Authors: C. H. E. Jordaan, M. van der Stelt, T. J. J. Maal, V. M. A. Stirler, R. Leijendekkers, T. Kachman, G. A. de Jong
Abstract: The quality of a transtibial prosthetic socket depends on the prosthetist's skills and expertise, as the fitting is performed manually. This study investigates multiple artificial intelligence (AI) approaches to help standardize transtibial prosthetic socket design. Data from 118 patients were collected by prosthetists working in the Dutch healthcare system. This data consists of a three-dimensional (3D) scan of the residual limb and a corresponding 3D model of the prosthetist-designed socket. Multiple data pre-processing steps are performed for alignment, standardization and optionally compression using Morphable Models and Principal Component Analysis. Afterward, three different algorithms - a 3D neural network, Feedforward neural network, and random forest - are developed to either predict 1) the final socket shape or 2) the adaptations performed by a prosthetist to predict the socket shape based on the 3D scan of the residual limb. Each algorithm's performance was evaluated by comparing the prosthetist-designed socket with the AI-generated socket, using two metrics in combination with the error location. First, we measure the surface-to-surface distance to assess the overall surface error between the AI-generated socket and the prosthetist-designed socket. Second, distance maps between the AI-generated and prosthetist sockets are utilized to analyze the error's location. For all algorithms, estimating the required adaptations outperformed direct prediction of the final socket shape. The random forest model applied to adaptation prediction yields the lowest error with a median surface-to-surface distance of 1.24 millimeters, a first quartile of 1.03 millimeters, and a third quartile of 1.54 millimeters.
Authors: Qiuyu Shi, Kangming Li, Yao Fehlis, Daniel Persaud, Robert Black, Jason Hattrick-Simpers
Abstract: Self-driving laboratories (SDLs) have shown promise to accelerate materials discovery by integrating machine learning with automated experimental platforms. However, errors in the capture of input parameters may corrupt the features used to model system performance, compromising current and future campaigns. This study develops an automated workflow to systematically detect noisy features, determine sample-feature pairings that can be corrected, and finally recover the correct feature values. A systematic study is then performed to examine how dataset size, noise intensity, and feature value distribution affect both the detectability and recoverability of noisy features. In general, high-intensity noise and large training datasets are conducive to the detection and correction of noisy features. Low-intensity noise reduces detection and recovery but can be compensated for by larger clean training data sets. Detection and correction results vary between features with continuous and dispersed feature distributions showing greater recoverability compared to features with discrete or narrow distributions. This systematic study not only demonstrates a model agnostic framework for rational data recovery in the presence of noise, limited data, and differing feature distributions but also provides a tangible benchmark of kNN imputation in materials data sets. Ultimately, it aims to enhance data quality and experimental precision in automated materials discovery.
Authors: Jie He, Vincent Theo Willem Kenbeek, Zhantao Yang, Meixun Qu, Ezio Bartocci, Dejan Ni\v{c}kovi\'c, Radu Grosu
Abstract: We introduce TD-Interpreter, a specialized ML tool that assists engineers in understanding complex timing diagrams (TDs), originating from a third party, during their design and verification process. TD-Interpreter is a visual question-answer environment which allows engineers to input a set of TDs and ask design and verification queries regarding these TDs. We implemented TD-Interpreter with multimodal learning by fine-tuning LLaVA, a lightweight 7B Multimodal Large Language Model (MLLM). To address limited training data availability, we developed a synthetic data generation workflow that aligns visual information with its textual interpretation. Our experimental evaluation demonstrates the usefulness of TD-Interpreter which outperformed untuned GPT-4o by a large margin on the evaluated benchmarks.
Authors: Tao Xu, Dung-Yang Lee, Momiao Xiong
Abstract: Multi-step reasoning is a fundamental challenge in artificial intelligence, with applications ranging from mathematical problem-solving to decision-making in dynamic environments. Reinforcement Learning (RL) has shown promise in enabling agents to perform multi-step reasoning by optimizing long-term rewards. However, conventional RL methods struggle with complex reasoning tasks due to issues such as credit assignment, high-dimensional state representations, and stability concerns. Recent advancements in Transformer architectures and hyperbolic geometry have provided novel solutions to these challenges. This paper introduces a new framework that integrates hyperbolic Transformers into RL for multi-step reasoning. The proposed approach leverages hyperbolic embeddings to model hierarchical structures effectively. We present theoretical insights, algorithmic details, and experimental results that include Frontier Math and nonlinear optimal control problems. Compared to RL with vanilla transformer, the hyperbolic RL largely improves accuracy by (32%~44%) on FrontierMath benchmark, (43%~45%) on nonlinear optimal control benchmark, while achieving impressive reduction in computational time by (16%~32%) on FrontierMath benchmark, (16%~17%) on nonlinear optimal control benchmark. Our work demonstrates the potential of hyperbolic Transformers in reinforcement learning, particularly for multi-step reasoning tasks that involve hierarchical structures.
Authors: Yunyi Zhao, Wei Zhang, Cheng Xiang, Hongyang Du, Dusit Niyato, Shuhua Gao
Abstract: This paper introduces DiffCarl, a diffusion-modeled carbon- and risk-aware reinforcement learning algorithm for intelligent operation of multi-microgrid systems. With the growing integration of renewables and increasing system complexity, microgrid communities face significant challenges in real-time energy scheduling and optimization under uncertainty. DiffCarl integrates a diffusion model into a deep reinforcement learning (DRL) framework to enable adaptive energy scheduling under uncertainty and explicitly account for carbon emissions and operational risk. By learning action distributions through a denoising generation process, DiffCarl enhances DRL policy expressiveness and enables carbon- and risk-aware scheduling in dynamic and uncertain microgrid environments. Extensive experimental studies demonstrate that it outperforms classic algorithms and state-of-the-art DRL solutions, with 2.3-30.1% lower operational cost. It also achieves 28.7% lower carbon emissions than those of its carbon-unaware variant and reduces performance variability. These results highlight DiffCarl as a practical and forward-looking solution. Its flexible design allows efficient adaptation to different system configurations and objectives to support real-world deployment in evolving energy systems.
Authors: Pietro Giuseppe Fr\'e, Federico Milanesio, Guido Sanguinetti, Matteo Santoro
Abstract: Recent work has identified non-compact symmetric spaces U/H as a promising class of homogeneous manifolds to develop a geometrically consistent theory of neural networks. An initial implementation of these concepts has been presented in a twin paper under the moniker of Cartan Neural Networks, showing both the feasibility and the performance of these geometric concepts in a machine learning context. The current paper expands on the mathematical structures underpinning Cartan Neural Networks, detailing the geometric properties of the layers and how the maps between layers interact with such structures to make Cartan Neural Networks covariant and geometrically interpretable. Together, these twin papers constitute a first step towards a fully geometrically interpretable theory of neural networks exploiting group-theoretic structures
Authors: Pengjiu Xia, Yidian Huang, Wenchao Wei, Yuwen Tan
Abstract: Probabilistic encoding introduces Gaussian noise into neural networks, enabling a smooth transition from deterministic to uncertain states and enhancing generalization ability. However, the randomness of Gaussian noise distorts point-based distance measurements in classification tasks. To mitigate this issue, we propose a confidence optimization probabilistic encoding (CPE) method that improves distance reliability and enhances representation learning. Specifically, we refine probabilistic encoding with two key strategies: First, we introduce a confidence-aware mechanism to adjust distance calculations, ensuring consistency and reliability in probabilistic encoding classification tasks. Second, we replace the conventional KL divergence-based variance regularization, which relies on unreliable prior assumptions, with a simpler L2 regularization term to directly constrain variance. The method we proposed is model-agnostic, and extensive experiments on natural language classification tasks demonstrate that our method significantly improves performance and generalization on both the BERT and the RoBERTa model.
Authors: Yi Guo, Wei Wang, Zhihang Yuan, Rong Cao, Kuan Chen, Zhengyang Chen, Yuanyuan Huo, Yang Zhang, Yuping Wang, Shouda Liu, Yuxuan Wang
Abstract: Generative models like Flow Matching have achieved state-of-the-art performance but are often hindered by a computationally expensive iterative sampling process. To address this, recent work has focused on few-step or one-step generation by learning the average velocity field, which directly maps noise to data. MeanFlow, a leading method in this area, learns this field by enforcing a differential identity that connects the average and instantaneous velocities. In this work, we argue that this differential formulation is a limiting special case of a more fundamental principle. We return to the first principles of average velocity and leverage the additivity property of definite integrals. This leads us to derive a novel, purely algebraic identity we term Interval Splitting Consistency. This identity establishes a self-referential relationship for the average velocity field across different time intervals without resorting to any differential operators. Based on this principle, we introduce SplitMeanFlow, a new training framework that enforces this algebraic consistency directly as a learning objective. We formally prove that the differential identity at the core of MeanFlow is recovered by taking the limit of our algebraic consistency as the interval split becomes infinitesimal. This establishes SplitMeanFlow as a direct and more general foundation for learning average velocity fields. From a practical standpoint, our algebraic approach is significantly more efficient, as it eliminates the need for JVP computations, resulting in simpler implementation, more stable training, and broader hardware compatibility. One-step and two-step SplitMeanFlow models have been successfully deployed in large-scale speech synthesis products (such as Doubao), achieving speedups of 20x.
Authors: Steven K. Esser, Jeffrey L. McKinstry, Deepika Bablani, Rathinakumar Appuswamy, Dharmendra S. Modha
Abstract: Large language models can be quantized to reduce inference time latency, model size, and energy consumption, thereby delivering a better user experience at lower cost. A challenge exists to deliver quantized models with minimal loss of accuracy in reasonable time, and in particular to do so without requiring mechanisms incompatible with specialized inference accelerators. Here, we demonstrate a simple, end-to-end quantization-aware training approach that, with an increase in total model training budget of less than 0.1%, outperforms the leading published quantization methods by large margins on several modern benchmarks, with both base and instruct model variants. The approach easily generalizes across different model architectures, can be applied to activations, cache, and weights, and requires the introduction of no additional operations to the model other than the quantization itself.
Authors: Sonny T. Jones, Grange M. Simpson, Patrick M. Pilarski, Ashley N. Dalrymple
Abstract: Rehabilitation technology is a natural setting to study the shared learning and decision-making of human and machine agents. In this work, we explore the use of Hierarchical Reinforcement Learning (HRL) to develop adaptive control strategies for lower-limb exoskeletons, aiming to enhance mobility and autonomy for individuals with motor impairments. Inspired by prominent models of biological sensorimotor processing, our investigated HRL approach breaks down the complex task of exoskeleton control adaptation into a higher-level framework for terrain strategy adaptation and a lower-level framework for providing predictive information; this latter element is implemented via the continual learning of general value functions (GVFs). GVFs generated temporal abstractions of future signal values from multiple wearable lower-limb sensors, including electromyography, pressure insoles, and goniometers. We investigated two methods for incorporating actual and predicted sensor signals into a policy network with the intent to improve the decision-making capacity of the control system of a lower-limb exoskeleton during ambulation across varied terrains. As a key result, we found that the addition of predictions made from GVFs increased overall network accuracy. Terrain-specific performance increases were seen while walking on even ground, uneven ground, up and down ramps, and turns, terrains that are often misclassified without predictive information. This suggests that predictive information can aid decision-making during uncertainty, e.g., on terrains that have a high chance of being misclassified. This work, therefore, contributes new insights into the nuances of HRL and the future development of exoskeletons to facilitate safe transitioning and traversing across different walking environments.
Authors: Matthias Fey, Jinu Sunil, Akihiro Nitta, Rishi Puri, Manan Shah, Bla\v{z} Stojanovi\v{c}, Ramona Bendias, Alexandria Barghi, Vid Kocijan, Zecheng Zhang, Xinwei He, Jan Eric Lenssen, Jure Leskovec
Abstract: PyG (PyTorch Geometric) has evolved significantly since its initial release, establishing itself as a leading framework for Graph Neural Networks. In this paper, we present Pyg 2.0 (and its subsequent minor versions), a comprehensive update that introduces substantial improvements in scalability and real-world application capabilities. We detail the framework's enhanced architecture, including support for heterogeneous and temporal graphs, scalable feature/graph stores, and various optimizations, enabling researchers and practitioners to tackle large-scale graph learning problems efficiently. Over the recent years, PyG has been supporting graph learning in a large variety of application areas, which we will summarize, while providing a deep dive into the important areas of relational deep learning and large language modeling.
Authors: Yan Li, Guangyi Chen, Yunlong Deng, Zijian Li, Zeyu Tang, Anpeng Wu, Kun Zhang
Abstract: Most existing methods for adapting models to out-of-distribution (OOD) domains rely on invariant representation learning to eliminate the influence of biased features. However, should bias always be eliminated -- and if not, when should it be retained, and how can it be leveraged? To address these questions, we first present a theoretical analysis that explores the conditions under which biased features can be identified and effectively utilized. Building on this theoretical foundation, we introduce a novel framework that strategically leverages bias to complement invariant representations during inference. The framework comprises two key components that leverage bias in both direct and indirect ways: (1) using invariance as guidance to extract predictive ingredients from bias, and (2) exploiting identified bias to estimate the environmental condition and then use it to explore appropriate bias-aware predictors to alleviate environment gaps. We validate our approach through experiments on both synthetic datasets and standard domain generalization benchmarks. Results consistently demonstrate that our method outperforms existing approaches, underscoring its robustness and adaptability.
Authors: Tobias Weber, B\'alint Mucs\'anyi, Lenard Rommel, Thomas Christie, Lars Kas\"uschke, Marvin Pf\"ortner, Philipp Hennig
Abstract: The Laplace approximation provides a scalable and efficient means of quantifying weight-space uncertainty in deep neural networks, enabling the application of Bayesian tools such as predictive uncertainty and model selection via Occam's razor. In this work, we introduce laplax, a new open-source Python package for performing Laplace approximations with jax. Designed with a modular and purely functional architecture and minimal external dependencies, laplax offers a flexible and researcher-friendly framework for rapid prototyping and experimentation. Its goal is to facilitate research on Bayesian neural networks, uncertainty quantification for deep learning, and the development of improved Laplace approximation techniques.
Authors: Omid Orang, Patricia O. Lucas, Gabriel I. F. Paiva, Petronio C. L. Silva, Felipe Augusto Rocha da Silva, Adriano Alonso Veloso, Frederico Gadelha Guimaraes
Abstract: In recent years, the application of Large Language Models (LLMs) to time series forecasting (TSF) has garnered significant attention among researchers. This study presents a new frame of LLMs named CGF-LLM using GPT-2 combined with fuzzy time series (FTS) and causal graph to predict multivariate time series, marking the first such architecture in the literature. The key objective is to convert numerical time series into interpretable forms through the parallel application of fuzzification and causal analysis, enabling both semantic understanding and structural insight as input for the pretrained GPT-2 model. The resulting textual representation offers a more interpretable view of the complex dynamics underlying the original time series. The reported results confirm the effectiveness of our proposed LLM-based time series forecasting model, as demonstrated across four different multivariate time series datasets. This initiative paves promising future directions in the domain of TSF using LLMs based on FTS.
