechanism enables training at a high replay number, maximizing the utility of each collected data batch. To counteract the risk of overfitting inherent in high-replay training, LoRR incorporates a periodic reset strategy with reusing initial data, which preserves network plasticity. Furthermore, it leverages a hybrid optimization objective, combining supervised fine-tuning (SFT) and preference-based losses to further bolster data exploitation. Our extensive experiments demonstrate that LoRR significantly boosts the performance of various preference optimization methods on both mathematical and general reasoning benchmarks. Notably, an iterative DPO approach augmented with LoRR achieves comparable performance on challenging math tasks, outperforming some complex and computationally intensive RL-based algorithms. These findings highlight that LoRR offers a practical, sample-efficient, and highly effective paradigm for LLM finetuning, unlocking greater performance from limited data.
Authors: Rui Liu, Haolin Zuo, Zheng Lian, Hongyu Yuan, Qi Fan
Abstract: Missing modalities have recently emerged as a critical research direction in multimodal emotion recognition (MER). Conventional approaches typically address this issue through missing modality reconstruction. However, these methods fail to account for variations in reconstruction difficulty across different samples, consequently limiting the model's ability to handle hard samples effectively. To overcome this limitation, we propose a novel Hardness-Aware Dynamic Curriculum Learning framework, termed HARDY-MER. Our framework operates in two key stages: first, it estimates the hardness level of each sample, and second, it strategically emphasizes hard samples during training to enhance model performance on these challenging instances. Specifically, we first introduce a Multi-view Hardness Evaluation mechanism that quantifies reconstruction difficulty by considering both Direct Hardness (modality reconstruction errors) and Indirect Hardness (cross-modal mutual information). Meanwhile, we introduce a Retrieval-based Dynamic Curriculum Learning strategy that dynamically adjusts the training curriculum by retrieving samples with similar semantic information and balancing the learning focus between easy and hard instances. Extensive experiments on benchmark datasets demonstrate that HARDY-MER consistently outperforms existing methods in missing-modality scenarios. Our code will be made publicly available at https://github.com/HARDY-MER/HARDY-MER.
Authors: Hao Liu, Chun Yang, Zhang xiaoxing, Xiaobin Zhu
Abstract: Time series forecasting plays a significant role in finance, energy, meteorology, and IoT applications. Recent studies have leveraged the generalization capabilities of large language models (LLMs) to adapt to time series forecasting, achieving promising performance. However, existing studies focus on token-level modal alignment, instead of bridging the intrinsic modality gap between linguistic knowledge structures and time series data patterns, greatly limiting the semantic representation. To address this issue, we propose a novel Semantic-Enhanced LLM (SE-LLM) that explores the inherent periodicity and anomalous characteristics of time series to embed into the semantic space to enhance the token embedding. This process enhances the interpretability of tokens for LLMs, thereby activating the potential of LLMs for temporal sequence analysis. Moreover, existing Transformer-based LLMs excel at capturing long-range dependencies but are weak at modeling short-term anomalies in time-series data. Hence, we propose a plugin module embedded within self-attention that models long-term and short-term dependencies to effectively adapt LLMs to time-series analysis. Our approach freezes the LLM and reduces the sequence dimensionality of tokens, greatly reducing computational consumption. Experiments demonstrate the superiority performance of our SE-LLM against the state-of-the-art (SOTA) methods.
Authors: Junyu Wu, Weiming Chang, Xiaotao Liu, Guanyou He, Tingfeng Xian, Haoqiang Hong, Boqi Chen, Haotao Tian, Tao Yang, Yunsheng Shi, Feng Lin, Ting Yao
Abstract: Reinforcement Learning from Human Feedback (RLHF) has emerged as a prominent paradigm for training large language models and multimodal systems. Despite notable advances enabled by existing RLHF training frameworks, significant challenges remain in scaling to complex multimodal workflows and adapting to dynamic workloads. In particular, current systems often encounter limitations related to controller scalability when managing large models, as well as inefficiencies in orchestrating intricate RLHF pipelines, especially in scenarios that require dynamic sampling and resource allocation. In this paper, we introduce WeChat-YATT (Yet Another Transformer Trainer in WeChat), a simple, scalable, and balanced RLHF training framework specifically designed to address these challenges. WeChat-YATT features a parallel controller programming model that enables flexible and efficient orchestration of complex RLHF workflows, effectively mitigating the bottlenecks associated with centralized controller architectures and facilitating scalability in large-scale data scenarios. In addition, we propose a dynamic placement schema that adaptively partitions computational resources and schedules workloads, thereby significantly reducing hardware idle time and improving GPU utilization under variable training conditions. We evaluate WeChat-YATT across a range of experimental scenarios, demonstrating that it achieves substantial improvements in throughput compared to state-of-the-art RLHF training frameworks. Furthermore, WeChat-YATT has been successfully deployed to train models supporting WeChat product features for a large-scale user base, underscoring its effectiveness and robustness in real-world applications.We have open-source WeChat-YATT at https://www.github.com/tencent/WeChat-YATT.
Authors: Supriyo Chakraborty, Krishnan Raghavan
Abstract: The stability-plasticity dilemma, closely related to a neural network's (NN) capacity-its ability to represent tasks-is a fundamental challenge in continual learning (CL). Within this context, we introduce CL's effective model capacity (CLEMC) that characterizes the dynamic behavior of the stability-plasticity balance point. We develop a difference equation to model the evolution of the interplay between the NN, task data, and optimization procedure. We then leverage CLEMC to demonstrate that the effective capacity-and, by extension, the stability-plasticity balance point is inherently non-stationary. We show that regardless of the NN architecture or optimization method, a NN's ability to represent new tasks diminishes when incoming task distributions differ from previous ones. We conduct extensive experiments to support our theoretical findings, spanning a range of architectures-from small feedforward network and convolutional networks to medium-sized graph neural networks and transformer-based large language models with millions of parameters.
Authors: Guangyin Jin, Sicong Lai, Xiaoshuai Hao, Mingtao Zhang, Jinlei Zhang
Abstract: Achieving accurate traffic prediction is a fundamental but crucial task in the development of current intelligent transportation systems.Most of the mainstream methods that have made breakthroughs in traffic prediction rely on spatio-temporal graph neural networks, spatio-temporal attention mechanisms, etc. The main challenges of the existing deep learning approaches are that they either depend on a complete traffic network structure or require intricate model designs to capture complex spatio-temporal dependencies. These limitations pose significant challenges for the efficient deployment and operation of deep learning models on large-scale datasets. To address these challenges, we propose a cost-effective graph-free Multilayer Perceptron (MLP) based model M3-Net for traffic prediction. Our proposed model not only employs time series and spatio-temporal embeddings for efficient feature processing but also first introduces a novel MLP-Mixer architecture with a mixture of experts (MoE) mechanism. Extensive experiments conducted on multiple real datasets demonstrate the superiority of the proposed model in terms of prediction performance and lightweight deployment.
Authors: Shuyi Zhang, Wei Shi, Sihang Li, Jiayi Liao, Tao Liang, Hengxing Cai, Xiang Wang
Abstract: Large language models (LLMs) have been widely deployed across numerous fields. Reinforcement Learning from Human Feedback (RLHF) leverages reward models (RMs) as proxies for human preferences to align LLM behaviors with human values, making the accuracy, reliability, and interpretability of RMs critical for effective alignment. However, traditional RMs lack interpretability, offer limited insight into the reasoning behind reward assignments, and are inflexible toward user preference shifts. While recent multidimensional RMs aim for improved interpretability, they often fail to provide feature-level attribution and require costly annotations. To overcome these limitations, we introduce the Sparse Autoencoder-enhanced Reward Model (SARM), a novel architecture that integrates a pretrained Sparse Autoencoder (SAE) into a reward model. SARM maps the hidden activations of LLM-based RM into an interpretable, sparse, and monosemantic feature space, from which a scalar head aggregates feature activations to produce transparent and conceptually meaningful reward scores. Empirical evaluations demonstrate that SARM facilitates direct feature-level attribution of reward assignments, allows dynamic adjustment to preference shifts, and achieves superior alignment performance compared to conventional reward models. Our code is available at https://github.com/schrieffer-z/sarm.
Authors: Shugang Hao, Hongbo Li, Lingjie Duan
Abstract: The binary exponential backoff scheme is widely used in WiFi 7 and still incurs poor throughput performance under dynamic channel environments. Recent model-based approaches (e.g., non-persistent and $p$-persistent CSMA) simply optimize backoff strategies under a known and fixed node density, still leading to a large throughput loss due to inaccurate node density estimation. This paper is the first to propose LLM transformer-based in-context learning (ICL) theory for optimizing channel access. We design a transformer-based ICL optimizer to pre-collect collision-threshold data examples and a query collision case. They are constructed as a prompt as the input for the transformer to learn the pattern, which then generates a predicted contention window threshold (CWT). To train the transformer for effective ICL, we develop an efficient algorithm and guarantee a near-optimal CWT prediction within limited training steps. As it may be hard to gather perfect data examples for ICL in practice, we further extend to allow erroneous data input in the prompt. We prove that our optimizer maintains minimal prediction and throughput deviations from the optimal values. Experimental results on NS-3 further demonstrate our approach's fast convergence and near-optimal throughput over existing model-based and DRL-based approaches under unknown node densities.
Authors: Siwen Jiao, Kangan Qian, Hao Ye, Yang Zhong, Ziang Luo, Sicong Jiang, Zilin Huang, Yangyi Fang, Jinyu Miao, Zheng Fu, Yunlong Wang, Kun Jiang, Diange Yang, Rui Fan, Baoyun Peng
Abstract: Autonomous driving faces significant challenges in achieving human-like iterative decision-making, which continuously generates, evaluates, and refines trajectory proposals. Current generation-evaluation frameworks isolate trajectory generation from quality assessment, preventing iterative refinement essential for planning, while reinforcement learning methods collapse multi-dimensional preferences into scalar rewards, obscuring critical trade-offs and yielding scalarization bias.To overcome these issues, we present EvaDrive, a novel multi-objective reinforcement learning framework that establishes genuine closed-loop co-evolution between trajectory generation and evaluation via adversarial optimization. EvaDrive frames trajectory planning as a multi-round adversarial game. In this game, a hierarchical generator continuously proposes candidate paths by combining autoregressive intent modeling for temporal causality with diffusion-based refinement for spatial flexibility. These proposals are then rigorously assessed by a trainable multi-objective critic that explicitly preserves diverse preference structures without collapsing them into a single scalarization bias.This adversarial interplay, guided by a Pareto frontier selection mechanism, enables iterative multi-round refinement, effectively escaping local optima while preserving trajectory diversity.Extensive experiments on NAVSIM and Bench2Drive benchmarks demonstrate SOTA performance, achieving 94.9 PDMS on NAVSIM v1 (surpassing DiffusionDrive by 6.8, DriveSuprim by 5.0, and TrajHF by 0.9) and 64.96 Driving Score on Bench2Drive. EvaDrive generates diverse driving styles via dynamic weighting without external preference data, introducing a closed-loop adversarial framework for human-like iterative decision-making, offering a novel scalarization-free trajectory optimization approach.
Authors: Rilwan Umar, Aydin Abadi, Basil Aldali, Benito Vincent, Elliot A. J. Hurley, Hotoon Aljazaeri, Jamie Hedley-Cook, Jamie-Lee Bell, Lambert Uwuigbusun, Mujeeb Ahmed, Shishir Nagaraja, Suleiman Sabo, Weaam Alrbeiqi
Abstract: Weather forecasting plays a vital role in disaster preparedness, agriculture, and resource management, yet current centralized forecasting systems are increasingly strained by security vulnerabilities, limited scalability, and susceptibility to single points of failure. To address these challenges, we propose a decentralized weather forecasting framework that integrates Federated Learning (FL) with blockchain technology. FL enables collaborative model training without exposing sensitive local data; this approach enhances privacy and reduces data transfer overhead. Meanwhile, the Ethereum blockchain ensures transparent and dependable verification of model updates. To further enhance the system's security, we introduce a reputation-based voting mechanism that assesses the trustworthiness of submitted models while utilizing the Interplanetary File System (IPFS) for efficient off-chain storage. Experimental results demonstrate that our approach not only improves forecasting accuracy but also enhances system resilience and scalability, making it a viable candidate for deployment in real-world, security-critical environments.
Authors: Vibhas K. Vats, Sripad Joshi, David J. Crandall, Md. Alimoor Reza, Soon-heung Jung
Abstract: Traditional multi-view stereo (MVS) methods rely heavily on photometric and geometric consistency constraints, but newer machine learning-based MVS methods check geometric consistency across multiple source views only as a post-processing step. In this paper, we present a novel approach that explicitly encourages geometric consistency of reference view depth maps across multiple source views at different scales during learning (see Fig. 1). We find that adding this geometric consistency loss significantly accelerates learning by explicitly penalizing geometrically inconsistent pixels, reducing the training iteration requirements to nearly half that of other MVS methods. Our extensive experiments show that our approach achieves a new state-of-the-art on the DTU and BlendedMVS datasets, and competitive results on the Tanks and Temples benchmark. To the best of our knowledge, GC-MVSNet is the first attempt to enforce multi-view, multi-scale geometric consistency during learning.
Authors: Yuma Ichikawa, Hiroaki Iwashita
Abstract: Finding the optimal solution is often the primary goal in combinatorial optimization (CO). However, real-world applications frequently require diverse solutions rather than a single optimum, particularly in two key scenarios. The first scenario occurs in real-world applications where strictly enforcing every constraint is neither necessary nor desirable. Allowing minor constraint violations can often lead to more cost-effective solutions. This is typically achieved by incorporating the constraints as penalty terms in the objective function, which requires careful tuning of penalty parameters. The second scenario involves cases where CO formulations tend to oversimplify complex real-world factors, such as domain knowledge, implicit trade-offs, or ethical considerations. To address these challenges, generating (i) penalty-diversified solutions by varying penalty intensities and (ii) variation-diversified solutions with distinct structural characteristics provides valuable insights, enabling practitioners to post-select the most suitable solution for their specific needs. However, efficiently discovering these diverse solutions is more challenging than finding a single optimal one. This study introduces Continual Parallel Relaxation Annealing (CPRA), a computationally efficient framework for unsupervised-learning (UL)-based CO solvers that generates diverse solutions within a single training run. CPRA leverages representation learning and parallelization to automatically discover shared representations, substantially accelerating the search for these diverse solutions. Numerical experiments demonstrate that CPRA outperforms existing UL-based solvers in generating these diverse solutions while significantly reducing computational costs.
Authors: Rui Wang, Yuesheng Xu, Mingsong Yan
Abstract: This paper introduces a hypothesis space for deep learning based on deep neural networks (DNNs). By treating a DNN as a function of two variables - the input variable and the parameter variable - we consider the set of DNNs where the parameter variable belongs to a space of weight matrices and biases determined by a prescribed depth and layer widths. To construct a Banach space of functions of the input variable, we take the weak* closure of the linear span of this DNN set. We prove that the resulting Banach space is a reproducing kernel Banach space (RKBS) and explicitly construct its reproducing kernel. Furthermore, we investigate two learning models - regularized learning and the minimum norm interpolation (MNI) problem - within the RKBS framework by establishing representer theorems. These theorems reveal that the solutions to these learning problems can be expressed as a finite sum of kernel expansions based on training data.
Authors: Weiye Zhao, Feihan Li, Changliu Liu
Abstract: Deep reinforcement learning (DRL) has demonstrated remarkable performance in many continuous control tasks. However, a significant obstacle to the real-world application of DRL is the lack of safety guarantees. Although DRL agents can satisfy system safety in expectation through reward shaping, designing agents to consistently meet hard constraints (e.g., safety specifications) at every time step remains a formidable challenge. In contrast, existing work in the field of safe control provides guarantees on persistent satisfaction of hard safety constraints. However, these methods require explicit analytical system dynamics models to synthesize safe control, which are typically inaccessible in DRL settings. In this paper, we present a model-free safe control algorithm, the implicit safe set algorithm, for synthesizing safeguards for DRL agents that ensure provable safety throughout training. The proposed algorithm synthesizes a safety index (barrier certificate) and a subsequent safe control law solely by querying a black-box dynamic function (e.g., a digital twin simulator). Moreover, we theoretically prove that the implicit safe set algorithm guarantees finite time convergence to the safe set and forward invariance for both continuous-time and discrete-time systems. We validate the proposed algorithm on the state-of-the-art Safety Gym benchmark, where it achieves zero safety violations while gaining $95\% \pm 9\%$ cumulative reward compared to state-of-the-art safe DRL methods. Furthermore, the resulting algorithm scales well to high-dimensional systems with parallel computing.