Authors: Ray Zirui Zhang, Christopher E. Miles, Xiaohui Xie, John S. Lowengrub
Abstract: Uncertainty quantification and inverse problems governed by partial differential equations (PDEs) are central to a wide range of scientific and engineering applications. In this second part of a two part series, we extend Bilevel Local Operator Learning (BiLO) for PDE-constrained optimization problems developed in Part 1 to the Bayesian inference framework. At the lower level, we train a network to approximate the local solution operator by minimizing the local operator loss with respect to the weights of the neural network. At the upper level, we sample the PDE parameters from the posterior distribution. We achieve efficient sampling through gradient-based Markov Chain Monte Carlo (MCMC) methods and low-rank adaptation (LoRA). Compared with existing methods based on Bayesian neural networks, our approach bypasses the challenge of sampling in the high-dimensional space of neural network weights and does not require specifying a prior distribution on the neural network solution. Instead, uncertainty propagates naturally from the data through the PDE constraints. By enforcing strong PDE constraints, the proposed method improves the accuracy of both parameter inference and uncertainty quantification. We analyze the dynamic error of the gradient in the MCMC sampler and the static error in the posterior distribution due to inexact minimization of the lower level problem and demonstrate a direct link between the tolerance for solving the lower level problem and the accuracy of the resulting uncertainty quantification. Through numerical experiments across a variety of PDE models, we demonstrate that our method delivers accurate inference and quantification of uncertainties while maintaining high computational efficiency.
Authors: Anton Matsson, Yaochen Rao, Heather J. Litman, Fredrik D. Johansson
Abstract: Offline reinforcement learning (RL) holds great promise for deriving optimal policies from observational data, but challenges related to interpretability and evaluation limit its practical use in safety-critical domains. Interpretability is hindered by the black-box nature of unconstrained RL policies, while evaluation -- typically performed off-policy -- is sensitive to large deviations from the data-collecting behavior policy, especially when using methods based on importance sampling. To address these challenges, we propose a simple yet practical alternative: deriving treatment policies from the most frequently chosen actions in each patient state, as estimated by an interpretable model of the behavior policy. By using a tree-based model, which is specifically designed to exploit patterns in the data, we obtain a natural grouping of states with respect to treatment. The tree structure ensures interpretability by design, while varying the number of actions considered controls the degree of overlap with the behavior policy, enabling reliable off-policy evaluation. This pragmatic approach to policy development standardizes frequent treatment patterns, capturing the collective clinical judgment embedded in the data. Using real-world examples in rheumatoid arthritis and sepsis care, we demonstrate that policies derived under this framework can outperform current practice, offering interpretable alternatives to those obtained via offline RL.
Authors: Jessup Byun, Xiaofeng Lin, Joshua Ward, Guang Cheng
Abstract: Synthetic tabular data is essential for machine learning workflows, especially for expanding small or imbalanced datasets and enabling privacy-preserving data sharing. However, state-of-the-art generative models (GANs, VAEs, diffusion models) rely on large datasets with thousands of examples. In low-data settings, often the primary motivation for synthetic data, these models can overfit, leak sensitive records, and require frequent retraining. Recent work uses large pre-trained transformers to generate rows via in-context learning (ICL), which needs only a few seed examples and no parameter updates, avoiding retraining. But ICL repeats seed rows verbatim, introducing a new privacy risk that has only been studied in text. The severity of this risk in tabular synthesis-where a single row may identify a person-remains unclear. We address this gap with the first benchmark of three foundation models (GPT-4o-mini, LLaMA 3.3 70B, TabPFN v2) against four baselines on 35 real-world tables from health, finance, and policy. We evaluate statistical fidelity, downstream utility, and membership inference leakage. Results show foundation models consistently have the highest privacy risk. LLaMA 3.3 70B reaches up to 54 percentage points higher true-positive rate at 1% FPR than the safest baseline. GPT-4o-mini and TabPFN are also highly vulnerable. We plot the privacy-utility frontier and show that CTGAN and GPT-4o-mini offer better tradeoffs. A factorial study finds that three zero-cost prompt tweaks-small batch size, low temperature, and using summary statistics-can reduce worst-case AUC by 14 points and rare-class leakage by up to 39 points while maintaining over 90% fidelity. Our benchmark offers a practical guide for safer low-data synthesis with foundation models.
Authors: Adithya Mohan, Dominik R\"o{\ss}le, Daniel Cremers, Torsten Sch\"on
Abstract: Recent advancements in Deep Reinforcement Learning (DRL) have demonstrated its applicability across various domains, including robotics, healthcare, energy optimization, and autonomous driving. However, a critical question remains: How robust are DRL models when exposed to adversarial attacks? While existing defense mechanisms such as adversarial training and distillation enhance the resilience of DRL models, there remains a significant research gap regarding the integration of multiple defenses in autonomous driving scenarios specifically. This paper addresses this gap by proposing a novel ensemble-based defense architecture to mitigate adversarial attacks in autonomous driving. Our evaluation demonstrates that the proposed architecture significantly enhances the robustness of DRL models. Compared to the baseline under FGSM attacks, our ensemble method improves the mean reward from 5.87 to 18.38 (over 213% increase) and reduces the mean collision rate from 0.50 to 0.09 (an 82% decrease) in the highway scenario and merge scenario, outperforming all standalone defense strategies.
Authors: Juntao Lin, Xianghao Zhan
Abstract: Due to environmental changes and sensor aging, sensor drift challenges the performance of electronic nose systems in gas classification during real-world deployment. Previous studies using the UCI Gas Sensor Array Drift Dataset reported promising drift compensation results but lacked robust statistical experimental validation and may overcompensate for sensor drift, losing class-related variance.To address these limitations and improve sensor drift compensation with statistical rigor, we first designed two domain adaptation tasks based on the same electronic nose dataset: using the first batch to predict the remaining batches, simulating a controlled laboratory setting; and predicting the next batch using all prior batches, simulating continuous training data updates for online training. We then systematically tested three methods: our proposed novel Knowledge Distillation (KD) method, the benchmark method Domain Regularized Component Analysis (DRCA), and a hybrid method KD-DRCA, across 30 random test set partitions on the UCI dataset. We showed that KD consistently outperformed both DRCA and KD-DRCA, achieving up to an 18% improvement in accuracy and 15% in F1-score, demonstrating KD's superior effectiveness in drift compensation. This is the first application of KD for electronic nose drift mitigation, significantly outperforming the previous state-of-the-art DRCA method and enhancing the reliability of sensor drift compensation in real-world environments.
Authors: Olivia Dry, Timothy L. Molloy, Wanxin Jin, Iman Shames
Abstract: We propose Zeroth-Order Random Matrix Search for Learning from Demonstrations (ZORMS-LfD). ZORMS-LfD enables the costs, constraints, and dynamics of constrained optimal control problems, in both continuous and discrete time, to be learned from expert demonstrations without requiring smoothness of the learning-loss landscape. In contrast, existing state-of-the-art first-order methods require the existence and computation of gradients of the costs, constraints, dynamics, and learning loss with respect to states, controls and/or parameters. Most existing methods are also tailored to discrete time, with constrained problems in continuous time receiving only cursory attention. We demonstrate that ZORMS-LfD matches or surpasses the performance of state-of-the-art methods in terms of both learning loss and compute time across a variety of benchmark problems. On unconstrained continuous-time benchmark problems, ZORMS-LfD achieves similar loss performance to state-of-the-art first-order methods with an over $80$\% reduction in compute time. On constrained continuous-time benchmark problems where there is no specialized state-of-the-art method, ZORMS-LfD is shown to outperform the commonly used gradient-free Nelder-Mead optimization method.
Authors: Andrii Balashov
Abstract: Reinforcement learning (RL) is a key post-pretraining step for aligning large language models (LLMs) with complex tasks and human preferences. While it is often assumed that RL fine-tuning requires updating most of a model's parameters, we challenge this assumption with a surprising finding: RL fine-tuning consistently modifies only a small subnetwork (typically 5-30% of weights), leaving most parameters unchanged. We call this phenomenon RL-induced parameter update sparsity. It arises naturally, without any sparsity constraints or parameter-efficient tuning, and appears across multiple RL algorithms (e.g., PPO, DPO, SimPO, PRIME) and model families (e.g., OpenAI, Meta, and open-source LLMs). Moreover, the subnetworks updated by RL show substantial overlap across different seeds, datasets, and algorithms-far exceeding chance-suggesting a partially transferable structure in the pretrained model. We show that fine-tuning only this sparse subnetwork recovers full model performance and yields parameters nearly identical to the fully fine-tuned model. Our analysis suggests this sparsity emerges because RL operates near the model's original distribution, requiring only targeted changes. KL penalties, gradient clipping, and on-policy dynamics have limited effect on the sparsity pattern. These findings shed new light on how RL adapts models: not by shifting all weights, but by focusing training on a small, consistently updated subnetwork. This insight enables more efficient RL methods and reframes sparsity through the lens of the lottery ticket hypothesis.
Authors: Jacqueline Maasch, Willie Neiswanger, Stefano Ermon, Volodymyr Kuleshov
Abstract: Probabilistic graphical modeling is a branch of machine learning that uses probability distributions to describe the world, make predictions, and support decision-making under uncertainty. Underlying this modeling framework is an elegant body of theory that bridges two mathematical traditions: probability and graph theory. This framework provides compact yet expressive representations of joint probability distributions, yielding powerful generative models for probabilistic reasoning. This tutorial provides a concise introduction to the formalisms, methods, and applications of this modeling framework. After a review of basic probability and graph theory, we explore three dominant themes: (1) the representation of multivariate distributions in the intuitive visual language of graphs, (2) algorithms for learning model parameters and graphical structures from data, and (3) algorithms for inference, both exact and approximate.
Authors: Jacob M. Delgado-L\'opez, Ricardo A. Morell-Rodriguez, Sebasti\'an O. Espinosa-Del Rosario, Wilfredo E. Lugo-Beauchamp
Abstract: The rapid diagnosis of infectious diseases, such as monkeypox, is crucial for effective containment and treatment, particularly in resource-constrained environments. This study presents an AI-driven diagnostic tool developed for deployment on the NVIDIA Jetson Orin Nano, leveraging the pre-trained MobileNetV2 architecture for binary classification. The model was trained on the open-source Monkeypox Skin Lesion Dataset, achieving a 93.07% F1-Score, which reflects a well-balanced performance in precision and recall. To optimize the model, the TensorRT framework was used to accelerate inference for FP32 and to perform post-training quantization for FP16 and INT8 formats. TensorRT's mixed-precision capabilities enabled these optimizations, which reduced the model size, increased inference speed, and lowered power consumption by approximately a factor of two, all while maintaining the original accuracy. Power consumption analysis confirmed that the optimized models used significantly less energy during inference, reinforcing their suitability for deployment in resource-constrained environments. The system was deployed with a Wi-Fi Access Point (AP) hotspot and a web-based interface, enabling users to upload and analyze images directly through connected devices such as mobile phones. This setup ensures simple access and seamless connectivity, making the tool practical for real-world applications. These advancements position the diagnostic tool as an efficient, scalable, and energy-conscious solution to address diagnosis challenges in underserved regions, paving the way for broader adoption in low-resource healthcare settings.
Authors: Jacob M. Delgado-L\'opez, Andrea P. Seda-Hernandez, Juan D. Guadalupe-Rosado, Luis E. Fernandez Ramirez, Miguel Giboyeaux-Camilo, Wilfredo E. Lugo-Beauchamp
Abstract: Skin cancer is one of the most prevalent and preventable types of cancer, yet its early detection remains a challenge, particularly in resource-limited settings where access to specialized healthcare is scarce. This study proposes an AI-driven diagnostic tool optimized for embedded systems to address this gap. Using transfer learning with the MobileNetV2 architecture, the model was adapted for binary classification of skin lesions into "Skin Cancer" and "Other." The TensorRT framework was employed to compress and optimize the model for deployment on the NVIDIA Jetson Orin Nano, balancing performance with energy efficiency. Comprehensive evaluations were conducted across multiple benchmarks, including model size, inference speed, throughput, and power consumption. The optimized models maintained their performance, achieving an F1-Score of 87.18% with a precision of 93.18% and recall of 81.91%. Post-compression results showed reductions in model size of up to 0.41, along with improvements in inference speed and throughput, and a decrease in energy consumption of up to 0.93 in INT8 precision. These findings validate the feasibility of deploying high-performing, energy-efficient diagnostic tools on resource-constrained edge devices. Beyond skin cancer detection, the methodologies applied in this research have broader applications in other medical diagnostics and domains requiring accessible, efficient AI solutions. This study underscores the potential of optimized AI systems to revolutionize healthcare diagnostics, thereby bridging the divide between advanced technology and underserved regions.
Authors: Yufei He, Ruoyu Li, Alex Chen, Yue Liu, Yulin Chen, Yuan Sui, Cheng Chen, Yi Zhu, Luca Luo, Frank Yang, Bryan Hooi
Abstract: Large language model (LLM) agents often struggle in environments where rules and required domain knowledge frequently change, such as regulatory compliance and user risk screening. Current approaches, like offline fine-tuning and standard prompting, are insufficient because they cannot effectively adapt to new knowledge during actual operation. To address this limitation, we propose the Adaptive Reflective Interactive Agent (ARIA), an LLM agent framework designed specifically to continuously learn updated domain knowledge at test time. ARIA assesses its own uncertainty through structured self-dialogue, proactively identifying knowledge gaps and requesting targeted explanations or corrections from human experts. It then systematically updates an internal, timestamped knowledge repository with provided human guidance, detecting and resolving conflicting or outdated knowledge through comparisons and clarification queries. We evaluate ARIA on the realistic customer due diligence name screening task on TikTok Pay, alongside publicly available dynamic knowledge tasks. Results demonstrate significant improvements in adaptability and accuracy compared to baselines using standard offline fine-tuning and existing self-improving agents. ARIA is deployed within TikTok Pay serving over 150 million monthly active users, confirming its practicality and effectiveness for operational use in rapidly evolving environments.
Authors: Ting Jiang, Yixiao Wang, Hancheng Ye, Zishan Shao, Jingwei Sun, Jingyang Zhang, Zekai Chen, Jianyi Zhang, Yiran Chen, Hai Li
Abstract: Diffusion models have achieved remarkable success in generative tasks but suffer from high computational costs due to their iterative sampling process and quadratic attention costs. Existing training-free acceleration strategies that reduce per-step computation cost, while effectively reducing sampling time, demonstrate low faithfulness compared to the original baseline. We hypothesize that this fidelity gap arises because (a) different prompts correspond to varying denoising trajectory, and (b) such methods do not consider the underlying ODE formulation and its numerical solution. In this paper, we propose Stability-guided Adaptive Diffusion Acceleration (SADA), a novel paradigm that unifies step-wise and token-wise sparsity decisions via a single stability criterion to accelerate sampling of ODE-based generative models (Diffusion and Flow-matching). For (a), SADA adaptively allocates sparsity based on the sampling trajectory. For (b), SADA introduces principled approximation schemes that leverage the precise gradient information from the numerical ODE solver. Comprehensive evaluations on SD-2, SDXL, and Flux using both EDM and DPM++ solvers reveal consistent $\ge 1.8\times$ speedups with minimal fidelity degradation (LPIPS $\leq 0.10$ and FID $\leq 4.5$) compared to unmodified baselines, significantly outperforming prior methods. Moreover, SADA adapts seamlessly to other pipelines and modalities: It accelerates ControlNet without any modifications and speeds up MusicLDM by $1.8\times$ with $\sim 0.01$ spectrogram LPIPS.
Authors: Anirudh Satheesh, Anant Khandelwal, Mucong Ding, Radu Balan
Abstract: Neural operators offer a powerful paradigm for solving partial differential equations (PDEs) that cannot be solved analytically by learning mappings between function spaces. However, there are two main bottlenecks in training neural operators: they require a significant amount of training data to learn these mappings, and this data needs to be labeled, which can only be accessed via expensive simulations with numerical solvers. To alleviate both of these issues simultaneously, we propose PICore, an unsupervised coreset selection framework that identifies the most informative training samples without requiring access to ground-truth PDE solutions. PICore leverages a physics-informed loss to select unlabeled inputs by their potential contribution to operator learning. After selecting a compact subset of inputs, only those samples are simulated using numerical solvers to generate labels, reducing annotation costs. We then train the neural operator on the reduced labeled dataset, significantly decreasing training time as well. Across four diverse PDE benchmarks and multiple coreset selection strategies, PICore achieves up to 78% average increase in training efficiency relative to supervised coreset selection methods with minimal changes in accuracy. We provide code at https://github.com/Asatheesh6561/PICore.