Authors: Man-Chung Yue, Yves Rychener, Daniel Kuhn, Viet Anh Nguyen
Abstract: The state-of-the-art methods for estimating high-dimensional covariance matrices all shrink the eigenvalues of the sample covariance matrix towards a data-insensitive shrinkage target. The underlying shrinkage transformation is either chosen heuristically - without compelling theoretical justification - or optimally in view of restrictive distributional assumptions. In this paper, we propose a principled approach to construct covariance estimators without imposing restrictive assumptions. That is, we study distributionally robust covariance estimation problems that minimize the worst-case Frobenius error with respect to all data distributions close to a nominal distribution, where the proximity of distributions is measured via a divergence on the space of covariance matrices. We identify mild conditions on this divergence under which the resulting minimizers represent shrinkage estimators. We show that the corresponding shrinkage transformations are intimately related to the geometrical properties of the underlying divergence. We also prove that our robust estimators are efficiently computable and asymptotically consistent and that they enjoy finite-sample performance guarantees. We exemplify our general methodology by synthesizing explicit estimators induced by the Kullback-Leibler, Fisher-Rao, and Wasserstein divergences. Numerical experiments based on synthetic and real data show that our robust estimators are competitive with state-of-the-art estimators.
Authors: Fran\c{c}ois Bachoc, Tommaso Cesari, Roberto Colomboni
Abstract: We study the role of contextual information in the online learning problem of brokerage between traders. In this sequential problem, at each time step, two traders arrive with secret valuations about an asset they wish to trade. The learner (a broker) suggests a trading (or brokerage) price based on contextual data about the asset and the market conditions. Then, the traders reveal their willingness to buy or sell based on whether their valuations are higher or lower than the brokerage price. A trade occurs if one of the two traders decides to buy and the other to sell, i.e., if the broker's proposed price falls between the smallest and the largest of their two valuations. We design algorithms for this problem and prove optimal theoretical regret guarantees under various standard assumptions.
Authors: Simon Hirsch, Jonathan Berrisch, Florian Ziel
Abstract: Large-scale streaming data are common in modern machine learning applications and have led to the development of online learning algorithms. Many fields, such as supply chain management, weather and meteorology, energy markets, and finance, have pivoted towards using probabilistic forecasts. This results in the need not only for accurate learning of the expected value but also for learning the conditional heteroskedasticity and conditional moments. Against this backdrop, we present a methodology for online estimation of regularized, linear distributional models. The proposed algorithm is based on a combination of recent developments for the online estimation of LASSO models and the well-known GAMLSS framework. We provide a case study on day-ahead electricity price forecasting, in which we show the competitive performance of the incremental estimation combined with strongly reduced computational effort. Our algorithms are implemented in a computationally efficient Python package ondil.
Authors: Yingzhen Yang, Ping Li
Abstract: We study nonparametric regression using an over-parameterized two-layer neural networks trained with algorithmic guarantees in this paper. We consider the setting where the training features are drawn uniformly from the unit sphere in $\RR^d$, and the target function lies in an interpolation space commonly studied in statistical learning theory. We demonstrate that training the neural network with a novel Preconditioned Gradient Descent (PGD) algorithm, equipped with early stopping, achieves a sharp regression rate of $\cO(n^{-\frac{2\alpha s'}{2\alpha s'+1}})$ when the target function is in the interpolation space $\bth{\cH_K}^{s'}$ with $s' \ge 3$. This rate is even sharper than the currently known nearly-optimal rate of $\cO(n^{-\frac{2\alpha s'}{2\alpha s'+1}})\log^2(1/\delta)$, where $n$ is the size of the training data and $\delta \in (0,1)$ is a small probability. This rate is also sharper than the standard kernel regression rate of $\cO(n^{-\frac{2\alpha}{2\alpha+1}})$ obtained under the regular Neural Tangent Kernel (NTK) regime when training the neural network with the vanilla gradient descent (GD), where $2\alpha = d/(d-1)$. Our analysis is based on two key technical contributions. First, we present a principled decomposition of the network output at each PGD step into a function in the reproducing kernel Hilbert space (RKHS) of a newly induced integral kernel, and a residual function with small $L^{\infty}$-norm. Second, leveraging this decomposition, we apply local Rademacher complexity theory to tightly control the complexity of the function class comprising all the neural network functions obtained in the PGD iterates. Our results further suggest that PGD enables the neural network to escape the linear NTK regime and achieve improved generalization, as it effectively induces a new kernel termed the integral kernel, compared to the regular NTK arising from the vanilla GD.
Authors: Sai Sathiesh Rajan, Ezekiel Soremekun, Sudipta Chattopadhyay
Abstract: In this work, we systematically expose and measure the inconsistency and knowledge gaps of Large Language Models (LLMs). Specifically, we propose an automated testing framework (called KonTest) which leverages a knowledge graph to construct test cases. KonTest probes and measures the inconsistencies in the LLM's knowledge of the world via a combination of semantically-equivalent queries and test oracles (metamorphic or ontological oracle). KonTest further mitigates knowledge gaps via a weighted LLM model ensemble. Using four state-of-the-art LLMs (Falcon, Gemini, GPT3.5, and Llama2), we show that KonTest generates 19.2% error inducing inputs (1917 errors from 9979 test inputs). It also reveals a 16.5% knowledge gap across all tested LLMs. A mitigation method informed by KonTest's test suite reduces LLM knowledge gap by 32.48%. Our ablation study further shows that GPT3.5 is not suitable for knowledge-based consistency testing because it is only 60%-68% effective in knowledge construction.
Authors: Alex Clinton, Yiding Chen, Xiaojin Zhu, Kirthevasan Kandasamy
Abstract: We study a collaborative learning problem where $m$ agents aim to estimate a vector $\mu =(\mu_1,\ldots,\mu_d)\in \mathbb{R}^d$ by sampling from associated univariate normal distributions $\{\mathcal{N}(\mu_k, \sigma^2)\}_{k\in[d]}$. Agent $i$ incurs a cost $c_{i,k}$ to sample from $\mathcal{N}(\mu_k, \sigma^2)$. Instead of working independently, agents can exchange data, collecting cheaper samples and sharing them in return for costly data, thereby reducing both costs and estimation error. We design a mechanism to facilitate such collaboration, while addressing two key challenges: ensuring individually rational (IR) and fair outcomes so all agents benefit, and preventing strategic behavior (e.g. non-collection, data fabrication) to avoid socially undesirable outcomes. We design a mechanism and an associated Nash equilibrium (NE) which minimizes the social penalty-sum of agents' estimation errors and collection costs-while being IR for all agents. We achieve a $\mathcal{O}(\sqrt{m})$-approximation to the minimum social penalty in the worst case and an $\mathcal{O}(1)$-approximation under favorable conditions. Additionally, we establish three hardness results: no nontrivial mechanism guarantees (i) a dominant strategy equilibrium where agents report truthfully, (ii) is IR for every strategy profile of other agents, (iii) or avoids a worst-case $\Omega(\sqrt{m})$ price of stability in any NE. Finally, by integrating concepts from axiomatic bargaining, we demonstrate that our mechanism supports fairer outcomes than one which minimizes social penalty.
Authors: Aditya Ashvin, Rimita Lahiri, Aditya Kommineni, Somer Bishop, Catherine Lord, Sudarsana Reddy Kadiri, Shrikanth Narayanan
Abstract: Reliable transcription of child-adult conversations in clinical settings is crucial for diagnosing developmental disorders like Autism. Recent advances in deep learning and availability of large scale transcribed data has led to development of speech foundation models that have shown dramatic improvements in ASR performance. However, their performance on conversational child-adult interactions remains underexplored. In this work, we provide a comprehensive evaluation of ASR performance on a dataset containing child-adult interactions from autism diagnostic sessions, using Whisper, Wav2Vec2, HuBERT, and WavLM. We find that speech foundation models show a noticeable performance drop (15-20% absolute WER) for child speech compared to adult speech in the conversational setting. Then, we fine-tune the best-performing zero-shot model (Whisper-large) using LoRA in a low-resource setting, yielding 8% and 13% absolute WER improvements for child and adult speech, respectively.
Authors: Yannis Montreuil, Shu Heng Yeo, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi
Abstract: The Two-Stage Learning-to-Defer (L2D) framework has been extensively studied for classification and, more recently, regression tasks. However, many real-world applications require solving both tasks jointly in a multi-task setting. We introduce a novel Two-Stage L2D framework for multi-task learning that integrates classification and regression through a unified deferral mechanism. Our method leverages a two-stage surrogate loss family, which we prove to be both Bayes-consistent and $(\mathcal{G}, \mathcal{R})$-consistent, ensuring convergence to the Bayes-optimal rejector. We derive explicit consistency bounds tied to the cross-entropy surrogate and the $L_1$-norm of agent-specific costs, and extend minimizability gap analysis to the multi-expert two-stage regime. We also make explicit how shared representation learning -- commonly used in multi-task models -- affects these consistency guarantees. Experiments on object detection and electronic health record analysis demonstrate the effectiveness of our approach and highlight the limitations of existing L2D methods in multi-task scenarios.