Authors: Vinura Galwaduge, Jagath Samarabandu
Abstract: Modern network intrusion detection systems (NIDS) frequently utilize the predictive power of complex deep learning models. However, the "black-box" nature of such deep learning methods adds a layer of opaqueness that hinders the proper understanding of detection decisions, trust in the decisions and prevent timely countermeasures against such attacks. Explainable AI (XAI) methods provide a solution to this problem by providing insights into the causes of the predictions. The majority of the existing XAI methods provide explanations which are not convenient to convert into actionable countermeasures. In this work, we propose a novel diffusion-based counterfactual explanation framework that can provide actionable explanations for network intrusion attacks. We evaluated our proposed algorithm against several other publicly available counterfactual explanation algorithms on 3 modern network intrusion datasets. To the best of our knowledge, this work also presents the first comparative analysis of existing counterfactual explanation algorithms within the context of network intrusion detection systems. Our proposed method provide minimal, diverse counterfactual explanations out of the tested counterfactual explanation algorithms in a more efficient manner by reducing the time to generate explanations. We also demonstrate how counterfactual explanations can provide actionable explanations by summarizing them to create a set of global rules. These rules are actionable not only at instance level but also at the global level for intrusion attacks. These global counterfactual rules show the ability to effectively filter out incoming attack queries which is crucial for efficient intrusion detection and defense mechanisms.
Authors: Shaohan Li, Hao Yang, Min Chen, Xiaolin Qin
Abstract: The increasing frequency of extreme weather events due to global climate change urges accurate weather prediction. Recently, great advances have been made by the \textbf{end-to-end methods}, thanks to deep learning techniques, but they face limitations of \textit{representation inconsistency} in multivariable integration and struggle to effectively capture the dependency between variables, which is required in complex weather systems. Treating different variables as distinct modalities and applying a \textbf{two-stage training approach} from multimodal models can partially alleviate this issue, but due to the inconformity in training tasks between the two stages, the results are often suboptimal. To address these challenges, we propose an implicit two-stage training method, configuring separate encoders and decoders for each variable. In detailed, in the first stage, the Translator is frozen while the Encoders and Decoders learn a shared latent space, in the second stage, the Encoders and Decoders are frozen, and the Translator captures inter-variable interactions for prediction. Besides, by introducing a self-attention mechanism for multivariable fusion in the latent space, the performance achieves further improvements. Empirically, extensive experiments show the state-of-the-art performance of our method. Specifically, it reduces the MSE for near-surface air temperature and relative humidity predictions by 28.82\% and 23.39\%, respectively. The source code is available at https://github.com/ShremG/Met2Net.
Authors: Zixuan Wang, Jinghao Shi, Hanzhong Liang, Xiang Shen, Vera Wen, Zhiqian Chen, Yifan Wu, Zhixin Zhang, Hongyu Xiong
Abstract: Effective content moderation is essential for video platforms to safeguard user experience and uphold community standards. While traditional video classification models effectively handle well-defined moderation tasks, they struggle with complicated scenarios such as implicit harmful content and contextual ambiguity. Multimodal large language models (MLLMs) offer a promising solution to these limitations with their superior cross-modal reasoning and contextual understanding. However, two key challenges hinder their industrial adoption. First, the high computational cost of MLLMs makes full-scale deployment impractical. Second, adapting generative models for discriminative classification remains an open research problem. In this paper, we first introduce an efficient method to transform a generative MLLM into a multimodal classifier using minimal discriminative training data. To enable industry-scale deployment, we then propose a router-ranking cascade system that integrates MLLMs with a lightweight router model. Offline experiments demonstrate that our MLLM-based approach improves F1 score by 66.50% over traditional classifiers while requiring only 2% of the fine-tuning data. Online evaluations show that our system increases automatic content moderation volume by 41%, while the cascading deployment reduces computational cost to only 1.5% of direct full-scale deployment.
Authors: Youneng Bao, Yiping Liu, Zhuo Chen, Yongsheng Liang, Mu Li, Kede Ma
Abstract: Driven by the ``scale-is-everything'' paradigm, modern machine learning increasingly demands ever-larger datasets and models, yielding prohibitive computational and storage requirements. Dataset distillation mitigates this by compressing an original dataset into a small set of synthetic samples, while preserving its full utility. Yet, existing methods either maximize performance under fixed storage budgets or pursue suitable synthetic data representations for redundancy removal, without jointly optimizing both objectives. In this work, we propose a joint rate-utility optimization method for dataset distillation. We parameterize synthetic samples as optimizable latent codes decoded by extremely lightweight networks. We estimate the Shannon entropy of quantized latents as the rate measure and plug any existing distillation loss as the utility measure, trading them off via a Lagrange multiplier. To enable fair, cross-method comparisons, we introduce bits per class (bpc), a precise storage metric that accounts for sample, label, and decoder parameter costs. On CIFAR-10, CIFAR-100, and ImageNet-128, our method achieves up to $170\times$ greater compression than standard distillation at comparable accuracy. Across diverse bpc budgets, distillation losses, and backbone architectures, our approach consistently establishes better rate-utility trade-offs.
Authors: Wei Fan, JinYi Yoon, Xiaochang Li, Huajie Shao, Bo Ji
Abstract: Split Learning (SL) is an emerging privacy-preserving machine learning technique that enables resource constrained edge devices to participate in model training by partitioning a model into client-side and server-side sub-models. While SL reduces computational overhead on edge devices, it encounters significant challenges in heterogeneous environments where devices vary in computing resources, communication capabilities, environmental conditions, and privacy requirements. Although recent studies have explored heterogeneous SL frameworks that optimize split points for devices with varying resource constraints, they often neglect personalized privacy requirements and local model customization under varying environmental conditions. To address these limitations, we propose P3SL, a Personalized Privacy-Preserving Split Learning framework designed for heterogeneous, resource-constrained edge device systems. The key contributions of this work are twofold. First, we design a personalized sequential split learning pipeline that allows each client to achieve customized privacy protection and maintain personalized local models tailored to their computational resources, environmental conditions, and privacy needs. Second, we adopt a bi-level optimization technique that empowers clients to determine their own optimal personalized split points without sharing private sensitive information (i.e., computational resources, environmental conditions, privacy requirements) with the server. This approach balances energy consumption and privacy leakage risks while maintaining high model accuracy. We implement and evaluate P3SL on a testbed consisting of 7 devices including 4 Jetson Nano P3450 devices, 2 Raspberry Pis, and 1 laptop, using diverse model architectures and datasets under varying environmental conditions.
Authors: Mattia Sabella, Monica Vitali
Abstract: The widespread adoption of Artificial Intelligence (AI) and Machine Learning (ML) comes with a significant environmental impact, particularly in terms of energy consumption and carbon emissions. This pressing issue highlights the need for innovative solutions to mitigate AI's ecological footprint. One of the key factors influencing the energy consumption of ML model training is the size of the training dataset. ML models are often trained on vast amounts of data continuously generated by sensors and devices distributed across multiple locations. To reduce data transmission costs and enhance privacy, Federated Learning (FL) enables model training without the need to move or share raw data. While FL offers these advantages, it also introduces challenges due to the heterogeneity of data sources (related to volume and quality), computational node capabilities, and environmental impact. This paper contributes to the advancement of Green AI by proposing a data-centric approach to Green Federated Learning. Specifically, we focus on reducing FL's environmental impact by minimizing the volume of training data. Our methodology involves the analysis of the characteristics of federated datasets, the selecting of an optimal subset of data based on quality metrics, and the choice of the federated nodes with the lowest environmental impact. We develop a comprehensive methodology that examines the influence of data-centric factors, such as data quality and volume, on FL training performance and carbon emissions. Building on these insights, we introduce an interactive recommendation system that optimizes FL configurations through data reduction, minimizing environmental impact during training. Applying this methodology to time series classification has demonstrated promising results in reducing the environmental impact of FL tasks.
Authors: Haolin Jin, Mengbai Xiao, Yuan Yuan, Xiao Zhang, Dongxiao Yu, Guanghui Zhang, Haoliang Wang
Abstract: The Transformer architecture has revolutionized deep learning, delivering the state-of-the-art performance in areas such as natural language processing, computer vision, and time series prediction. However, its core component, self-attention, has the quadratic time complexity relative to input sequence length, which hinders the scalability of Transformers. The exsiting approaches on optimizing self-attention either discard full-contextual information or lack of flexibility. In this work, we design DistrAttention, an effcient and flexible self-attention mechanism with the full context. DistrAttention achieves this by grouping data on the embedding dimensionality, usually referred to as $d$. We realize DistrAttention with a lightweight sampling and fusion method that exploits locality-sensitive hashing to group similar data. A block-wise grouping framework is further designed to limit the errors introduced by locality sensitive hashing. By optimizing the selection of block sizes, DistrAttention could be easily integrated with FlashAttention-2, gaining high-performance on modern GPUs. We evaluate DistrAttention with extensive experiments. The results show that our method is 37% faster than FlashAttention-2 on calculating self-attention. In ViT inference, DistrAttention is the fastest and the most accurate among approximate self-attention mechanisms. In Llama3-1B, DistrAttention still achieves the lowest inference time with only 1% accuray loss.
Authors: Songxuan Shi
Abstract: This paper explores the generative capabilities of Autoencoders (AEs) and establishes connections between Variational Autoencoders (VAEs) and Vector Quantized-Variational Autoencoders (VQ-VAEs) through a reformulated training framework. We demonstrate that AEs exhibit generative potential via latent space interpolation and perturbation, albeit limited by undefined regions in the encoding space. To address this, we propose a new VAE-like training method that introduces clustering centers to enhance data compactness and ensure well-defined latent spaces without relying on traditional KL divergence or reparameterization techniques. Experimental results on MNIST, CelebA, and FashionMNIST datasets show smooth interpolative transitions, though blurriness persists. Extending this approach to multiple learnable vectors, we observe a natural progression toward a VQ-VAE-like model in continuous space. However, when the encoder outputs multiple vectors, the model degenerates into a discrete Autoencoder (VQ-AE), which combines image fragments without learning semantic representations. Our findings highlight the critical role of encoding space compactness and dispersion in generative modeling and provide insights into the intrinsic connections between VAEs and VQ-VAEs, offering a new perspective on their design and limitations.
Authors: Rishi Parekh, Saisubramaniam Gopalakrishnan, Zishan Ahmad, Anirudh Deodhar
Abstract: Analyzing large, complex output datasets from Discrete Event Simulations (DES) of warehouse operations to identify bottlenecks and inefficiencies is a critical yet challenging task, often demanding significant manual effort or specialized analytical tools. Our framework integrates Knowledge Graphs (KGs) and Large Language Model (LLM)-based agents to analyze complex Discrete Event Simulation (DES) output data from warehouse operations. It transforms raw DES data into a semantically rich KG, capturing relationships between simulation events and entities. An LLM-based agent uses iterative reasoning, generating interdependent sub-questions. For each sub-question, it creates Cypher queries for KG interaction, extracts information, and self-reflects to correct errors. This adaptive, iterative, and self-correcting process identifies operational issues mimicking human analysis. Our DES approach for warehouse bottleneck identification, tested with equipment breakdowns and process irregularities, outperforms baseline methods. For operational questions, it achieves near-perfect pass rates in pinpointing inefficiencies. For complex investigative questions, we demonstrate its superior diagnostic ability to uncover subtle, interconnected issues. This work bridges simulation modeling and AI (KG+LLM), offering a more intuitive method for actionable insights, reducing time-to-insight, and enabling automated warehouse inefficiency evaluation and diagnosis.
Authors: Aritz P\'erez, Carlos Echegoyen, Guzm\'an Santaf\'e
Abstract: Federated learning is a paradigm of increasing relevance in real world applications, aimed at building a global model across a network of heterogeneous users without requiring the sharing of private data. We focus on model learning over decentralized architectures, where users collaborate directly to update the global model without relying on a central server. In this context, the current paper proposes a novel approach to collaboratively learn probabilistic generative classifiers with a parametric form. The framework is composed by a communication network over a set of local nodes, each of one having its own local data, and a local updating rule. The proposal involves sharing local statistics with neighboring nodes, where each node aggregates the neighbors' information and iteratively learns its own local classifier, which progressively converges to a global model. Extensive experiments demonstrate that the algorithm consistently converges to a globally competitive model across a wide range of network topologies, network sizes, local dataset sizes, and extreme non-i.i.d. data distributions.
Authors: Zhuokun Chen, Zeren Chen, Jiahao He, Mingkui Tan, Jianfei Cai, Bohan Zhuang
Abstract: Chain-of-thought (CoT) reasoning enhances the problem-solving capabilities of large language models by encouraging step-by-step intermediate reasoning during inference. While effective, CoT introduces substantial computational overhead due to its reliance on autoregressive decoding over long token sequences. Existing acceleration strategies either reduce sequence length through early stopping or compressive reward designs, or improve decoding speed via speculative decoding with smaller models. However, speculative decoding suffers from limited speedup when the agreement between small and large models is low, and fails to exploit the potential advantages of small models in producing concise intermediate reasoning. In this paper, we present R-Stitch, a token-level, confidence-based hybrid decoding framework that accelerates CoT inference by switching between a small language model (SLM) and a large language model (LLM) along the reasoning trajectory. R-Stitch uses the SLM to generate tokens by default and delegates to the LLM only when the SLM's confidence falls below a threshold. This design avoids full-sequence rollback and selectively invokes the LLM on uncertain steps, preserving both efficiency and answer quality. R-Stitch is model-agnostic, training-free, and compatible with standard decoding pipelines. Experiments on math reasoning benchmarks demonstrate that R-Stitch achieves up to 85\% reduction in inference latency with negligible accuracy drop, highlighting its practical effectiveness in accelerating CoT reasoning.
Authors: Yan Zeng, Shenglan Nie, Feng Xie, Libo Huang, Peng Wu, Zhi Geng
Abstract: Imitation learning from demonstrations usually suffers from the confounding effects of unmeasured variables (i.e., unmeasured confounders) on the states and actions. If ignoring them, a biased estimation of the policy would be entailed. To break up this confounding gap, in this paper, we take the best of the strong power of instrumental variables (IV) and propose a Confounded Causal Imitation Learning (C2L) model. This model accommodates confounders that influence actions across multiple timesteps, rather than being restricted to immediate temporal dependencies. We develop a two-stage imitation learning framework for valid IV identification and policy optimization. In particular, in the first stage, we construct a testing criterion based on the defined pseudo-variable, with which we achieve identifying a valid IV for the C2L models. Such a criterion entails the sufficient and necessary identifiability conditions for IV validity. In the second stage, with the identified IV, we propose two candidate policy learning approaches: one is based on a simulator, while the other is offline. Extensive experiments verified the effectiveness of identifying the valid IV as well as learning the policy.
Authors: Zijie Guo, Jiong Wang, Xiaoyu Yue, Wangxu Wei, Zhe Jiang, Wanghan Xu, Ben Fei, Wenlong Zhang, Xinyu Gu, Lijing Cheng, Jing-Jia Luo, Chao Li, Yaqiang Wang, Tao Chen, Wanli Ouyang, Fenghua Ling, Lei Bai
Abstract: Modern Earth science is at an inflection point. The vast, fragmented, and complex nature of Earth system data, coupled with increasingly sophisticated analytical demands, creates a significant bottleneck for rapid scientific discovery. Here we introduce EarthLink, the first AI agent designed as an interactive copilot for Earth scientists. It automates the end-to-end research workflow, from planning and code generation to multi-scenario analysis. Unlike static diagnostic tools, EarthLink can learn from user interaction, continuously refining its capabilities through a dynamic feedback loop. We validated its performance on a number of core scientific tasks of climate change, ranging from model-observation comparisons to the diagnosis of complex phenomena. In a multi-expert evaluation, EarthLink produced scientifically sound analyses and demonstrated an analytical competency that was rated as comparable to specific aspects of a human junior researcher's workflow. Additionally, its transparent, auditable workflows and natural language interface empower scientists to shift from laborious manual execution to strategic oversight and hypothesis generation. EarthLink marks a pivotal step towards an efficient, trustworthy, and collaborative paradigm for Earth system research in an era of accelerating global change. The system is accessible at our website https://earthlink.intern-ai.org.cn.
Authors: Rui Wu, Nikola Kovachki, Burigede Liu
Abstract: Recent developments in mechanical, aerospace, and structural engineering have driven a growing need for efficient ways to model and analyse structures at much larger and more complex scales than before. While established numerical methods like the Finite Element Method remain reliable, they often struggle with computational cost and scalability when dealing with large and geometrically intricate problems. In recent years, neural network-based methods have shown promise because of their ability to efficiently approximate nonlinear mappings. However, most existing neural approaches are still largely limited to simple domains, which makes it difficult to apply to real-world PDEs involving complex geometries. In this paper, we propose a learning-based domain decomposition method (L-DDM) that addresses this gap. Our approach uses a single, pre-trained neural operator-originally trained on simple domains-as a surrogate model within a domain decomposition scheme, allowing us to tackle large and complicated domains efficiently. We provide a general theoretical result on the existence of neural operator approximations in the context of domain decomposition solution of abstract PDEs. We then demonstrate our method by accurately approximating solutions to elliptic PDEs with discontinuous microstructures in complex geometries, using a physics-pretrained neural operator (PPNO). Our results show that this approach not only outperforms current state-of-the-art methods on these challenging problems, but also offers resolution-invariance and strong generalization to microstructural patterns unseen during training.
Authors: Rongwei Lu, Jingyan Jiang, Chunyang Li, Haotian Dong, Xingguang Wei, Delin Cai, Zhi Wang
Abstract: Distributed machine learning in high end-to-end latency and low, varying bandwidth network environments undergoes severe throughput degradation. Due to its low communication requirements, distributed SGD (D-SGD) remains the mainstream optimizer in such challenging networks, but it still suffers from significant throughput reduction. To mitigate these limitations, existing approaches typically employ gradient compression and delayed aggregation to alleviate low bandwidth and high latency, respectively. To address both challenges simultaneously, these strategies are often combined, introducing a complex three-way trade-off among compression ratio, staleness (delayed synchronization steps), and model convergence rate. To achieve the balance under varying bandwidth conditions, an adaptive policy is required to dynamically adjust these parameters. Unfortunately, existing works rely on static heuristic strategies due to the lack of theoretical guidance, which prevents them from achieving this goal. This study fills in this theoretical gap by introducing a new theoretical tool, decomposing the joint optimization problem into a traditional convergence rate analysis with multiple analyzable noise terms. We are the first to reveal that staleness exponentially amplifies the negative impact of gradient compression on training performance, filling a critical gap in understanding how compressed and delayed gradients affect training. Furthermore, by integrating the convergence rate with a network-aware time minimization condition, we propose DeCo-SGD, which dynamically adjusts the compression ratio and staleness based on the real-time network condition and training task. DeCo-SGD achieves up to 5.07 and 1.37 speed-ups over D-SGD and static strategy in high-latency and low, varying bandwidth networks, respectively.
Authors: Rafael Ayll\'on-Gavil\'an, David Guijo-Rubio, Antonio Manuel G\'omez-Orellana, Francisco B\'erchez-Moreno, V\'ictor Manuel Vargas-Yun, Pedro A. Guti\'errez
Abstract: An ordinal classification (OC) problem corresponds to a special type of classification characterised by the presence of a natural order relationship among the classes. This type of problem can be found in a number of real-world applications, motivating the design and development of many ordinal methodologies over the last years. However, it is important to highlight that the development of the OC field suffers from one main disadvantage: the lack of a comprehensive set of datasets on which novel approaches to the literature have to be benchmarked. In order to approach this objective, this manuscript from the University of C\'ordoba (UCO), which have previous experience on the OC field, provides the literature with a publicly available repository of tabular data for a robust validation of novel OC approaches, namely TOC-UCO (Tabular Ordinal Classification repository of the UCO). Specifically, this repository includes a set of $46$ tabular ordinal datasets, preprocessed under a common framework and ensured to have a reasonable number of patterns and an appropriate class distribution. We also provide the sources and preprocessing steps of each dataset, along with details on how to benchmark a novel approach using the TOC-UCO repository. For this, indices for $30$ different randomised train-test partitions are provided to facilitate the reproducibility of the experiments.
Authors: Chuzhan Hao, Wenfeng Feng, Yuewei Zhang, Hao Wang
Abstract: Multi-step agentic retrieval systems based on large language models (LLMs) have demonstrated remarkable performance in complex information search tasks. However, these systems still face significant challenges in practical applications, particularly in generating factually inconsistent intermediate queries and inefficient search trajectories, which can lead to reasoning deviations or redundant computations. To address these issues, we propose DynaSearcher, an innovative search agent enhanced by dynamic knowledge graphs and multi-reward reinforcement learning (RL). Specifically, our system leverages knowledge graphs as external structured knowledge to guide the search process by explicitly modeling entity relationships, thereby ensuring factual consistency in intermediate queries and mitigating biases from irrelevant information. Furthermore, we employ a multi-reward RL framework for fine-grained control over training objectives such as retrieval accuracy, efficiency, and response quality. This framework promotes the generation of high-quality intermediate queries and comprehensive final answers, while discouraging unnecessary exploration and minimizing information omissions or redundancy. Experimental results demonstrate that our approach achieves state-of-the-art answer accuracy on six multi-hop question answering datasets, matching frontier LLMs while using only small-scale models and limited computational resources. Furthermore, our approach demonstrates strong generalization and robustness across diverse retrieval environments and larger-scale models, highlighting its broad applicability.
Authors: Hao Dai, Chong Tang, Jagmohan Chauhan
Abstract: Continual learning (CL) with long-tailed data distributions remains a critical challenge for real-world AI systems, where models must sequentially adapt to new classes while retaining knowledge of old ones, despite severe class imbalance. Existing methods struggle to balance stability and plasticity, often collapsing under extreme sample scarcity. To address this, we propose ViRN, a novel CL framework that integrates variational inference (VI) with distributional trilateration for robust long-tailed learning. First, we model class-conditional distributions via a Variational Autoencoder to mitigate bias toward head classes. Second, we reconstruct tail-class distributions via Wasserstein distance-based neighborhood retrieval and geometric fusion, enabling sample-efficient alignment of tail-class representations. Evaluated on six long-tailed classification benchmarks, including speech (e.g., rare acoustic events, accents) and image tasks, ViRN achieves a 10.24% average accuracy gain over state-of-the-art methods.
Authors: Hao Dai, Jagmohan Chauhan
Abstract: Continual Generalized Category Discovery (C-GCD) faces a critical challenge: incrementally learning new classes from unlabeled data streams while preserving knowledge of old classes. Existing methods struggle with catastrophic forgetting, especially when unlabeled data mixes known and novel categories. We address this by analyzing C-GCD's forgetting dynamics through a Bayesian lens, revealing that covariance misalignment between old and new classes drives performance degradation. Building on this insight, we propose Variational Bayes C-GCD (VB-CGCD), a novel framework that integrates variational inference with covariance-aware nearest-class-mean classification. VB-CGCD adaptively aligns class distributions while suppressing pseudo-label noise via stochastic variational updates. Experiments show VB-CGCD surpasses prior art by +15.21% with the overall accuracy in the final session on standard benchmarks. We also introduce a new challenging benchmark with only 10% labeled data and extended online phases, VB-CGCD achieves a 67.86% final accuracy, significantly higher than state-of-the-art (38.55%), demonstrating its robust applicability across diverse scenarios. Code is available at: https://github.com/daihao42/VB-CGCD
Authors: Yutong Liu, Cairong Zhao, Guosheng Hu
Abstract: For large language models (LLMs), post-training quantization (PTQ) can significantly reduce memory footprint and computational overhead. Model quantization is a rapidly evolving research field. Though many papers have reported breakthrough performance, they may not conduct experiments on the same ground since one quantization method usually contains multiple components. In addition, analyzing the theoretical connections among existing methods is crucial for in-depth understanding. To bridge these gaps, we conduct an extensive review of state-of-the-art methods and perform comprehensive evaluations on the same ground to ensure fair comparisons. To our knowledge, this fair and extensive investigation remains critically important yet underexplored. To better understand the theoretical connections, we decouple the published quantization methods into two steps: pre-quantization transformation and quantization error mitigation. We define the former as a preprocessing step applied before quantization to reduce the impact of outliers, making the data distribution flatter and more suitable for quantization. Quantization error mitigation involves techniques that offset the errors introduced during quantization, thereby enhancing model performance. We evaluate and analyze the impact of different components of quantization methods. Additionally, we analyze and evaluate the latest MXFP4 data format and its performance. Our experimental results demonstrate that optimized rotation and scaling yield the best performance for pre-quantization transformation, and combining low-rank compensation with GPTQ occasionally outperforms using GPTQ alone for quantization error mitigation. Furthermore, we explore the potential of the latest MXFP4 quantization and reveal that the optimal pre-quantization transformation strategy for INT4 does not generalize well to MXFP4, inspiring further investigation.
Authors: Arsha Niksa, Hooman Zare, Ali Shahrabi, Hanieh Hatami, Mohammadreza Razvan
Abstract: We present a topological pipeline for automated multiclass emotion recognition from eye-tracking data. Delay embeddings of gaze trajectories are analyzed using persistent homology. From the resulting persistence diagrams, we extract shape-based features such as mean persistence, maximum persistence, and entropy. A random forest classifier trained on these features achieves up to $75.6\%$ accuracy on four emotion classes, which are the quadrants the Circumplex Model of Affect. The results demonstrate that persistence diagram geometry effectively encodes discriminative gaze dynamics, suggesting a promising topological approach for affective computing and human behavior analysis.
Authors: Guanqin Zhang, Kota Fukuda, Zhenya Zhang, H. M. N. Dilum Bandara, Shiping Chen, Jianjun Zhao, Yulei Sui
Abstract: The vulnerability of neural networks to adversarial perturbations has necessitated formal verification techniques that can rigorously certify the quality of neural networks. As the state-of-the-art, branch and bound (BaB) is a "divide-and-conquer" strategy that applies off-the-shelf verifiers to sub-problems for which they perform better. While BaB can identify the sub-problems that are necessary to be split, it explores the space of these sub-problems in a naive "first-come-first-serve" manner, thereby suffering from an issue of inefficiency to reach a verification conclusion. To bridge this gap, we introduce an order over different sub-problems produced by BaB, concerning with their different likelihoods of containing counterexamples. Based on this order, we propose a novel verification framework Oliva that explores the sub-problem space by prioritizing those sub-problems that are more likely to find counterexamples, in order to efficiently reach the conclusion of the verification. Even if no counterexample can be found in any sub-problem, it only changes the order of visiting different sub-problem and so will not lead to a performance degradation. Specifically, Oliva has two variants, including $Oliva^{GR}$, a greedy strategy that always prioritizes the sub-problems that are more likely to find counterexamples, and $Oliva^{SA}$, a balanced strategy inspired by simulated annealing that gradually shifts from exploration to exploitation to locate the globally optimal sub-problems. We experimentally evaluate the performance of Oliva on 690 verification problems spanning over 5 models with datasets MNIST and CIFAR10. Compared to the state-of-the-art approaches, we demonstrate the speedup of Oliva for up to 25X in MNIST, and up to 80X in CIFAR10.
Authors: Shusen Ma, Yun-Bo Zhao, Yu Kang
Abstract: Multivariate time series forecasting has drawn increasing attention due to its practical importance. Existing approaches typically adopt either channel-mixing (CM) or channel-independence (CI) strategies. CM strategy can capture inter-variable dependencies but fails to discern variable-specific temporal patterns. CI strategy improves this aspect but fails to fully exploit cross-variable dependencies like CM. Hybrid strategies based on feature fusion offer limited generalization and interpretability. To address these issues, we propose C3RL, a novel representation learning framework that jointly models both CM and CI strategies. Motivated by contrastive learning in computer vision, C3RL treats the inputs of the two strategies as transposed views and builds a siamese network architecture: one strategy serves as the backbone, while the other complements it. By jointly optimizing contrastive and prediction losses with adaptive weighting, C3RL balances representation and forecasting performance. Extensive experiments on seven models show that C3RL boosts the best-case performance rate to 81.4\% for models based on CI strategy and to 76.3\% for models based on CM strategy, demonstrating strong generalization and effectiveness. The code will be available once the paper is accepted.
Authors: Junhua Liu, Roy Ka-Wei Lee, Kwan Hui Lim
Abstract: Human decision-making in high-stakes domains often relies on expertise and heuristics, but is vulnerable to hard-to-detect cognitive biases that threaten fairness and long-term outcomes. This work presents a novel approach to enhancing complex decision-making workflows through the integration of hierarchical learning alongside various enhancements. Focusing on university admissions as a representative high-stakes domain, we propose BGM-HAN, an enhanced Byte-Pair Encoded, Gated Multi-head Hierarchical Attention Network, designed to effectively model semi-structured applicant data. BGM-HAN captures multi-level representations that are crucial for nuanced assessment, improving both interpretability and predictive performance. Experimental results on real admissions data demonstrate that our proposed model significantly outperforms both state-of-the-art baselines from traditional machine learning to large language models, offering a promising framework for augmenting decision-making in domains where structure, context, and fairness matter. Source code is available at: https://github.com/junhua/bgm-han.
Authors: Xianbiao Qi, Marco Chen, Wenjie Xiao, Jiaquan Ye, Yelin He, Chun-Guang Li, Zhouchen Lin
Abstract: Transformers have become the de facto backbone of modern deep learning, yet their training typically demands an advanced optimizer with adaptive learning rate like AdamW, rather than a momentum SGDW (mSGDW). Previous works show that it is mainly due to a heavy-tailed distribution of the gradients. In this paper, we introduce a Deeply Normalized Transformer (DNT), which is meticulously engineered to overcome this limitation enabling seamless training with vanilla mSGDW while yielding comparable performance to the Transformers trained via AdamW. To be specific, in DNT, we strategically integrate normalization techniques at proper positions in the Transformers to effectively modulate the Jacobian matrices of each layer, balance the influence of weights, activations, and their interactions, and thus enable the distributions of gradients concentrated. We provide both theoretical justifications of the normalization technique used in our DNT and extensive empirical evaluation on two popular Transformer architectures to validate that: a) DNT outperforms its counterparts (\ie, ViT and GPT), and b) DNT can be effectively trained with vanilla mSGDW.
Authors: Nazar Buzun, Daniil Shlenskii, Maxim Bobrin, Dmitry V. Dylov
Abstract: Optimal transport (OT) has become a natural framework for guiding the probability flows. Yet, the majority of recent generative models assume trivial geometry (e.g., Euclidean) and rely on strong density-estimation assumptions, yielding trajectories that do not respect the true principles of optimality in the underlying manifold. We present Hamiltonian Optimal Transport Advection (HOTA), a Hamilton-Jacobi-Bellman based method that tackles the dual dynamical OT problem explicitly through Kantorovich potentials, enabling efficient and scalable trajectory optimization. Our approach effectively evades the need for explicit density modeling, performing even when the cost functionals are non-smooth. Empirically, HOTA outperforms all baselines in standard benchmarks, as well as in custom datasets with non-differentiable costs, both in terms of feasibility and optimality.