Authors: Junho Kim, Hyungjin Chung, Byung-Hoon Kim
Abstract: Category-agnostic pose estimation (CAPE) has traditionally relied on support images with annotated keypoints, a process that is often cumbersome and may fail to fully capture the necessary correspondences across diverse object categories. Recent efforts have explored the use of text queries, leveraging their enhanced stability and generalization capabilities. However, existing approaches often remain constrained by their reliance on support queries, their failure to fully utilize the rich priors embedded in pre-trained large language models, and the limitations imposed by their parametric distribution assumptions. To address these challenges, we introduce CapeLLM, the first multimodal large language model (MLLM) designed for CAPE. Our method only employs query image and detailed text descriptions as an input to estimate category-agnostic keypoints. Our method encompasses effective training strategies and carefully designed instructions for applying the MLLM to CAPE. Moreover, we propose an inference mechanism that further enhances the reasoning process for unseen keypoints. while flexibly modeling their underlying spatial distribution and uncertainty, allowing for adaptive refinement based on contextual cues. We conducted extensive experiments to apply the MLLM to CAPE effectively, focusing not only on the model architecture and prompt design but also on ensuring robustness across input variations. Our approach sets a new state-of-the-art on the MP-100 benchmark in the 1-shot and even 5-shot setting, marking a significant advancement in the field of category-agnostic pose estimation. Code is available at https://github.com/Junhojuno/CapeLLM.
Authors: Vasisht Duddu, Rui Zhang, N. Asokan
Abstract: Machine learning (ML) defenses protect against various risks to security, privacy, and fairness. Real-life models need simultaneous protection against multiple different risks which necessitates combining multiple defenses. But combining defenses with conflicting interactions in an ML model can be ineffective, incurring a significant drop in the effectiveness of one or more defenses being combined. Practitioners need a way to determine if a given combination can be effective. Experimentally identifying effective combinations can be time-consuming and expensive, particularly when multiple defenses need to be combined. We need an inexpensive, easy-to-use combination technique to identify effective combinations. Ideally, a combination technique should be (a) accurate (correctly identifies whether a combination is effective or not), (b) scalable (allows combining multiple defenses), (c) non-invasive (requires no change to the defenses being combined), and (d) general (is applicable to different types of defenses). Prior works have identified several ad-hoc techniques but none satisfy all the requirements above. We propose a principled combination technique, Def\Con, to identify effective defense combinations. Def\Con meets all requirements, achieving 90% accuracy on eight combinations explored in prior work and 81% in 30 previously unexplored combinations that we empirically evaluate in this paper.
Authors: Heehwan Wang, Joonwoo Kwon, Sooyoung Kim, Shinjae Yoo, Yuewei Lin, Jiook Cha
Abstract: Music style transfer enables personalized music creation by combining the structure of one piece with the stylistic characteristics of another. While recent approaches have explored text-conditioned generation and diffusion-based synthesis, most require extensive training, paired datasets, or detailed textual annotations. In this work, we introduce Stylus, a novel training-free framework for music style transfer that directly manipulates the self-attention layers of a pre-trained Latent Diffusion Model (LDM). Operating in the mel-spectrogram domain, Stylus transfers musical style by replacing key and value representations from the content audio with those of the style reference, without any fine-tuning. To enhance stylization quality and controllability, we further incorporate query preservation, CFG-inspired guidance scaling, multi-style interpolation, and phase-preserving reconstruction. Our method significantly improves perceptual quality and structural preservation compared to prior work, while remaining lightweight and easy to deploy. This work highlights the potential of diffusion-based attention manipulation for efficient, high-fidelity, and interpretable music generation-without training. Codes will be released upon acceptance.
Authors: Jan Rathjens, Shirin Reyhanian, David Kappel, Laurenz Wiskott
Abstract: Understanding the mechanisms underlying deep neural networks remains a fundamental challenge in machine learning and computer vision. One promising, yet only preliminarily explored approach, is feature inversion, which attempts to reconstruct images from intermediate representations using trained inverse neural networks. In this study, we revisit feature inversion, introducing a novel, modular variation that enables significantly more efficient application of the technique. We demonstrate how our method can be systematically applied to the large-scale transformer-based vision models, Detection Transformer and Vision Transformer, and how reconstructed images can be qualitatively interpreted in a meaningful way. We further quantitatively evaluate our method, thereby uncovering underlying mechanisms of representing image features that emerge in the two transformer architectures. Our analysis reveals key insights into how these models encode contextual shape and image details, how their layers correlate, and their robustness against color perturbations. These findings contribute to a deeper understanding of transformer-based vision models and their internal representations. The code for reproducing our experiments is available at github.com/wiskott-lab/inverse-tvm.
Authors: Felipe Montealegre-Mora, Carl Boettiger, Carl J. Walters, Christopher L. Cahill
Abstract: In fishery science, harvest management of size-structured stochastic populations is a long-standing and difficult problem. Rectilinear precautionary policies based on biomass and harvesting reference points have now become a standard approach to this problem. While these standard feedback policies are adapted from analytical or dynamic programming solutions assuming relatively simple ecological dynamics, they are often applied to more complicated ecological settings in the real world. In this paper we explore the problem of designing harvest control rules for partially observed, age-structured, spasmodic fish populations using tools from reinforcement learning (RL) and Bayesian optimization. Our focus is on the case of Walleye fisheries in Alberta, Canada, whose highly variable recruitment dynamics have perplexed managers and ecologists. We optimized and evaluated policies using several complementary performance metrics. The main questions we addressed were: 1. How do standard policies based on reference points perform relative to numerically optimized policies? 2. Can an observation of mean fish weight, in addition to stock biomass, aid policy decisions?
Authors: Xianchao Xiu, Chenyi Huang, Pan Shang, Wanquan Liu
Abstract: To deal with high-dimensional unlabeled datasets in many areas, principal component analysis (PCA) has become a rising technique for unsupervised feature selection (UFS). However, most existing PCA-based methods only consider the structure of datasets by embedding a single sparse regularization or constraint on the transformation matrix. In this paper, we introduce a novel bi-sparse method called BSUFS to improve the performance of UFS. The core idea of BSUFS is to incorporate $\ell_{2,p}$-norm and $\ell_q$-norm into the classical PCA, which enables our method to select relevant features and filter out irrelevant noises, thereby obtaining discriminative features. Here, the parameters $p$ and $q$ are within the range of $[0, 1)$. Therefore, BSUFS not only constructs a unified framework for bi-sparse optimization, but also includes some existing works as special cases. To solve the resulting non-convex model, we propose an efficient proximal alternating minimization (PAM) algorithm using Stiefel manifold optimization and sparse optimization techniques. In addition, the computational complexity analysis is presented. Extensive numerical experiments on synthetic and real-world datasets demonstrate the effectiveness of our proposed BSUFS. The results reveal the advantages of bi-sparse optimization in feature selection and show its potential for other fields in image processing. Our code is available at https://github.com/xianchaoxiu/BSUFS.
Authors: Arun K. Sharma, Shubhobrata Bhattacharya, Motahar Reza, Bishakh Bhattacharya
Abstract: Traditional biometric systems have encountered significant setbacks due to various unavoidable factors, for example, wearing of face masks in face recognition-based biometrics and hygiene concerns in fingerprint-based biometrics. This paper proposes a novel lightweight vision transformer with phase-only cross-attention (POC-ViT) using dual biometric traits of forehead and periocular portions of the face, capable of performing well even with face masks and without any physical touch, offering a promising alternative to traditional methods. The POC-ViT framework is designed to handle two biometric traits and to capture inter-dependencies in terms of relative structural patterns. Each channel consists of a Cross-Attention using phase-only correlation (POC) that captures both their individual and correlated structural patterns. The computation of cross-attention using POC extracts the phase correlation in the spatial features. Therefore, it is robust against variations in resolution and intensity, as well as illumination changes in the input images. The lightweight model is suitable for edge device deployment. The performance of the proposed framework was successfully demonstrated using the Forehead Subcutaneous Vein Pattern and Periocular Biometric Pattern (FSVP-PBP) database, having 350 subjects. The POC-ViT framework outperformed state-of-the-art methods with an outstanding classification accuracy of $98.8\%$ with the dual biometric traits.