Authors: Yao Wang, Jiannan Li, Yue Kang, Shanxing Gao, Zhenxin Xiao
Abstract: The matrix contextual bandit (CB), as an extension of the well-known multi-armed bandit, is a powerful framework that has been widely applied in sequential decision-making scenarios involving low-rank structure. In many real-world scenarios, such as online advertising and recommender systems, additional graph information often exists beyond the low-rank structure, that is, the similar relationships among users/items can be naturally captured through the connectivity among nodes in the corresponding graphs. However, existing matrix CB methods fail to explore such graph information, and thereby making them difficult to generate effective decision-making policies. To fill in this void, we propose in this paper a novel matrix CB algorithmic framework that builds upon the classical upper confidence bound (UCB) framework. This new framework can effectively integrate both the low-rank structure and graph information in a unified manner. Specifically, it involves first solving a joint nuclear norm and matrix Laplacian regularization problem, followed by the implementation of a graph-based generalized linear version of the UCB algorithm. Rigorous theoretical analysis demonstrates that our procedure outperforms several popular alternatives in terms of cumulative regret bound, owing to the effective utilization of graph information. A series of synthetic and real-world data experiments are conducted to further illustrate the merits of our procedure.
Authors: Shahil Shaik, Jonathon M. Smereka, Yue Wang
Abstract: Generalized Advantage Estimation (GAE) has been used to mitigate the computational complexity of reinforcement learning (RL) by employing an exponentially weighted estimation of the advantage function to reduce the variance in policy gradient estimates. Despite its effectiveness, GAE is not designed to handle value distributions integral to distributional RL, which can capture the inherent stochasticity in systems and is hence more robust to system noises. To address this gap, we propose a novel approach that utilizes the optimal transport theory to introduce a Wasserstein-like directional metric, which measures both the distance and the directional discrepancies between probability distributions. Using the exponentially weighted estimation, we leverage this Wasserstein-like directional metric to derive distributional GAE (DGAE). Similar to traditional GAE, our proposed DGAE provides a low-variance advantage estimate with controlled bias, making it well-suited for policy gradient algorithms that rely on advantage estimation for policy updates. We integrated DGAE into three different policy gradient methods. Algorithms were evaluated across various OpenAI Gym environments and compared with the baselines with traditional GAE to assess the performance.
Authors: Aymeric Dieuleveut, Gersende Fort, Mahmoud Hegazy, Hoi-To Wai
Abstract: This paper proposes a unified approach for designing stochastic optimization algorithms that robustly scale to the federated learning setting. Our work studies a class of Majorize-Minimization (MM) problems, which possesses a linearly parameterized family of majorizing surrogate functions. This framework encompasses (proximal) gradient-based algorithms for (regularized) smooth objectives, the Expectation Maximization algorithm, and many problems seen as variational surrogate MM. We show that our framework motivates a unifying algorithm called Stochastic Approximation Stochastic Surrogate MM (\SSMM), which includes previous stochastic MM procedures as special instances. We then extend \SSMM\ to the federated setting, while taking into consideration common bottlenecks such as data heterogeneity, partial participation, and communication constraints; this yields \QSMM. The originality of \QSMM\ is to learn locally and then aggregate information characterizing the \textit{surrogate majorizing function}, contrary to classical algorithms which learn and aggregate the \textit{original parameter}. Finally, to showcase the flexibility of this methodology beyond our theoretical setting, we use it to design an algorithm for computing optimal transport maps in the federated setting.
Authors: Amandeep Singh Bhatia, Sabre Kais
Abstract: Federated Learning (FL) has become increasingly popular across different sectors, offering a way for clients to work together to train a global model without sharing sensitive data. It involves multiple rounds of communication between the global model and participating clients, which introduces several challenges like high communication costs, heterogeneous client data, prolonged processing times, and increased vulnerability to privacy threats. In recent years, the convergence of federated learning and parameterized quantum circuits has sparked significant research interest, with promising implications for fields such as healthcare and finance. By enabling decentralized training of quantum models, it allows clients or institutions to collaboratively enhance model performance and outcomes while preserving data privacy. Recognizing that Fisher information can quantify the amount of information that a quantum state carries under parameter changes, thereby providing insight into its geometric and statistical properties. We intend to leverage this property to address the aforementioned challenges. In this work, we propose a Quantum Federated Learning (QFL) algorithm that makes use of the Fisher information computed on local client models, with data distributed across heterogeneous partitions. This approach identifies the critical parameters that significantly influence the quantum model's performance, ensuring they are preserved during the aggregation process. Our research assessed the effectiveness and feasibility of QFL by comparing its performance against other variants, and exploring the benefits of incorporating Fisher information in QFL settings. Experimental results on ADNI and MNIST datasets demonstrate the effectiveness of our approach in achieving better performance and robustness against the quantum federated averaging method.
Authors: Moncef Garouani, Ayah Barhrhouj, Olivier Teste
Abstract: Ensemble Machine Learning (EML) techniques, especially stacking, have been shown to improve predictive performance by combining multiple base models. However, they are often criticized for their lack of interpretability. In this paper, we introduce XStacking, an effective and inherently explainable framework that addresses this limitation by integrating dynamic feature transformation with model-agnostic Shapley additive explanations. This enables stacked models to retain their predictive accuracy while becoming inherently explainable. We demonstrate the effectiveness of the framework on 29 datasets, achieving improvements in both the predictive effectiveness of the learning space and the interpretability of the resulting models. XStacking offers a practical and scalable solution for responsible ML.
Authors: Alexander David Goldie, Zilin Wang, Jakob Nicolaus Foerster, Shimon Whiteson
Abstract: The process of meta-learning algorithms from data, instead of relying on manual design, is growing in popularity as a paradigm for improving the performance of machine learning systems. Meta-learning shows particular promise for reinforcement learning (RL), where algorithms are often adapted from supervised or unsupervised learning despite their suboptimality for RL. However, until now there has been a severe lack of comparison between different meta-learning algorithms, such as using evolution to optimise over black-box functions or LLMs to propose code. In this paper, we carry out this empirical comparison of the different approaches when applied to a range of meta-learned algorithms which target different parts of the RL pipeline. In addition to meta-train and meta-test performance, we also investigate factors including the interpretability, sample cost and train time for each meta-learning algorithm. Based on these findings, we propose several guidelines for meta-learning new RL algorithms which will help ensure that future learned algorithms are as performant as possible.
Authors: Penukonda Naga Chandana, Tejas Srivastava, Gowtham R. Kurri, V. Lalitha
Abstract: Dual discriminator generative adversarial networks (D2 GANs) were introduced to mitigate the problem of mode collapse in generative adversarial networks. In D2 GANs, two discriminators are employed alongside a generator: one discriminator rewards high scores for samples from the true data distribution, while the other favors samples from the generator. In this work, we first introduce dual discriminator $\alpha$-GANs (D2 $\alpha$-GANs), which combines the strengths of dual discriminators with the flexibility of a tunable loss function, $\alpha$-loss. We further generalize this approach to arbitrary functions defined on positive reals, leading to a broader class of models we refer to as generalized dual discriminator generative adversarial networks. For each of these proposed models, we provide theoretical analysis and show that the associated min-max optimization reduces to the minimization of a linear combination of an $f$-divergence and a reverse $f$-divergence. This generalizes the known simplification for D2-GANs, where the objective reduces to a linear combination of the KL-divergence and the reverse KL-divergence. Finally, we perform experiments on 2D synthetic data and use multiple performance metrics to capture various advantages of our GANs.
Authors: Jiazhen Chen, Zheng Ma, Sichao Fu, Mingbin Feng, Tony S. Wirjanto, Weihua Ou
Abstract: Graph class-incremental learning (GCIL) allows graph neural networks (GNNs) to adapt to evolving graph analytical tasks by incrementally learning new class knowledge while retaining knowledge of old classes. Existing GCIL methods primarily focus on a closed-set assumption, where all test samples are presumed to belong to previously known classes. Such an assumption restricts their applicability in real-world scenarios, where unknown classes naturally emerge during inference, and are absent during training. In this paper, we explore a more challenging open-set graph class-incremental learning scenario with two intertwined challenges: catastrophic forgetting of old classes, which impairs the detection of unknown classes, and inadequate open-set recognition, which destabilizes the retention of learned knowledge. To address the above problems, a novel OGCIL framework is proposed, which utilizes pseudo-sample embedding generation to effectively mitigate catastrophic forgetting and enable robust detection of unknown classes. To be specific, a prototypical conditional variational autoencoder is designed to synthesize node embeddings for old classes, enabling knowledge replay without storing raw graph data. To handle unknown classes, we employ a mixing-based strategy to generate out-of-distribution (OOD) samples from pseudo in-distribution and current node embeddings. A novel prototypical hypersphere classification loss is further proposed, which anchors in-distribution embeddings to their respective class prototypes, while repelling OOD embeddings away. Instead of assigning all unknown samples into one cluster, our proposed objective function explicitly models them as outliers through prototype-aware rejection regions, ensuring a robust open-set recognition. Extensive experiments on five benchmarks demonstrate the effectiveness of OGCIL over existing GCIL and open-set GNN methods.
Authors: Jialiang Wang, Xianming Liu, Xiong Zhou, Gangfeng Hu, Deming Zhai, Junjun Jiang, Xiangyang Ji
Abstract: Learning with noisy labels is a crucial task for training accurate deep neural networks. To mitigate label noise, prior studies have proposed various robust loss functions, particularly symmetric losses. Nevertheless, symmetric losses usually suffer from the underfitting issue due to the overly strict constraint. To address this problem, the Active Passive Loss (APL) jointly optimizes an active and a passive loss to mutually enhance the overall fitting ability. Within APL, symmetric losses have been successfully extended, yielding advanced robust loss functions. Despite these advancements, emerging theoretical analyses indicate that asymmetric losses, a new class of robust loss functions, possess superior properties compared to symmetric losses. However, existing asymmetric losses are not compatible with advanced optimization frameworks such as APL, limiting their potential and applicability. Motivated by this theoretical gap and the prospect of asymmetric losses, we extend the asymmetric loss to the more complex passive loss scenario and propose the Asymetric Mean Square Error (AMSE), a novel asymmetric loss. We rigorously establish the necessary and sufficient condition under which AMSE satisfies the asymmetric condition. By substituting the traditional symmetric passive loss in APL with our proposed AMSE, we introduce a novel robust loss framework termed Joint Asymmetric Loss (JAL). Extensive experiments demonstrate the effectiveness of our method in mitigating label noise. Code available at: https://github.com/cswjl/joint-asymmetric-loss
Authors: Taha Ceritli, Ondrej Bohdal, Mete Ozay, Jijoong Moon, Kyeng-Hun Lee, Hyeonmok Ko, Umberto Michieli
Abstract: Large language models (LLMs) often leverage adapters, such as low-rank-based adapters, to achieve strong performance on downstream tasks. However, storing a separate adapter for each task significantly increases memory requirements, posing a challenge for resource-constrained environments such as mobile devices. Although model merging techniques can reduce storage costs, they typically result in substantial performance degradation. In this work, we introduce HydraOpt, a new model merging technique that capitalizes on the inherent similarities between the matrices of low-rank adapters. Unlike existing methods that produce a fixed trade-off between storage size and performance, HydraOpt allows us to navigate this spectrum of efficiency and performance. Our experiments show that HydraOpt significantly reduces storage size (48% reduction) compared to storing all adapters, while achieving competitive performance (0.2-1.8% drop). Furthermore, it outperforms existing merging techniques in terms of performance at the same or slightly worse storage efficiency.
Authors: Melih Barsbey, Ant\^onio H. Ribeiro, Umut \c{S}im\c{s}ekli, Tolga Birdal
Abstract: Modern neural networks are expected to simultaneously satisfy a host of desirable properties: accurate fitting to training data, generalization to unseen inputs, parameter and computational efficiency, and robustness to adversarial perturbations. While compressibility and robustness have each been studied extensively, a unified understanding of their interaction still remains elusive. In this work, we develop a principled framework to analyze how different forms of compressibility - such as neuron-level sparsity and spectral compressibility - affect adversarial robustness. We show that these forms of compression can induce a small number of highly sensitive directions in the representation space, which adversaries can exploit to construct effective perturbations. Our analysis yields a simple yet instructive robustness bound, revealing how neuron and spectral compressibility impact $L_\infty$ and $L_2$ robustness via their effects on the learned representations. Crucially, the vulnerabilities we identify arise irrespective of how compression is achieved - whether via regularization, architectural bias, or implicit learning dynamics. Through empirical evaluations across synthetic and realistic tasks, we confirm our theoretical predictions, and further demonstrate that these vulnerabilities persist under adversarial training and transfer learning, and contribute to the emergence of universal adversarial perturbations. Our findings show a fundamental tension between structured compressibility and robustness, and suggest new pathways for designing models that are both efficient and secure.
Authors: Zihao Li, Zhichen Zeng, Xiao Lin, Feihao Fang, Yanru Qu, Zhe Xu, Zhining Liu, Xuying Ning, Tianxin Wei, Ge Liu, Hanghang Tong, Jingrui He
Abstract: Over the past decade, advances in generative modeling, such as generative adversarial networks, masked autoencoders, and diffusion models, have significantly transformed biological research and discovery, enabling breakthroughs in molecule design, protein generation, drug discovery, and beyond. At the same time, biological applications have served as valuable testbeds for evaluating the capabilities of generative models. Recently, flow matching has emerged as a powerful and efficient alternative to diffusion-based generative modeling, with growing interest in its application to problems in biology and life sciences. This paper presents the first comprehensive survey of recent developments in flow matching and its applications in biological domains. We begin by systematically reviewing the foundations and variants of flow matching, and then categorize its applications into three major areas: biological sequence modeling, molecule generation and design, and peptide and protein generation. For each, we provide an in-depth review of recent progress. We also summarize commonly used datasets and software tools, and conclude with a discussion of potential future directions. The corresponding curated resources are available at https://github.com/Violet24K/Awesome-Flow-Matching-Meets-Biology.
URLs: https://github.com/Violet24K/Awesome-Flow-Matching-Meets-Biology.
Authors: Anisha Gunjal, Anthony Wang, Elaine Lau, Vaskar Nath, Bing Liu, Sean Hendryx
Abstract: Extending Reinforcement Learning with Verifiable Rewards (RLVR) to real-world tasks often requires balancing objective and subjective evaluation criteria. However, many such tasks lack a single, unambiguous ground truth-making it difficult to define reliable reward signals for post-training language models. While traditional preference-based methods offer a workaround, they rely on opaque reward functions that are difficult to interpret and prone to spurious correlations. We introduce $\textbf{Rubrics as Rewards}$ (RaR), a framework that uses structured, checklist-style rubrics as interpretable reward signals for on-policy training with GRPO. Our best RaR method yields up to a $28\%$ relative improvement on HealthBench-1k compared to simple Likert-based approaches, while matching or surpassing the performance of reward signals derived from expert-written references. By treating rubrics as structured reward signals, we show that RaR enables smaller-scale judge models to better align with human preferences and sustain robust performance across model scales.
Authors: Melih Barsbey, Lucas Prieto, Stefanos Zafeiriou, Tolga Birdal
Abstract: Robustness and resource-efficiency are two highly desirable properties for modern machine learning models. However, achieving them jointly remains a challenge. In this paper, we position high learning rates as a facilitator for simultaneously achieving robustness to spurious correlations and network compressibility. We demonstrate that large learning rates also produce desirable representation properties such as invariant feature utilization, class separation, and activation sparsity. Importantly, our findings indicate that large learning rates compare favorably to other hyperparameters and regularization methods, in consistently satisfying these properties in tandem. In addition to demonstrating the positive effect of large learning rates across diverse spurious correlation datasets, models, and optimizers, we also present strong evidence that the previously documented success of large learning rates in standard classification tasks is likely due to its effect on addressing hidden/rare spurious correlations in the training dataset.
Authors: Paul Saves
Abstract: The objective of this Philosophiae Doctor (Ph.D) thesis is to propose an efficient approach for optimizing a multidisciplinary black-box model when the optimization problem is constrained and involves a large number of mixed integer design variables (typically 100 variables). The targeted optimization approach, called EGO, is based on a sequential enrichment of an adaptive surrogate model and, in this context, GP surrogate models are one of the most widely used in engineering problems to approximate time-consuming high fidelity models. EGO is a heuristic BO method that performs well in terms of solution quality. However, like any other global optimization method, EGO suffers from the curse of dimensionality, meaning that its performance is satisfactory on lower dimensional problems, but deteriorates as the dimensionality of the optimization search space increases. For realistic aircraft design problems, the typical size of the design variables can even exceed 100 and, thus, trying to solve directly the problems using EGO is ruled out. The latter is especially true when the problems involve both continuous and categorical variables increasing even more the size of the search space. In this Ph.D thesis, effective parameterization tools are investigated, including techniques like partial least squares regression, to significantly reduce the number of design variables. Additionally, Bayesian optimization is adapted to handle discrete variables and high-dimensional spaces in order to reduce the number of evaluations when optimizing innovative aircraft concepts such as the "DRAGON" hybrid airplane to reduce their climate impact.
Authors: Peter Plantinga, Briac Cordelle, Dominique Lou\"er, Mirco Ravanelli, Denise Klein
Abstract: Using speech samples as a biomarker is a promising avenue for detecting and monitoring the progression of Parkinson's disease (PD), but there is considerable disagreement in the literature about how best to collect and analyze such data. Early research in detecting PD from speech used a sustained vowel phonation (SVP) task, while some recent research has explored recordings of more cognitively demanding tasks. To assess the role of language in PD detection, we tested pretrained models with varying data types and pretraining objectives and found that (1) text-only models match the performance of vocal-feature models, (2) multilingual Whisper outperforms self-supervised models whereas monolingual Whisper does worse, and (3) AudioSet pretraining improves performance on SVP but not spontaneous speech. These findings together highlight the critical role of language for the early detection of Parkinson's disease.
Authors: Peter Plantinga, Jen-Kai Chen, Roozbeh Sattari, Mirco Ravanelli, Denise Klein
Abstract: Speech holds promise as a cost-effective and non-invasive biomarker for neurological conditions such as Parkinson's disease (PD). While deep learning systems trained on raw audio can find subtle signals not available from hand-crafted features, their black-box nature hinders clinical adoption. To address this, we apply sparse autoencoders (SAEs) to uncover interpretable internal representations from a speech-based PD detection system. We introduce a novel mask-based activation for adapting SAEs to small biomedical datasets, creating sparse disentangled dictionary representations. These dictionary entries are found to have strong associations with characteristic articulatory deficits in PD speech, such as reduced spectral flux and increased spectral flatness in the low-energy regions highlighted by the model attention. We further show that the spectral flux is related to volumetric measurements of the putamen from MRI scans, demonstrating the potential of SAEs to reveal clinically relevant biomarkers for disease monitoring and diagnosis.
Authors: Xiaoran Xua, In-Ho Rab, Ravi Sankarc
Abstract: Lung diseases, including lung cancer and COPD, are significant health concerns globally. Traditional diagnostic methods can be costly, time-consuming, and invasive. This study investigates the use of semi supervised learning methods for lung sound signal detection using a model combination of MFCC+CNN. By introducing semi supervised learning modules such as Mix Match, Co-Refinement, and Co Refurbishing, we aim to enhance the detection performance while reducing dependence on manual annotations. With the add-on semi-supervised modules, the accuracy rate of the MFCC+CNN model is 92.9%, an increase of 3.8% to the baseline model. The research contributes to the field of lung disease sound detection by addressing challenges such as individual differences, feature insufficient labeled data.
Authors: Ismail Hossain, Sai Puppala, Md Jahangir Alam, Sajedul Talukder
Abstract: Social media platforms serve as a significant medium for sharing personal emotions, daily activities, and various life events, ensuring individuals stay informed about the latest developments. From the initiation of an account, users progressively expand their circle of friends or followers, engaging actively by posting, commenting, and sharing content. Over time, user behavior on these platforms evolves, influenced by demographic attributes and the networks they form. In this study, we present a novel approach that leverages open-source models Llama-3-Instruct, Mistral-7B-Instruct, Gemma-7B-IT through prompt engineering, combined with GPT-2, BERT, and RoBERTa using a joint embedding technique, to analyze and predict the evolution of user behavior on social media over their lifetime. Our experiments demonstrate the potential of these models to forecast future stages of a user's social evolution, including network changes, future connections, and shifts in user activities. Experimental results highlight the effectiveness of our approach, with GPT-2 achieving the lowest perplexity (8.21) in a Cross-modal configuration, outperforming RoBERTa (9.11) and BERT, and underscoring the importance of leveraging Cross-modal configurations for superior performance. This approach addresses critical challenges in social media, such as friend recommendations and activity predictions, offering insights into the trajectory of user behavior. By anticipating future interactions and activities, this research aims to provide early warnings about potential negative outcomes, enabling users to make informed decisions and mitigate risks in the long term.
Authors: \'Alvaro Ruiz-R\'odenas, Jaime Pujante S\'aez, Daniel Garc\'ia-Algora, Mario Rodr\'iguez B\'ejar, Jorge Blasco, Jos\'e Luis Hern\'andez-Ramos
Abstract: Cyber Threat Intelligence (CTI) mining involves extracting structured insights from unstructured threat data, enabling organizations to understand and respond to evolving adversarial behavior. A key task in CTI mining is mapping threat descriptions to MITRE ATT\&CK techniques. However, this process is often performed manually, requiring expert knowledge and substantial effort. Automated approaches face two major challenges: the scarcity of high-quality labeled CTI data and class imbalance, where many techniques have very few examples. While domain-specific Large Language Models (LLMs) such as SecureBERT have shown improved performance, most recent work focuses on model architecture rather than addressing the data limitations. In this work, we present SynthCTI, a data augmentation framework designed to generate high-quality synthetic CTI sentences for underrepresented MITRE ATT\&CK techniques. Our method uses a clustering-based strategy to extract semantic context from training data and guide an LLM in producing synthetic CTI sentences that are lexically diverse and semantically faithful. We evaluate SynthCTI on two publicly available CTI datasets, CTI-to-MITRE and TRAM, using LLMs with different capacity. Incorporating synthetic data leads to consistent macro-F1 improvements: for example, ALBERT improves from 0.35 to 0.52 (a relative gain of 48.6\%), and SecureBERT reaches 0.6558 (up from 0.4412). Notably, smaller models augmented with SynthCTI outperform larger models trained without augmentation, demonstrating the value of data generation methods for building efficient and effective CTI classification systems.
Authors: Isha Pandey, Pranav Gaikwad, Amruta Parulekar, Ganesh Ramakrishnan
Abstract: High-quality speech generation for low-resource languages, such as many Indian languages, remains a significant challenge due to limited data and diverse linguistic structures. Duration prediction is a critical component in many speech generation pipelines, playing a key role in modeling prosody and speech rhythm. While some recent generative approaches choose to omit explicit duration modeling, often at the cost of longer training times. We retain and explore this module to better understand its impact in the linguistically rich and data-scarce landscape of India. We train a non-autoregressive Continuous Normalizing Flow (CNF) based speech model using publicly available Indian language data and evaluate multiple duration prediction strategies for zero-shot, speaker-specific generation. Our comparative analysis on speech-infilling tasks reveals nuanced trade-offs: infilling based predictors improve intelligibility in some languages, while speaker-prompted predictors better preserve speaker characteristics in others. These findings inform the design and selection of duration strategies tailored to specific languages and tasks, underscoring the continued value of interpretable components like duration prediction in adapting advanced generative architectures to low-resource, multilingual settings.
Authors: Charlotte Jennings (National Pathology Imaging Cooperative, Leeds Teaching Hospitals NHS Trust, Leeds, UK), Andrew Broad (National Pathology Imaging Cooperative, Leeds Teaching Hospitals NHS Trust, Leeds, UK), Lucy Godson (National Pathology Imaging Cooperative, Leeds Teaching Hospitals NHS Trust, Leeds, UK), Emily Clarke (National Pathology Imaging Cooperative, Leeds Teaching Hospitals NHS Trust, Leeds, UK), David Westhead (University of Leeds, Leeds, UK), Darren Treanor (National Pathology Imaging Cooperative, Leeds Teaching Hospitals NHS Trust, Leeds, UK)
Abstract: Multimodal machine learning integrating histopathology and molecular data shows promise for cancer prognostication. We systematically reviewed studies combining whole slide images (WSIs) and high-throughput omics to predict overall survival. Searches of EMBASE, PubMed, and Cochrane CENTRAL (12/08/2024), plus citation screening, identified eligible studies. Data extraction used CHARMS; bias was assessed with PROBAST+AI; synthesis followed SWiM and PRISMA 2020. Protocol: PROSPERO (CRD42024594745). Forty-eight studies (all since 2017) across 19 cancer types met criteria; all used The Cancer Genome Atlas. Approaches included regularised Cox regression (n=4), classical ML (n=13), and deep learning (n=31). Reported c-indices ranged 0.550-0.857; multimodal models typically outperformed unimodal ones. However, all studies showed unclear/high bias, limited external validation, and little focus on clinical utility. Multimodal WSI-omics survival prediction is a fast-growing field with promising results but needs improved methodological rigor, broader datasets, and clinical evaluation. Funded by NPIC, Leeds Teaching Hospitals NHS Trust, UK (Project 104687), supported by UKRI Industrial Strategy Challenge Fund.
Authors: Antoni Kowalczuk, Dominik Hintersdorf, Lukas Struppek, Kristian Kersting, Adam Dziedzic, Franziska Boenisch
Abstract: Text-to-image diffusion models (DMs) have achieved remarkable success in image generation. However, concerns about data privacy and intellectual property remain due to their potential to inadvertently memorize and replicate training data. Recent mitigation efforts have focused on identifying and pruning weights responsible for triggering replication, based on the assumption that memorization can be localized. Our research assesses the robustness of these pruning-based approaches. We demonstrate that even after pruning, minor adjustments to text embeddings of input prompts are sufficient to re-trigger data replication, highlighting the fragility of these defenses. Furthermore, we challenge the fundamental assumption of memorization locality, by showing that replication can be triggered from diverse locations within the text embedding space, and follows different paths in the model. Our findings indicate that existing mitigation strategies are insufficient and underscore the need for methods that truly remove memorized content, rather than attempting to suppress its retrieval. As a first step in this direction, we introduce a novel adversarial fine-tuning method that iteratively searches for replication triggers and updates the model to increase robustness. Through our research, we provide fresh insights into the nature of memorization in text-to-image DMs and a foundation for building more trustworthy and compliant generative AI.
Authors: Youpeng Li, Weiliang Qi, Xuyu Wang, Fuxun Yu, Xinda Wang
Abstract: The rapid advancement of pre-trained language models (PLMs) has demonstrated promising results for various code-related tasks. However, their effectiveness in detecting real-world vulnerabilities remains a critical challenge. % for the security community. While existing empirical studies evaluate PLMs for vulnerability detection (VD), their inadequate consideration in data preparation, evaluation setups, and experimental settings undermines the accuracy and comprehensiveness of evaluations. This paper introduces RevisitVD, an extensive evaluation of 17 PLMs spanning smaller code-specific PLMs and large-scale PLMs using newly constructed datasets. Specifically, we compare the performance of PLMs under both fine-tuning and prompt engineering, assess their effectiveness and generalizability across various training and testing settings, and analyze their robustness against code normalization, abstraction, and semantic-preserving transformations. Our findings reveal that, for VD tasks, PLMs incorporating pre-training tasks designed to capture the syntactic and semantic patterns of code outperform both general-purpose PLMs and those solely pre-trained or fine-tuned on large code corpora. However, these models face notable challenges in real-world scenarios, such as difficulties in detecting vulnerabilities with complex dependencies, handling perturbations introduced by code normalization and abstraction, and identifying semantic-preserving vulnerable code transformations. Also, the truncation caused by the limited context windows of PLMs can lead to a non-negligible amount of labeling errors. This study underscores the importance of thorough evaluations of model performance in practical scenarios and outlines future directions to help enhance the effectiveness of PLMs for realistic VD applications.
Authors: April Herwig, Matthew J. Colbrook, Oliver Junge, P\'eter Koltai, Julia Slipantschuk
Abstract: Koopman operator theory enables linear analysis of nonlinear dynamical systems by lifting their evolution to infinite-dimensional function spaces. However, finite-dimensional approximations of Koopman and transfer (Frobenius--Perron) operators are prone to spectral pollution, introducing spurious eigenvalues that can compromise spectral computations. While recent advances have yielded provably convergent methods for Koopman operators, analogous tools for general transfer operators remain limited. In this paper, we present algorithms for computing spectral properties of transfer operators without spectral pollution, including extensions to the Hardy-Hilbert space. Case studies--ranging from families of Blaschke maps with known spectrum to a molecular dynamics model of protein folding--demonstrate the accuracy and flexibility of our approach. Notably, we demonstrate that spectral features can arise even when the corresponding eigenfunctions lie outside the chosen space, highlighting the functional-analytic subtleties in defining the "true" Koopman spectrum. Our methods offer robust tools for spectral estimation across a broad range of applications.
Authors: Nima Fathi, Amar Kumar, Tal Arbel
Abstract: Recent advancements in Large Language Models (LLMs) have catalyzed a paradigm shift from static prediction systems to agentic AI agents capable of reasoning, interacting with tools, and adapting to complex tasks. While LLM-based agentic systems have shown promise across many domains, their application to medical imaging remains in its infancy. In this work, we introduce AURA, the first visual linguistic explainability agent designed specifically for comprehensive analysis, explanation, and evaluation of medical images. By enabling dynamic interactions, contextual explanations, and hypothesis testing, AURA represents a significant advancement toward more transparent, adaptable, and clinically aligned AI systems. We highlight the promise of agentic AI in transforming medical image analysis from static predictions to interactive decision support. Leveraging Qwen-32B, an LLM-based architecture, AURA integrates a modular toolbox comprising: (i) a segmentation suite with phase grounding, pathology segmentation, and anatomy segmentation to localize clinically meaningful regions; (ii) a counterfactual image-generation module that supports reasoning through image-level explanations; and (iii) a set of evaluation tools including pixel-wise difference-map analysis, classification, and advanced state-of-the-art components to assess diagnostic relevance and visual interpretability.
Authors: Mohammad Reza Rahmani, Mohammad Hossein Yassaee, Mohammad Reza Aref
Abstract: Estimating high-dimensional covariance matrices is a key task across many fields. This paper explores the theoretical limits of distributed covariance estimation in a feature-split setting, where communication between agents is constrained. Specifically, we study a scenario in which multiple agents each observe different components of i.i.d. samples drawn from a sub-Gaussian random vector. A central server seeks to estimate the complete covariance matrix using a limited number of bits communicated by each agent. We obtain a nearly tight minimax lower bound for covariance matrix estimation under operator norm and Frobenius norm. Our main technical tool is a novel generalization of the strong data processing inequality (SDPI), termed the Conditional Strong Data Processing Inequality (C-SDPI) coefficient, introduced in this work. The C-SDPI coefficient shares key properties such as tensorization with the conventional SDPI. Crucially, it quantifies the average contraction in a state-dependent channel and can be significantly lower than the worst-case SDPI coefficient over the state input. Utilizing the doubling trick of Geng-Nair and an operator Jensen inequality, we compute this coefficient for Gaussian mixture channels. We then employ it to establish minimax lower bounds on estimation error, capturing the trade-offs among sample size, communication cost, and data dimensionality. Building on this, we present a nearly optimal estimation protocol whose sample and communication requirements match the lower bounds up to logarithmic factors. Unlike much of the existing literature, our framework does not assume infinite samples or Gaussian distributions, making it broadly applicable. Finally, we extend our analysis to interactive protocols, showing interaction can significantly reduce communication requirements compared to non-interactive schemes.