Authors: Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi
Abstract: Two-stage Learning-to-Defer (L2D) enables optimal task delegation by assigning each input to either a fixed main model or one of several offline experts, supporting reliable decision-making in complex, multi-agent environments. However, existing L2D frameworks assume clean inputs and are vulnerable to adversarial perturbations that can manipulate query allocation--causing costly misrouting or expert overload. We present the first comprehensive study of adversarial robustness in two-stage L2D systems. We introduce two novel attack strategie--untargeted and targeted--which respectively disrupt optimal allocations or force queries to specific agents. To defend against such threats, we propose SARD, a convex learning algorithm built on a family of surrogate losses that are provably Bayes-consistent and $(\mathcal{R}, \mathcal{G})$-consistent. These guarantees hold across classification, regression, and multi-task settings. Empirical results demonstrate that SARD significantly improves robustness under adversarial attacks while maintaining strong clean performance, marking a critical step toward secure and trustworthy L2D deployment.
Authors: Wonduk Seo, Hyunjin An, Seunghyun Lee
Abstract: Query expansion is widely used in Information Retrieval (IR) to improve search outcomes by supplementing initial queries with richer information. While recent Large Language Model (LLM) based methods generate pseudo-relevant content and expanded terms via multiple prompts, they often yield homogeneous, narrow expansions that lack the diverse context needed to retrieve relevant information. In this paper, we propose AMD: a new Agent-Mediated Dialogic Framework that engages in a dialogic inquiry involving three specialized roles: (1) a Socratic Questioning Agent reformulates the initial query into three sub-questions, with each question inspired by a specific Socratic questioning dimension, including clarification, assumption probing, and implication probing, (2) a Dialogic Answering Agent generates pseudo-answers, enriching the query representation with multiple perspectives aligned to the user's intent, and (3) a Reflective Feedback Agent evaluates and refines these pseudo-answers, ensuring that only the most relevant and informative content is retained. By leveraging a multi-agent process, AMD effectively crafts richer query representations through inquiry and feedback refinement. Extensive experiments on benchmarks including BEIR and TREC demonstrate that our framework outperforms previous methods, offering a robust solution for retrieval tasks.
Authors: Ragini Gupta, Shinan Liu, Ruixiao Zhang, Xinyue Hu, Xiaoyang Wang, Hadjer Benkraouda, Pranav Kommaraju, Phuong Cao, Nick Feamster, Klara Nahrstedt
Abstract: Machine learning has shown promise in network intrusion detection systems, yet its performance often degrades due to concept drift and imbalanced data. These challenges are compounded by the labor-intensive process of labeling network traffic, especially when dealing with evolving and rare attack types, which makes preparing the right data for adaptation difficult. To address these issues, we propose a generative active adaptation framework that minimizes labeling effort while enhancing model robustness. Our approach employs density-aware dataset prior selection to identify the most informative samples for annotation, and leverages deep generative models to conditionally synthesize diverse samples, thereby augmenting the training set and mitigating the effects of concept drift. We evaluate our end-to-end framework \NetGuard on both simulated IDS data and a real-world ISP dataset, demonstrating significant improvements in intrusion detection performance. Our method boosts the overall F1-score from 0.60 (without adaptation) to 0.86. Rare attacks such as Infiltration, Web Attack, and FTP-BruteForce, which originally achieved F1 scores of 0.001, 0.04, and 0.00, improve to 0.30, 0.50, and 0.71, respectively, with generative active adaptation in the CIC-IDS 2018 dataset. Our framework effectively enhances rare attack detection while reducing labeling costs, making it a scalable and practical solution for intrusion detection.
Authors: Amr Mousa, Neil Karavis, Michele Caprio, Wei Pan, Richard Allmendinger
Abstract: Quadrupedal locomotion via Reinforcement Learning (RL) is commonly addressed using the teacher-student paradigm, where a privileged teacher guides a proprioceptive student policy. However, key challenges such as representation misalignment between privileged teacher and proprioceptive-only student, covariate shift due to behavioral cloning, and lack of deployable adaptation; lead to poor generalization in real-world scenarios. We propose Teacher-Aligned Representations via Contrastive Learning (TAR), a framework that leverages privileged information with self-supervised contrastive learning to bridge this gap. By aligning representations to a privileged teacher in simulation via contrastive objectives, our student policy learns structured latent spaces and exhibits robust generalization to Out-of-Distribution (OOD) scenarios, surpassing the fully privileged "Teacher". Results showed accelerated training by 2x compared to state-of-the-art baselines to achieve peak performance. OOD scenarios showed better generalization by 40% on average compared to existing methods. Moreover, TAR transitions seamlessly into learning during deployment without requiring privileged states, setting a new benchmark in sample-efficient, adaptive locomotion and enabling continual fine-tuning in real-world scenarios. Open-source code and videos are available at https://amrmousa.com/TARLoco/.
Authors: Yuping Wang, Xiangyu Huang, Xiaokang Sun, Mingxuan Yan, Shuo Xing, Zhengzhong Tu, Jiachen Li
Abstract: We introduce UniOcc, a comprehensive, unified benchmark and toolkit for occupancy forecasting (i.e., predicting future occupancies based on historical information) and occupancy prediction (i.e., predicting current-frame occupancy from camera images. UniOcc unifies the data from multiple real-world datasets (i.e., nuScenes, Waymo) and high-fidelity driving simulators (i.e., CARLA, OpenCOOD), providing 2D/3D occupancy labels and annotating innovative per-voxel flows. Unlike existing studies that rely on suboptimal pseudo labels for evaluation, UniOcc incorporates novel evaluation metrics that do not depend on ground-truth labels, enabling robust assessment on additional aspects of occupancy quality. Through extensive experiments on state-of-the-art models, we demonstrate that large-scale, diverse training data and explicit flow information significantly enhance occupancy prediction and forecasting performance. Our data and code are available at https://uniocc.github.io/.
Authors: Xingshan Zeng, Weiwen Liu, Xu Huang, Zezhong Wang, Lingzhi Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Ruiming Tang, Qun Liu
Abstract: Tool learning, which allows Large Language Models (LLMs) to leverage external tools for solving complex user tasks, has emerged as a promising avenue for extending model capabilities. However, existing approaches primarily focus on data synthesis for fine-tuning LLMs to invoke tools effectively, largely ignoring how to fully stimulate the potential of the model. In this paper, we propose ToolACE-R, a novel framework that includes both model-aware iterative training and adaptive refinement for tool learning. ToolACE-R features a model-aware iterative training procedure that progressively adjust training samples based on the model's evolving capabilities to maximize its potential. Additionally, it incorporates self-refinement training corpus which emphasizes LLM's ability to iteratively refine their tool calls, optimizing performance without requiring external feedback. Furthermore, we introduce adaptive self-refinement mechanism for efficient test-time scaling, where the trained model can autonomously determine when to stop the process based on iterative self-refinement. We conduct extensive experiments across several benchmark datasets, showing that ToolACE-R achieves competitive performance compared to advanced API-based models. The performance of tool invocation can be further improved efficiently through adaptive self-refinement. These results highlight the effectiveness and generalizability of ToolACE-R, offering a promising direction for more efficient and scalable tool learning.
Authors: Eugen Barbulescu, Antonio Alexoaie, Lucian Busoniu
Abstract: Network pruning is used to reduce inference latency and power consumption in large neural networks. However, most existing methods use ad-hoc heuristics, lacking much insight and justified mainly by empirical results. We introduce Hyperflux, a conceptually grounded L0 pruning approach that estimates each weight's importance through its flux, the gradient's response to the weight's removal. A global pressure term continuously drives all weights toward pruning, with those critical for accuracy being automatically regrown based on their flux. We postulate several properties that naturally follow from our framework and experimentally validate each of them. One such property is the relationship between final sparsity and pressure, for which we derive a generalized scaling-law equation that is used to design our sparsity-controlling scheduler. Empirically, we demonstrate state-of-the-art results with ResNet-50 and VGG-19 on CIFAR-10 and CIFAR-100.