Authors: Yalda Zafari, Roaa Elalfy, Mohamed Mabrok, Somaya Al-Maadeed, Tamer Khattab, Essam A. Rashed
Abstract: Early and accurate interpretation of screening mammograms is essential for effective breast cancer detection, yet it remains a complex challenge due to subtle imaging findings and diagnostic ambiguity. Many existing AI approaches fall short by focusing on single view inputs or single-task outputs, limiting their clinical utility. To address these limitations, we propose a novel multi-view, multitask hybrid deep learning framework that processes all four standard mammography views and jointly predicts diagnostic labels and BI-RADS scores for each breast. Our architecture integrates a hybrid CNN VSSM backbone, combining convolutional encoders for rich local feature extraction with Visual State Space Models (VSSMs) to capture global contextual dependencies. To improve robustness and interpretability, we incorporate a gated attention-based fusion module that dynamically weights information across views, effectively handling cases with missing data. We conduct extensive experiments across diagnostic tasks of varying complexity, benchmarking our proposed hybrid models against baseline CNN architectures and VSSM models in both single task and multi task learning settings. Across all tasks, the hybrid models consistently outperform the baselines. In the binary BI-RADS 1 vs. 5 classification task, the shared hybrid model achieves an AUC of 0.9967 and an F1 score of 0.9830. For the more challenging ternary classification, it attains an F1 score of 0.7790, while in the five-class BI-RADS task, the best F1 score reaches 0.4904. These results highlight the effectiveness of the proposed hybrid framework and underscore both the potential and limitations of multitask learning for improving diagnostic performance and enabling clinically meaningful mammography analysis.
Authors: Mohammad Saleh Refahi, Gavin Hearne, Harrison Muller, Kieran Lynch, Bahrad A. Sokhansanj, James R. Brown, Gail Rosen
Abstract: The exponential growth of DNA sequencing data has outpaced traditional heuristic-based methods, which struggle to scale effectively. Efficient computational approaches are urgently needed to support large-scale similarity search, a foundational task in bioinformatics for detecting homology, functional similarity, and novelty among genomic and proteomic sequences. Although tools like BLAST have been widely used and remain effective in many scenarios, they suffer from limitations such as high computational cost and poor performance on divergent sequences. In this work, we explore embedding-based similarity search methods that learn latent representations capturing deeper structural and functional patterns beyond raw sequence alignment. We systematically evaluate two state-of-the-art vector search libraries, FAISS and ScaNN, on biologically meaningful gene embeddings. Unlike prior studies, our analysis focuses on bioinformatics-specific embeddings and benchmarks their utility for detecting novel sequences, including those from uncharacterized taxa or genes lacking known homologs. Our results highlight both computational advantages (in memory and runtime efficiency) and improved retrieval quality, offering a promising alternative to traditional alignment-heavy tools.
Authors: Felix Huber, Sebastian Rojas Gonzalez, Raul Astudillo
Abstract: We present a novel approach to help decision-makers efficiently identify preferred solutions from the Pareto set of a multi-objective optimization problem. Our method uses a Bayesian model to estimate the decision-maker's utility function based on pairwise comparisons. Aided by this model, a principled elicitation strategy selects queries interactively to balance exploration and exploitation, guiding the discovery of high-utility solutions. The approach is flexible: it can be used interactively or a posteriori after estimating the Pareto front through standard multi-objective optimization techniques. Additionally, at the end of the elicitation phase, it generates a reduced menu of high-quality solutions, simplifying the decision-making process. Through experiments on test problems with up to nine objectives, our method demonstrates superior performance in finding high-utility solutions with a small number of queries. We also provide an open-source implementation of our method to support its adoption by the broader community.
Authors: Jacob Piland, Chris Sweet, Adam Czajka
Abstract: Existing saliency-guided training approaches improve model generalization by incorporating a loss term that compares the model's class activation map (CAM) for a sample's true-class ({\it i.e.}, correct-label class) against a human reference saliency map. However, prior work has ignored the false-class CAM(s), that is the model's saliency obtained for incorrect-label class. We hypothesize that in binary tasks the true and false CAMs should diverge on the important classification features identified by humans (and reflected in human saliency maps). We use this hypothesis to motivate three new saliency-guided training methods incorporating both true- and false-class model's CAM into the training strategy and a novel post-hoc tool for identifying important features. We evaluate all introduced methods on several diverse binary close-set and open-set classification tasks, including synthetic face detection, biometric presentation attack detection, and classification of anomalies in chest X-ray scans, and find that the proposed methods improve generalization capabilities of deep learning models over traditional (true-class CAM only) saliency-guided training approaches. We offer source codes and model weights\footnote{GitHub repository link removed to preserve anonymity} to support reproducible research.
Authors: H M Mohaimanul Islam, Huynh Q. N. Vo, Aditya Rane
Abstract: In the era of synthetic media, deepfake manipulations pose a significant threat to information integrity. To address this challenge, we propose TrustDefender, a two-stage framework comprising (i) a lightweight convolutional neural network (CNN) that detects deepfake imagery in real-time extended reality (XR) streams, and (ii) an integrated succinct zero-knowledge proof (ZKP) protocol that validates detection results without disclosing raw user data. Our design addresses both the computational constraints of XR platforms while adhering to the stringent privacy requirements in sensitive settings. Experimental evaluations on multiple benchmark deepfake datasets demonstrate that TrustDefender achieves 95.3% detection accuracy, coupled with efficient proof generation underpinned by rigorous cryptography, ensuring seamless integration with high-performance artificial intelligence (AI) systems. By fusing advanced computer vision models with provable security mechanisms, our work establishes a foundation for reliable AI in immersive and privacy-sensitive applications.
Authors: Vansh Bansal, Tianyu Chen, James G. Scott
Abstract: Neural Posterior Estimation (NPE) has emerged as a powerful approach for amortized Bayesian inference when the true posterior $p(\theta \mid y)$ is intractable or difficult to sample. But evaluating the accuracy of neural posterior estimates remains challenging, with existing methods suffering from major limitations. One appealing and widely used method is the classifier two-sample test (C2ST), where a classifier is trained to distinguish samples from the true posterior $p(\theta \mid y)$ versus the learned NPE approximation $q(\theta \mid y)$. Yet despite the appealing simplicity of the C2ST, its theoretical and practical reliability depend upon having access to a near-Bayes-optimal classifier -- a requirement that is rarely met and, at best, difficult to verify. Thus a major open question is: can a weak classifier still be useful for neural posterior validation? We show that the answer is yes. Building on the work of Hu and Lei, we present several key results for a conformal variant of the C2ST, which converts any trained classifier's scores -- even those of weak or over-fitted models -- into exact finite-sample p-values. We establish two key theoretical properties of the conformal C2ST: (i) finite-sample Type-I error control, and (ii) non-trivial power that degrades gently in tandem with the error of the trained classifier. The upshot is that even weak, biased, or overfit classifiers can still yield powerful and reliable tests. Empirically, the Conformal C2ST outperforms classical discriminative tests across a wide range of benchmarks. These results reveal the under appreciated strength of weak classifiers for validating neural posterior estimates, establishing the conformal C2ST as a practical, theoretically grounded diagnostic for modern simulation-based inference.
Authors: Tianyu Chen, Vansh Bansal, James G. Scott
Abstract: We consider the problem of validating whether a neural posterior estimate \( q(\theta \mid x) \) is an accurate approximation to the true, unknown true posterior \( p(\theta \mid x) \). Existing methods for evaluating the quality of an NPE estimate are largely derived from classifier-based tests or divergence measures, but these suffer from several practical drawbacks. As an alternative, we introduce the \emph{Conditional Localization Test} (CoLT), a principled method designed to detect discrepancies between \( p(\theta \mid x) \) and \( q(\theta \mid x) \) across the full range of conditioning inputs. Rather than relying on exhaustive comparisons or density estimation at every \( x \), CoLT learns a localization function that adaptively selects points $\theta_l(x)$ where the neural posterior $q$ deviates most strongly from the true posterior $p$ for that $x$. This approach is particularly advantageous in typical simulation-based inference settings, where only a single draw \( \theta \sim p(\theta \mid x) \) from the true posterior is observed for each conditioning input, but where the neural posterior \( q(\theta \mid x) \) can be sampled an arbitrary number of times. Our theoretical results establish necessary and sufficient conditions for assessing distributional equality across all \( x \), offering both rigorous guarantees and practical scalability. Empirically, we demonstrate that CoLT not only performs better than existing methods at comparing $p$ and $q$, but also pinpoints regions of significant divergence, providing actionable insights for model refinement. These properties position CoLT as a state-of-the-art solution for validating neural posterior estimates.
Authors: Jannis B\"ahler, Diego Paez-Granados, Jorge Pe\~na-Queralta
Abstract: Smart electric wheelchairs can improve user experience by supporting the driver with shared control. State-of-the-art work showed the potential of shared control in improving safety in navigation for non-holonomic robots. However, for holonomic systems, current approaches often lead to unintuitive behavior for the user and fail to utilize the full potential of omnidirectional driving. Therefore, we propose a reinforcement learning-based method, which takes a 2D user input and outputs a 3D motion while ensuring user comfort and reducing cognitive load on the driver. Our approach is trained in Isaac Gym and tested in simulation in Gazebo. We compare different RL agent architectures and reward functions based on metrics considering cognitive load and user comfort. We show that our method ensures collision-free navigation while smartly orienting the wheelchair and showing better or competitive smoothness compared to a previous non-learning-based method. We further perform a sim-to-real transfer and demonstrate, to the best of our knowledge, the first real-world implementation of RL-based shared control for an omnidirectional mobility platform.
Authors: Jayadeep Jacob, Wenzheng Zhang, Houston Warren, Paulo Borges, Tirthankar Bandyopadhyay, Fabio Ramos
Abstract: Manipulating clusters of deformable objects presents a substantial challenge with widespread applicability, but requires contact-rich whole-arm interactions. A potential solution must address the limited capacity for realistic model synthesis, high uncertainty in perception, and the lack of efficient spatial abstractions, among others. We propose a novel framework for learning model-free policies integrating two modalities: 3D point clouds and proprioceptive touch indicators, emphasising manipulation with full body contact awareness, going beyond traditional end-effector modes. Our reinforcement learning framework leverages a distributional state representation, aided by kernel mean embeddings, to achieve improved training efficiency and real-time inference. Furthermore, we propose a novel context-agnostic occlusion heuristic to clear deformables from a target region for exposure tasks. We deploy the framework in a power line clearance scenario and observe that the agent generates creative strategies leveraging multiple arm links for de-occlusion. Finally, we perform zero-shot sim-to-real policy transfer, allowing the arm to clear real branches with unknown occlusion patterns, unseen topology, and uncertain dynamics.
Authors: Faisal Ahmed, Mohammad Alfrad Nobel Bhuiyan
Abstract: Diabetic retinopathy (DR) is a leading cause of vision loss worldwide, and early diagnosis through automated retinal image analysis can significantly reduce the risk of blindness. This paper presents a robust deep learning framework for both binary and five-class DR classification, leveraging transfer learning and extensive data augmentation to address the challenges of class imbalance and limited training data. We evaluate a range of pretrained convolutional neural network architectures, including variants of ResNet and EfficientNet, on the APTOS 2019 dataset. For binary classification, our proposed model achieves a state-of-the-art accuracy of 98.9%, with a precision of 98.6%, recall of 99.3%, F1-score of 98.9%, and an AUC of 99.4%. In the more challenging five-class severity classification task, our model obtains a competitive accuracy of 84.6% and an AUC of 94.1%, outperforming several existing approaches. Our findings also demonstrate that EfficientNet-B0 and ResNet34 offer optimal trade-offs between accuracy and computational efficiency across both tasks. These results underscore the effectiveness of combining class-balanced augmentation with transfer learning for high-performance DR diagnosis. The proposed framework provides a scalable and accurate solution for DR screening, with potential for deployment in real-world clinical environments.
Authors: Masayoshi Someya, Taisuke Yamada, Tomohisa Okazaki
Abstract: The Okada model is a widely used analytical solution for displacements and strains caused by a point or rectangular dislocation source in a 3D elastic half-space. We present OkadaTorch, a PyTorch implementation of the Okada model, where the entire code is differentiable; gradients with respect to input can be easily computed using automatic differentiation (AD). Our work consists of two components: a direct translation of the original Okada model into PyTorch, and a convenient wrapper interface for efficiently computing gradients and Hessians with respect to either observation station coordinates or fault parameters. This differentiable framework is well suited for fault parameter inversion, including gradient-based optimization, Bayesian inference, and integration with scientific machine learning (SciML) models. Our code is available here: https://github.com/msomeya1/OkadaTorch
Authors: Bo Fang, Jianan Fan, Dongnan Liu, Hang Chang, Gerald J. Shami, Filip Braet, Weidong Cai
Abstract: The significant morphological and distributional variability among subcellular components poses a long-standing challenge for learning-based organelle segmentation models, significantly increasing the risk of biased feature learning. Existing methods often rely on single mapping relationships, overlooking feature diversity and thereby inducing biased training. Although the Segment Anything Model (SAM) provides rich feature representations, its application to subcellular scenarios is hindered by two key challenges: (1) The variability in subcellular morphology and distribution creates gaps in the label space, leading the model to learn spurious or biased features. (2) SAM focuses on global contextual understanding and often ignores fine-grained spatial details, making it challenging to capture subtle structural alterations and cope with skewed data distributions. To address these challenges, we introduce ScSAM, a method that enhances feature robustness by fusing pre-trained SAM with Masked Autoencoder (MAE)-guided cellular prior knowledge to alleviate training bias from data imbalance. Specifically, we design a feature alignment and fusion module to align pre-trained embeddings to the same feature space and efficiently combine different representations. Moreover, we present a cosine similarity matrix-based class prompt encoder to activate class-specific features to recognize subcellular categories. Extensive experiments on diverse subcellular image datasets demonstrate that ScSAM outperforms state-of-the-art methods.
Authors: Fangze Lin, Ying He, Fei Yu, Hong Zhang
Abstract: Predicting the future motion of road participants is a critical task in autonomous driving. In this work, we address the challenge of low-quality generation of low-probability modes in multi-agent joint prediction. To tackle this issue, we propose a two-stage multi-agent interactive prediction framework named \textit{keypoint-guided joint prediction after classification-aware marginal proposal} (JAM). The first stage is modeled as a marginal prediction process, which classifies queries by trajectory type to encourage the model to learn all categories of trajectories, providing comprehensive mode information for the joint prediction module. The second stage is modeled as a joint prediction process, which takes the scene context and the marginal proposals from the first stage as inputs to learn the final joint distribution. We explicitly introduce key waypoints to guide the joint prediction module in better capturing and leveraging the critical information from the initial predicted trajectories. We conduct extensive experiments on the real-world Waymo Open Motion Dataset interactive prediction benchmark. The results show that our approach achieves competitive performance. In particular, in the framework comparison experiments, the proposed JAM outperforms other prediction frameworks and achieves state-of-the-art performance in interactive trajectory prediction. The code is available at https://github.com/LinFunster/JAM to facilitate future research.