Authors: Junmo Kim, Namkyeong Lee, Jiwon Kim, Kwangsoo Kim
Abstract: Electronic health record (EHR) foundation models have been an area ripe for exploration with their improved performance in various medical tasks. Despite the rapid advances, there exists a fundamental limitation: Processing unseen medical codes out of vocabulary. This problem limits the generalizability of EHR foundation models and the integration of models trained with different vocabularies. To alleviate this problem, we propose a set of novel medical concept representations (MedRep) for EHR foundation models based on the observational medical outcome partnership (OMOP) common data model (CDM). For concept representation learning, we enrich the information of each concept with a minimal definition through large language model (LLM) prompts and complement the text-based representations through the graph ontology of OMOP vocabulary. Our approach outperforms the vanilla EHR foundation model and the model with a previously introduced medical code tokenizer in diverse prediction tasks. We also demonstrate the generalizability of MedRep through external validation.
Authors: Xinyu Wang, Jijun Chi, Zhenghan Tai, Tung Sum Thomas Kwok, Muzhi Li, Zhuhong Li, Hailin He, Yuchen Hua, Peng Lu, Suyuchen Wang, Yihong Wu, Jerry Huang, Jingrui Tian, Fengran Mo, Yufei Cui, Ling Zhou
Abstract: Leveraging large language models in real-world settings often entails a need to utilize domain-specific data and tools in order to follow the complex regulations that need to be followed for acceptable use. Within financial sectors, modern enterprises increasingly rely on Retrieval-Augmented Generation (RAG) systems to address complex compliance requirements in financial document workflows. However, existing solutions struggle to account for the inherent heterogeneity of data (e.g., text, tables, diagrams) and evolving nature of regulatory standards used in financial filings, leading to compromised accuracy in critical information extraction. We propose the FinSage framework as a solution, utilizing a multi-aspect RAG framework tailored for regulatory compliance analysis in multi-modal financial documents. FinSage introduces three innovative components: (1) a multi-modal pre-processing pipeline that unifies diverse data formats and generates chunk-level metadata summaries, (2) a multi-path sparse-dense retrieval system augmented with query expansion (HyDE) and metadata-aware semantic search, and (3) a domain-specialized re-ranking module fine-tuned via Direct Preference Optimization (DPO) to prioritize compliance-critical content. Extensive experiments demonstrate that FinSage achieves an impressive recall of 92.51% on 75 expert-curated questions derived from surpasses the best baseline method on the FinanceBench question answering datasets by 24.06% in accuracy. Moreover, FinSage has been successfully deployed as financial question-answering agent in online meetings, where it has already served more than 1,200 people.
Authors: Shotaro Takasu, Toshio Aoyagi
Abstract: Reservoir computing is a powerful framework for real-time information processing, characterized by its high computational ability and quick learning, with applications ranging from machine learning to biological systems. In this paper, we investigate how the computational ability of reservoir recurrent neural networks (RNNs) scales with an increasing number of readout neurons. First, we demonstrate that the memory capacity of a reservoir RNN scales sublinearly with the number of readout neurons. To elucidate this observation, we develop a theoretical framework for analytically deriving memory capacity that incorporates the effect of neuronal correlations, which have been ignored in prior theoretical work for analytical simplicity. Our theory successfully relates the sublinear scaling of memory capacity to the strength of neuronal correlations. Furthermore, we show this principle holds across diverse types of RNNs, even those beyond the direct applicability of our theory. Next, we numerically investigate the scaling behavior of nonlinear computational ability, which, alongside memory capacity, is crucial for overall computational performance. Our numerical simulations reveal that as memory capacity growth becomes sublinear, increasing the number of readout neurons successively enables nonlinear processing at progressively higher polynomial orders. Our theoretical framework suggests that neuronal correlations govern not only memory capacity but also the sequential growth of nonlinear computational capabilities. Our findings establish a foundation for designing scalable and cost-effective reservoir computing, providing novel insights into the interplay among neuronal correlations, linear memory, and nonlinear processing.
Authors: Yuhang Huang, Jiazhao Zhang, Shilong Zou, Xinwang Liu, Ruizhen Hu, Kai Xu
Abstract: Predictive manipulation has recently gained considerable attention in the Embodied AI community due to its potential to improve robot policy performance by leveraging predicted states. However, generating accurate future visual states of robot-object interactions from world models remains a well-known challenge, particularly in achieving high-quality pixel-level representations. To this end, we propose LaDi-WM, a world model that predicts the latent space of future states using diffusion modeling. Specifically, LaDi-WM leverages the well-established latent space aligned with pre-trained Visual Foundation Models (VFMs), which comprises both geometric features (DINO-based) and semantic features (CLIP-based). We find that predicting the evolution of the latent space is easier to learn and more generalizable than directly predicting pixel-level images. Building on LaDi-WM, we design a diffusion policy that iteratively refines output actions by incorporating forecasted states, thereby generating more consistent and accurate results. Extensive experiments on both synthetic and real-world benchmarks demonstrate that LaDi-WM significantly enhances policy performance by 27.9\% on the LIBERO-LONG benchmark and 20\% on the real-world scenario. Furthermore, our world model and policies achieve impressive generalizability in real-world experiments.
Authors: Sina Sharifi, Erfan Yazdandoost Hamedani, Mahyar Fazlyab
Abstract: Bilevel optimization involves a hierarchical structure where one problem is nested within another, leading to complex interdependencies between levels. We propose a single-loop, tuning-free algorithm that guarantees anytime feasibility, i.e., approximate satisfaction of the lower-level optimality condition, while ensuring descent of the upper-level objective. At each iteration, a convex quadratically-constrained quadratic program (QCQP) with a closed-form solution yields the search direction, followed by a backtracking line search inspired by control barrier functions to ensure safe, uniformly positive step sizes. The resulting method is scalable, requires no hyperparameter tuning, and converges under mild local regularity assumptions. We establish an O(1/k) ergodic convergence rate in terms of a first-order stationary metric and demonstrate the algorithm's effectiveness on representative bilevel tasks.
Authors: Fernando Lucatelli Nunes, Gordon Plotkin, Matthijs V\'ak\'ar
Abstract: Combinatory Homomorphic Automatic Differentiation (CHAD) was originally formulated as a semantics-driven source-to-source transformation for reverse-mode AD of total (terminating) functional programs. In this work, we extend CHAD to encompass programs featuring constructs such as partial (potentially non-terminating) operations, data-dependent conditionals (e.g., real-valued tests), and iteration constructs (i.e. while-loops), while maintaining CHAD's core principle of structure-preserving semantics. A central contribution is the introduction of iteration-extensive indexed categories, which provide a principled integration of iteration into dependently typed programming languages. This integration is achieved by requiring that iteration in the base category lifts to parameterized initial algebras in the indexed category, yielding an op-fibred iterative structure that models while-loops and other iteration constructs in the total category, which corresponds to the category of containers of our dependently typed language. Through the idea of iteration-extensive indexed categories, we extend the CHAD transformation to looping programs as the unique structure-preserving functor in a suitable sense. Specifically, it is the unique iterative Freyd category morphism from the iterative Freyd category corresponding to the source language to the category of containers obtained from the target language, such that each primitive operation is mapped to its (transposed) derivative. We establish the correctness of this extended transformation via the universal property of the syntactic categorical model of the source language, showing that the differentiated programs compute correct reverse-mode derivatives of their originals.
Authors: Shuhai Zhang, Zeng You, Yaofo Chen, Zhiquan Wen, Qianyue Wang, Zhijie Qiu, Yuanqing Li, Mingkui Tan
Abstract: Transformer-based large language models (LLMs) excel in natural language processing tasks by capturing long-range dependencies through self-attention mechanisms. However, long-context modeling faces significant computational inefficiencies due to \textit{redundant} attention computations: while attention weights are often \textit{sparse}, all tokens consume \textit{equal} computational resources. In this paper, we reformulate traditional probabilistic sequence modeling as a \textit{supervised learning task}, enabling the separation of relevant and irrelevant tokens and providing a clearer understanding of redundancy. Based on this reformulation, we theoretically analyze attention sparsity, revealing that only a few tokens significantly contribute to predictions. Building on this, we formulate attention optimization as a linear coding problem and propose a \textit{group coding strategy}, theoretically showing its ability to improve robustness against random noise and enhance learning efficiency. Motivated by this, we propose \textit{Dynamic Group Attention} (DGA), which leverages the group coding to explicitly reduce redundancy by aggregating less important tokens during attention computation. Empirical results show that our DGA significantly reduces computational costs while maintaining competitive performance.Code is available at https://github.com/bolixinyu/DynamicGroupAttention.
Authors: Yeonwoo Jang, Shariqah Hossain, Ashwin Sreevatsa, Diogo Cruz
Abstract: In this work, we demonstrate that certain machine unlearning methods may fail under straightforward prompt attacks. We systematically evaluate eight unlearning techniques across three model families using output-based, logit-based, and probe analysis to assess the extent to which supposedly unlearned knowledge can be retrieved. While methods like RMU and TAR exhibit robust unlearning, ELM remains vulnerable to specific prompt attacks (e.g., prepending Hindi filler text to the original prompt recovers 57.3% accuracy). Our logit analysis further indicates that unlearned models are unlikely to hide knowledge through changes in answer formatting, given the strong correlation between output and logit accuracy. These findings challenge prevailing assumptions about unlearning effectiveness and highlight the need for evaluation frameworks that can reliably distinguish between genuine knowledge removal and superficial output suppression. To facilitate further research, we publicly release our evaluation framework to easily evaluate prompting techniques to retrieve unlearned knowledge.
Authors: Yuqi Zhu, Yi Zhong, Jintian Zhang, Ziheng Zhang, Shuofei Qiao, Yujie Luo, Lun Du, Da Zheng, Ningyu Zhang, Huajun Chen
Abstract: Large Language Models (LLMs) hold promise in automating data analysis tasks, yet open-source models face significant limitations in these kinds of reasoning-intensive scenarios. In this work, we investigate strategies to enhance the data analysis capabilities of open-source LLMs. By curating a seed dataset of diverse, realistic scenarios, we evaluate model behavior across three core dimensions: data understanding, code generation, and strategic planning. Our analysis reveals three key findings: (1) Strategic planning quality serves as the primary determinant of model performance; (2) Interaction design and task complexity significantly influence reasoning capabilities; (3) Data quality demonstrates a greater impact than diversity in achieving optimal performance. We leverage these insights to develop a data synthesis methodology, demonstrating significant improvements in open-source LLMs' analytical reasoning capabilities. Code is available at https://github.com/zjunlp/DataMind.
Authors: Sadia Kamal, Lalu Prasad Yadav Prakash, S M Rafiuddin, Mohammed Rakib, Atriya Sen, Sagnik Ray Choudhury
Abstract: Political Compass Test (PCT) or similar questionnaires have been used to quantify LLM's political leanings. Building on a recent line of work that examines the validity of PCT tests, we demonstrate that variation in standard generation parameters does not significantly impact the models' PCT scores. However, external factors such as prompt variations and fine-tuning individually and in combination affect the same. Finally, we demonstrate that when models are fine-tuned on text datasets with higher political content than others, the PCT scores are not differentially affected. This calls for a thorough investigation into the validity of PCT and similar tests, as well as the mechanism by which political leanings are encoded in LLMs.
Authors: V Team, Wenyi Hong, Wenmeng Yu, Xiaotao Gu, Guo Wang, Guobing Gan, Haomiao Tang, Jiale Cheng, Ji Qi, Junhui Ji, Lihang Pan, Shuaiqi Duan, Weihan Wang, Yan Wang, Yean Cheng, Zehai He, Zhe Su, Zhen Yang, Ziyang Pan, Aohan Zeng, Baoxu Wang, Bin Chen, Boyan Shi, Changyu Pang, Chenhui Zhang, Da Yin, Fan Yang, Guoqing Chen, Jiazheng Xu, Jiale Zhu, Jiali Chen, Jing Chen, Jinhao Chen, Jinghao Lin, Jinjiang Wang, Junjie Chen, Leqi Lei, Letian Gong, Leyi Pan, Mingdao Liu, Mingde Xu, Mingzhi Zhang, Qinkai Zheng, Sheng Yang, Shi Zhong, Shiyu Huang, Shuyuan Zhao, Siyan Xue, Shangqin Tu, Shengbiao Meng, Tianshu Zhang, Tianwei Luo, Tianxiang Hao, Tianyu Tong, Wenkai Li, Wei Jia, Xiao Liu, Xiaohan Zhang, Xin Lyu, Xinyue Fan, Xuancheng Huang, Yanling Wang, Yadong Xue, Yanfeng Wang, Yanzi Wang, Yifan An, Yifan Du, Yiming Shi, Yiheng Huang, Yilin Niu, Yuan Wang, Yuanchang Yue, Yuchen Li, Yutao Zhang, Yuting Wang, Yu Wang, Yuxuan Zhang, Zhao Xue, Zhenyu Hou, Zhengxiao Du, Zihan Wang, Peng Zhang, Debing Liu, Bin Xu, Juanzi Li, Minlie Huang, Yuxiao Dong, Jie Tang
Abstract: We present GLM-4.1V-Thinking and GLM-4.5V, a family of vision-language models (VLMs) designed to advance general-purpose multimodal understanding and reasoning. In this report, we share our key findings in the development of the reasoning-centric training framework. We first develop a capable vision foundation model with significant potential through large-scale pre-training, which arguably sets the upper bound for the final performance. We then propose Reinforcement Learning with Curriculum Sampling (RLCS) to unlock the full potential of the model, leading to comprehensive capability enhancement across a diverse range of tasks, including STEM problem solving, video understanding, content recognition, coding, grounding, GUI-based agents, and long document interpretation. In a comprehensive evaluation across 42 public benchmarks, GLM-4.5V achieves state-of-the-art performance on nearly all tasks among open-source models of similar size, and demonstrates competitive or even superior results compared to closed-source models such as Gemini-2.5-Flash on challenging tasks including Coding and GUI Agents. Meanwhile, the smaller GLM-4.1V-9B-Thinking remains highly competitive-achieving superior results to the much larger Qwen2.5-VL-72B on 29 benchmarks. We open-source both GLM-4.1V-9B-Thinking and GLM-4.5V. Code, models and more information are released at https://github.com/zai-org/GLM-V.
Authors: Myeongjang Pyeon, Janghyeon Lee, Minsoo Lee, Juseung Yun, Hwanil Choi, Jonghyun Kim, Jiwon Kim, Yi Hu, Jongseong Jang, Soonyoung Lee
Abstract: In digital pathology, whole-slide images (WSIs) are often difficult to handle due to their gigapixel scale, so most approaches train patch encoders via self-supervised learning (SSL) and then aggregate the patch-level embeddings via multiple instance learning (MIL) or slide encoders for downstream tasks. However, patch-level SSL may overlook complex domain-specific features that are essential for biomarker prediction, such as mutation status and molecular characteristics, as SSL methods rely only on basic augmentations selected for natural image domains on small patch-level area. Moreover, SSL methods remain less data efficient than fully supervised approaches, requiring extensive computational resources and datasets to achieve competitive performance. To address these limitations, we present EXAONE Path 2.0, a pathology foundation model that learns patch-level representations under direct slide-level supervision. Using only 37k WSIs for training, EXAONE Path 2.0 achieves state-of-the-art average performance across 10 biomarker prediction tasks, demonstrating remarkable data efficiency.
Authors: Jihye Lee, Minseo Kang, Dongha Kim
Abstract: The presence of missing values often reflects variations in data collection policies, which may shift across time or locations, even when the underlying feature distribution remains stable. Such shifts in the missingness distribution between training and test inputs pose a significant challenge to achieving robust predictive performance. In this study, we propose a novel deep learning framework designed to address this challenge, particularly in the common yet challenging scenario where the test-time dataset is unseen. We begin by introducing a set of mutual information-based conditions, called MI robustness conditions, which guide the prediction model to extract label-relevant information. This promotes robustness against distributional shifts in missingness at test-time. To enforce these conditions, we design simple yet effective loss terms that collectively define our final objective, called MIRRAMS. Importantly, our method does not rely on any specific missingness assumption such as MCAR, MAR, or MNAR, making it applicable to a broad range of scenarios. Furthermore, it can naturally extend to cases where labels are also missing in training data, by generalizing the framework to a semi-supervised learning setting. Extensive experiments across multiple benchmark tabular datasets demonstrate that MIRRAMS consistently outperforms existing state-of-the-art baselines and maintains stable performance under diverse missingness conditions. Moreover, it achieves superior performance even in fully observed settings, highlighting MIRRAMS as a powerful, off-the-shelf framework for general-purpose tabular learning.
Authors: Ngoc Hung Nguyen, Nguyen Van Thieu, Quang-Trung Luu, Anh Tuan Nguyen, Senura Wanasekara, Nguyen Cong Luong, Fatemeh Kavehmadavani, Van-Dinh Nguyen
Abstract: In this paper, we explore mission assignment and task offloading in an Open Radio Access Network (Open RAN)-based intelligent transportation system (ITS), where autonomous vehicles leverage mobile edge computing for efficient processing. Existing studies often overlook the intricate interdependencies between missions and the costs associated with offloading tasks to edge servers, leading to suboptimal decision-making. To bridge this gap, we introduce Oranits, a novel system model that explicitly accounts for mission dependencies and offloading costs while optimizing performance through vehicle cooperation. To achieve this, we propose a twofold optimization approach. First, we develop a metaheuristic-based evolutionary computing algorithm, namely the Chaotic Gaussian-based Global ARO (CGG-ARO), serving as a baseline for one-slot optimization. Second, we design an enhanced reward-based deep reinforcement learning (DRL) framework, referred to as the Multi-agent Double Deep Q-Network (MA-DDQN), that integrates both multi-agent coordination and multi-action selection mechanisms, significantly reducing mission assignment time and improving adaptability over baseline methods. Extensive simulations reveal that CGG-ARO improves the number of completed missions and overall benefit by approximately 7.1% and 7.7%, respectively. Meanwhile, MA-DDQN achieves even greater improvements of 11.0% in terms of mission completions and 12.5% in terms of the overall benefit. These results highlight the effectiveness of Oranits in enabling faster, more adaptive, and more efficient task processing in dynamic ITS environments.
Authors: Wenxuan Ye, Xueli An, Junfan Wang, Xueqiang Yan, Georg Carle
Abstract: Native AI support is a key objective in the evolution of 6G networks, with Federated Learning (FL) emerging as a promising paradigm. FL allows decentralized clients to collaboratively train an AI model without directly sharing their data, preserving privacy. Clients train local models on private data and share model updates, which a central server aggregates to refine the global model and redistribute it for the next iteration. However, client data heterogeneity slows convergence and reduces model accuracy, and frequent client participation imposes communication and computational burdens. To address these challenges, we propose FedABC, an innovative client selection algorithm designed to take a long-term view in managing data heterogeneity and optimizing client participation. Inspired by attention mechanisms, FedABC prioritizes informative clients by evaluating both model similarity and each model's unique contributions to the global model. Moreover, considering the evolving demands of the global model, we formulate an optimization problem to guide FedABC throughout the training process. Following the "later-is-better" principle, FedABC adaptively adjusts the client selection threshold, encouraging greater participation in later training stages. Extensive simulations on CIFAR-10 demonstrate that FedABC significantly outperforms existing approaches in model accuracy and client participation efficiency, achieving comparable performance with 32% fewer clients than the classical FL algorithm FedAvg, and 3.5% higher accuracy with 2% fewer clients than the state-of-the-art. This work marks a step toward deploying FL in heterogeneous, resource-constrained environments, thereby supporting native AI capabilities in 6G networks.
Authors: Sahar Salimpour, Lei Fu, Farhad Keramat, Leonardo Militano, Giovanni Toffetti, Harry Edelman, Jorge Pe\~na Queralta
Abstract: Foundation models, including large language models (LLMs) and vision-language models (VLMs), have recently enabled novel approaches to robot autonomy and human-robot interfaces. In parallel, vision-language-action models (VLAs) or large behavior models (LBMs) are increasing the dexterity and capabilities of robotic systems. This survey paper focuses on those works advancing towards agentic applications and architectures. This includes initial efforts exploring GPT-style interfaces to tooling, as well as more complex system where AI agents are coordinators, planners, perception actors, or generalist interfaces. Such agentic architectures allow robots to reason over natural language instructions, invoke APIs, plan task sequences, or assist in operations and diagnostics. In addition to peer-reviewed research, due to the fast-evolving nature of the field, we highlight and include community-driven projects, ROS packages, and industrial frameworks that show emerging trends. We propose a taxonomy for classifying model integration approaches and present a comparative analysis of the role that agents play in different solutions in today's literature.
Authors: Meital Shlezinger, Shay Akirav, Lei Zhou, Liang Guo, Avi Kessel, Guoliang Li
Abstract: Database systems are extensively used to store critical data across various domains. However, the frequency of abnormal database access behaviors, such as database intrusion by internal and external attacks, continues to rise. Internal masqueraders often have greater organizational knowledge, making it easier to mimic employee behavior effectively. In contrast, external masqueraders may behave differently due to their lack of familiarity with the organization. Current approaches lack the granularity needed to detect anomalies at the operational level, frequently misclassifying entire sequences of operations as anomalies, even though most operations are likely to represent normal behavior. On the other hand, some anomalous behaviors often resemble normal activities, making them difficult for existing detection methods to identify. This paper introduces a two-tiered anomaly detection approach for Structured Query Language (SQL) using the Bidirectional Encoder Representations from Transformers (BERT) model, specifically DistilBERT, a more efficient, pre-trained version. Our method combines both unsupervised and supervised machine learning techniques to accurately identify anomalous activities while minimizing the need for data labeling. First, the unsupervised method uses ensemble anomaly detectors that flag embedding vectors distant from learned normal patterns of typical user behavior across the database (out-of-scope queries). Second, the supervised method uses fine-tuned transformer-based models to detect internal attacks with high precision (in-scope queries), using role-labeled classification, even on limited labeled SQL data. Our findings make a significant contribution by providing an effective solution for safeguarding critical database systems from sophisticated threats.
Authors: Murtaza Rangwala, KR Venugopal, Rajkumar Buyya
Abstract: Blockchain-enabled federated learning (BCFL) addresses fundamental challenges of trust, privacy, and coordination in collaborative AI systems. This chapter provides comprehensive architectural analysis of BCFL systems through a systematic four-dimensional taxonomy examining coordination structures, consensus mechanisms, storage architectures, and trust models. We analyze design patterns from blockchain-verified centralized coordination to fully decentralized peer-to-peer networks, evaluating trade-offs in scalability, security, and performance. Through detailed examination of consensus mechanisms designed for federated learning contexts, including Proof of Quality and Proof of Federated Learning, we demonstrate how computational work can be repurposed from arbitrary cryptographic puzzles to productive machine learning tasks. The chapter addresses critical storage challenges by examining multi-tier architectures that balance blockchain's transaction constraints with neural networks' large parameter requirements while maintaining cryptographic integrity. A technical case study of the TrustMesh framework illustrates practical implementation considerations in BCFL systems through distributed image classification training, demonstrating effective collaborative learning across IoT devices with highly non-IID data distributions while maintaining complete transparency and fault tolerance. Analysis of real-world deployments across healthcare consortiums, financial services, and IoT security applications validates the practical viability of BCFL systems, achieving performance comparable to centralized approaches while providing enhanced security guarantees and enabling new models of trustless collaborative intelligence.
Authors: Denis Janiak, Jakub Binkowski, Albert Sawczyn, Bogdan Gabrys, Ravid Shwartz-Ziv, Tomasz Kajdanowicz
Abstract: Large language models (LLMs) have revolutionized natural language processing, yet their tendency to hallucinate poses serious challenges for reliable deployment. Despite numerous hallucination detection methods, their evaluations often rely on ROUGE, a metric based on lexical overlap that misaligns with human judgments. Through comprehensive human studies, we demonstrate that while ROUGE exhibits high recall, its extremely low precision leads to misleading performance estimates. In fact, several established detection methods show performance drops of up to 45.9\% when assessed using human-aligned metrics like LLM-as-Judge. Moreover, our analysis reveals that simple heuristics based on response length can rival complex detection techniques, exposing a fundamental flaw in current evaluation practices. We argue that adopting semantically aware and robust evaluation frameworks is essential to accurately gauge the true performance of hallucination detection methods, ultimately ensuring the trustworthiness of LLM outputs.
Authors: Sekh Mainul Islam, Nadav Borenstein, Siddhesh Milind Pawar, Haeun Yu, Arnav Arora, Isabelle Augenstein
Abstract: Understanding biases and stereotypes encoded in the weights of Large Language Models (LLMs) is crucial for developing effective mitigation strategies. Biased behaviour is often subtle and non-trivial to isolate, even when deliberately elicited, making systematic analysis and debiasing particularly challenging. To address this, we introduce BiasGym, a simple, cost-effective, and generalizable framework for reliably injecting, analyzing, and mitigating conceptual associations within LLMs. BiasGym consists of two components: BiasInject, which injects specific biases into the model via token-based fine-tuning while keeping the model frozen, and BiasScope, which leverages these injected signals to identify and steer the components responsible for biased behavior. Our method enables consistent bias elicitation for mechanistic analysis, supports targeted debiasing without degrading performance on downstream tasks, and generalizes to biases unseen during token-based fine-tuning. We demonstrate the effectiveness of BiasGym in reducing real-world stereotypes (e.g., people from Italy being `reckless drivers') and in probing fictional associations (e.g., people from a fictional country having `blue skin'), showing its utility for both safety interventions and interpretability research.