Authors: Myeongwon Jung, Takanori Fujiwara, Jaemin Jo
Abstract: Despite the widespread use of Uniform Manifold Approximation and Projection (UMAP), the impact of its stochastic optimization process on the results remains underexplored. We observed that it often produces unstable results where the projections of data points are determined mostly by chance rather than reflecting neighboring structures. To address this limitation, we introduce (r,d)-stability to UMAP: a framework that analyzes the stochastic positioning of data points in the projection space. To assess how stochastic elements, specifically initial projection positions and negative sampling, impact UMAP results, we introduce "ghosts", or duplicates of data points representing potential positional variations due to stochasticity. We define a data point's projection as (r,d)-stable if its ghosts perturbed within a circle of radius r in the initial projection remain confined within a circle of radius d for their final positions. To efficiently compute the ghost projections, we develop an adaptive dropping scheme that reduces a runtime up to 60% compared to an unoptimized baseline while maintaining approximately 90% of unstable points. We also present a visualization tool that supports the interactive exploration of the (r,d)-stability of data points. Finally, we demonstrate the effectiveness of our framework by examining the stability of projections of real-world datasets and present usage guidelines for the effective use of our framework.
Authors: Tianyi Wang, Bingqian Dai, Kin Wong, Yaochen Li, Yang Cheng, Qingyuan Shu, Haoran He, Puyang Huang, Hanshen Huang, Kang L. Wang
Abstract: As artificial intelligence (AI) advances into diverse applications, ensuring reliability of AI models is increasingly critical. Conventional neural networks offer strong predictive capabilities but produce deterministic outputs without inherent uncertainty estimation, limiting their reliability in safety-critical domains. Probabilistic neural networks (PNNs), which introduce randomness, have emerged as a powerful approach for enabling intrinsic uncertainty quantification. However, traditional CMOS architectures are inherently designed for deterministic operation and actively suppress intrinsic randomness. This poses a fundamental challenge for implementing PNNs, as probabilistic processing introduces significant computational overhead. To address this challenge, we introduce a Magnetic Probabilistic Computing (MPC) platform-an energy-efficient, scalable hardware accelerator that leverages intrinsic magnetic stochasticity for uncertainty-aware computing. This physics-driven strategy utilizes spintronic systems based on magnetic domain walls (DWs) and their dynamics to establish a new paradigm of physical probabilistic computing for AI. The MPC platform integrates three key mechanisms: thermally induced DW stochasticity, voltage controlled magnetic anisotropy (VCMA), and tunneling magnetoresistance (TMR), enabling fully electrical and tunable probabilistic functionality at the device level. As a representative demonstration, we implement a Bayesian Neural Network (BNN) inference structure and validate its functionality on CIFAR-10 classification tasks. Compared to standard 28nm CMOS implementations, our approach achieves a seven orders of magnitude improvement in the overall figure of merit, with substantial gains in area efficiency, energy consumption, and speed. These results underscore the MPC platform's potential to enable reliable and trustworthy physical AI systems.
Authors: Haoran Jiang, Shaohan Shi, Yunjie Yao, Chang Jiang, Quan Li
Abstract: Modern scientific discovery faces growing challenges in integrating vast and heterogeneous knowledge critical to breakthroughs in biomedicine and drug development. Traditional hypothesis-driven research, though effective, is constrained by human cognitive limits, the complexity of biological systems, and the high cost of trial-and-error experimentation. Deep learning models, especially graph neural networks (GNNs), have accelerated prediction generation, but the sheer volume of outputs makes manual selection for validation unscalable. Large language models (LLMs) offer promise in filtering and hypothesis generation, yet suffer from hallucinations and lack grounding in structured knowledge, limiting their reliability. To address these issues, we propose HypoChainer, a collaborative visualization framework that integrates human expertise, LLM-driven reasoning, and knowledge graphs (KGs) to enhance hypothesis generation and validation. HypoChainer operates in three stages: First, exploration and contextualization -- experts use retrieval-augmented LLMs (RAGs) and dimensionality reduction to navigate large-scale GNN predictions, assisted by interactive explanations. Second, hypothesis chain formation -- experts iteratively examine KG relationships around predictions and semantically linked entities, refining hypotheses with LLM and KG suggestions. Third, validation prioritization -- refined hypotheses are filtered based on KG-supported evidence to identify high-priority candidates for experimentation, with visual analytics further strengthening weak links in reasoning. We demonstrate HypoChainer's effectiveness through case studies in two domains and expert interviews, highlighting its potential to support interpretable, scalable, and knowledge-grounded scientific discovery.
Authors: Fatemeh Hasanzadeh Fard, Sanaz Hasanzadeh Fard, Mehdi Jonoobi
Abstract: The wood processing industry, particularly in facilities such as sawmills and MDF production lines, requires accurate and efficient identification of species and thickness of the wood. Although traditional methods rely heavily on expert human labor, they are slow, inconsistent, and prone to error, especially when processing large volumes. This study focuses on practical and cost-effective machine learning frameworks that automate the estimation of timber log diameter using standard RGB images captured under real-world working conditions. We employ the YOLOv5 object detection algorithm, fine-tuned on a public dataset (TimberSeg 1.0), to detect individual timber logs and estimate thickness through bounding-box dimensions. Unlike previous methods that require expensive sensors or controlled environments, this model is trained on images taken in typical industrial sheds during timber delivery. Experimental results show that the model achieves a mean Average Precision (mAP@0.5) of 0.64, demonstrating reliable log detection even with modest computing resources. This lightweight, scalable solution holds promise for practical integration into existing workflows, including on-site inventory management and preliminary sorting, particularly in small and medium-sized operations.
Authors: Po-Yen Wu, Cheng-Yu Kuo, Yuki Kadokawa, Takamitsu Matsubara
Abstract: In inaccessible environments with uncertain task demands, robots often rely on general-purpose tools that lack predefined usage strategies. These tools are not tailored for particular operations, making their longevity highly sensitive to how they are used. This creates a fundamental challenge: how can a robot learn a tool-use policy that both completes the task and prolongs the tool's lifespan? In this work, we address this challenge by introducing a reinforcement learning (RL) framework that incorporates tool lifespan as a factor during policy optimization. Our framework leverages Finite Element Analysis (FEA) and Miner's Rule to estimate Remaining Useful Life (RUL) based on accumulated stress, and integrates the RUL into the RL reward to guide policy learning toward lifespan-guided behavior. To handle the fact that RUL can only be estimated after task execution, we introduce an Adaptive Reward Normalization (ARN) mechanism that dynamically adjusts reward scaling based on estimated RULs, ensuring stable learning signals. We validate our method across simulated and real-world tool use tasks, including Object-Moving and Door-Opening with multiple general-purpose tools. The learned policies consistently prolong tool lifespan (up to 8.01x in simulation) and transfer effectively to real-world settings, demonstrating the practical value of learning lifespan-guided tool use strategies.
Authors: Saiful Khan, Joyraj Chakraborty, Philip Beaucamp, Niraj Bhujel, Min Chen
Abstract: Nowadays, machine learning (ML) teams have multiple concurrent ML workflows for different applications. Each workflow typically involves many experiments, iterations, and collaborative activities and commonly takes months and sometimes years from initial data wrangling to model deployment. Organizationally, there is a large amount of intermediate data to be stored, processed, and maintained. \emph{Data virtualization} becomes a critical technology in an infrastructure to serve ML workflows. In this paper, we present the design and implementation of a data virtualization service, focusing on its service architecture and service operations. The infrastructure currently supports six ML applications, each with more than one ML workflow. The data virtualization service allows the number of applications and workflows to grow in the coming years.
Authors: Jianxin Bi, Kevin Yuchen Ma, Ce Hao, Mike Zheng Shou, Harold Soh
Abstract: Tactile feedback is generally recognized to be crucial for effective interaction with the physical world. However, state-of-the-art Vision-Language-Action (VLA) models lack the ability to interpret and use tactile signals, limiting their effectiveness in contact-rich tasks. Incorporating tactile feedback into these systems is challenging due to the absence of large multi-modal datasets. We present VLA-Touch, an approach that enhances generalist robot policies with tactile sensing \emph{without fine-tuning} the base VLA. Our method introduces two key innovations: (1) a pipeline that leverages a pretrained tactile-language model that provides semantic tactile feedback for high-level task planning, and (2) a diffusion-based controller that refines VLA-generated actions with tactile signals for contact-rich manipulation. Through real-world experiments, we demonstrate that our dual-level integration of tactile feedback improves task planning efficiency while enhancing execution precision. Code is open-sourced at \href{https://github.com/jxbi1010/VLA-Touch}{this URL}.
Authors: Tobias Morocutti, Jonathan Greif, Paul Primus, Florian Schmid, Gerhard Widmer
Abstract: Spatial semantic segmentation of sound scenes (S5) involves the accurate identification of active sound classes and the precise separation of their sources from complex acoustic mixtures. Conventional systems rely on a two-stage pipeline - audio tagging followed by label-conditioned source separation - but are often constrained by the absence of fine-grained temporal information critical for effective separation. In this work, we address this limitation by introducing a novel approach for S5 that enhances the synergy between the event detection and source separation stages. Our key contributions are threefold. First, we fine-tune a pre-trained Transformer to detect active sound classes. Second, we utilize a separate instance of this fine-tuned Transformer to perform sound event detection (SED), providing the separation module with detailed, time-varying guidance. Third, we implement an iterative refinement mechanism that progressively enhances separation quality by recursively reusing the separator's output from previous iterations. These advancements lead to significant improvements in both audio tagging and source separation performance, as demonstrated by our system's second-place finish in Task 4 of the DCASE Challenge 2025. Our implementation and model checkpoints are available in our GitHub repository: https://github.com/theMoro/dcase25task4 .
Authors: Dirk van der Hoeven, Julia Olkhovskaia, Tim van Erven
Abstract: We consider the problem of estimating a discrete distribution $p$ with support of size $K$ and provide both upper and lower bounds with high probability in KL divergence. We prove that in the worst case, for any estimator $\widehat{p}$, with probability at least $\delta$, $\text{KL}(p \| \widehat{p}) \geq C\max\{K,\ln(K)\ln(1/\delta) \}/n $, where $n$ is the sample size and $C > 0$ is a constant. We introduce a computationally efficient estimator $p^{\text{OTB}}$, based on Online to Batch conversion and suffix averaging, and show that with probability at least $1 - \delta$ $\text{KL}(p \| \widehat{p}) \leq C(K\log(\log(K)) + \ln(K)\ln(1/\delta)) /n$. Furthermore, we also show that with sufficiently many observations relative to $\log(1/\delta)$, the maximum likelihood estimator $\bar{p}$ guarantees that with probability at least $1-\delta$ $$ 1/6 \chi^2(\bar{p}\|p) \leq 1/4 \chi^2(p\|\bar{p}) \leq \text{KL}(p|\bar{p}) \leq C(K + \log(1/\delta))/n\,, $$ where $\chi^2$ denotes the $\chi^2$-divergence.
Authors: Xiaohao Liu, Xiaobo Xia, See-Kiong Ng, Tat-Seng Chua
Abstract: Multimodal representation learning seeks to create a unified representation space by integrating diverse data modalities to improve multimodal understanding. Traditional methods often depend on pairwise contrastive learning, which relies on a predefined anchor modality, restricting alignment across all modalities. Recent advances have investigated the simultaneous alignment of multiple modalities, yet several challenges remain, such as limitations imposed by fixed anchor points and instability arising from optimizing the product of singular values. To address the challenges, in this paper, we propose Principled Multimodal Representation Learning (PMRL), a novel framework that achieves simultaneous alignment of multiple modalities without anchor dependency in a more stable manner. Specifically, grounded in the theoretical insight that full alignment corresponds to a rank-1 Gram matrix, PMRL optimizes the dominant singular value of the representation matrix to align modalities along a shared leading direction. We propose a softmax-based loss function that treats singular values as logits to prioritize the largest singular value. Besides, instance-wise contrastive regularization on the leading eigenvectors maintains inter-instance separability and prevents representation collapse. Extensive experiments across diverse tasks demonstrate PMRL's superiority compared to baseline methods. The source code will be publicly available.
Authors: Thomas P Zollo, Richard Zemel
Abstract: Trustworthy robot behavior requires not only high levels of task success but also that the robot can reliably quantify how likely it is to succeed. To this end, we present the first systematic study of confidence calibration in vision-language-action (VLA) foundation models, which map visual observations and natural-language instructions to low-level robot motor commands. We begin with extensive benchmarking to understand the critical relationship between task success and calibration error across multiple datasets and VLA variants, finding that task performance and calibration are not in tension. Next, we introduce prompt ensembles for VLAs, a lightweight, Bayesian-inspired algorithm that averages confidence across paraphrased instructions and consistently improves calibration. We further analyze calibration over the task time horizon, showing that confidence is often most reliable after making some progress, suggesting natural points for risk-aware intervention. Finally, we reveal differential miscalibration across action dimensions and propose action-wise Platt scaling, a method to recalibrate each action dimension independently to produce better confidence estimates. Our aim in this study is to begin to develop the tools and conceptual understanding necessary to render VLAs both highly performant and highly trustworthy via reliable uncertainty quantification.
Authors: Junlang Huang, Hao Chen, Zhong Guan
Abstract: This paper proposes a neural framework for power and timing prediction of multi-stage data path, distinguishing itself from traditional lib-based analytical methods dependent on driver characterization and load simplifications. To the best of our knowledge, this is the first language-based, netlist-aware neural network designed explicitly for standard cells. Our approach employs two pre-trained neural models of waveform prediction and delay estimation that directly infer transient waveforms and propagation delays from SPICE netlists, conditioned on critical physical parameters such as load capacitance, input slew, and gate size. This method accurately captures both intrinsic and coupling-induced delay effects without requiring simplification or interpolation. For multi-stage timing prediction, we implement a recursive propagation strategy where predicted waveforms from each stage feed into subsequent stages, cumulatively capturing delays across the logic chain. This approach ensures precise timing alignment and complete waveform visibility throughout complex signal pathways. The waveform prediction utilizes a hybrid CNN-Transformer architecture with netlist-aware node-level encoding, addressing traditional Transformers' fixed input dimensionality constraints. Additionally, specialized subnetworks separately handle primary delay estimation and crosstalk correction. Experimental results demonstrate SPICE-level accuracy, consistently achieving RMSE below 0.0098 across diverse industrial circuits. The proposed framework provides a scalable, structurally adaptable neural alternative to conventional power and timing engines, demonstrating high fidelity to physical circuit behaviors.
Authors: Joobin Jin, Seokjun Hong, Gyeongseon Baek, Yeeun Kim, Byeongjoon Noh
Abstract: Precise modeling of microscopic vehicle trajectories is critical for traffic behavior analysis and autonomous driving systems. We propose Ctx2TrajGen, a context-aware trajectory generation framework that synthesizes realistic urban driving behaviors using GAIL. Leveraging PPO and WGAN-GP, our model addresses nonlinear interdependencies and training instability inherent in microscopic settings. By explicitly conditioning on surrounding vehicles and road geometry, Ctx2TrajGen generates interaction-aware trajectories aligned with real-world context. Experiments on the drone-captured DRIFT dataset demonstrate superior performance over existing methods in terms of realism, behavioral diversity, and contextual fidelity, offering a robust solution to data scarcity and domain shift without simulation.
Authors: Joonsung Kang
Abstract: Outliers can severely distort causal effect estimation in observational studies, yet this issue has received limited attention in the literature. Their influence is especially pronounced in small sample sizes, where detecting and removing outliers becomes increasingly difficult. Therefore, it is essential to estimate treatment effects robustly without excluding these influential data points. To address this, we propose a doubly robust point estimator for the average treatment effect under a contaminated model that includes outliers. Robustness in outcome regression is achieved through a robust estimating equation, while covariate balancing propensity scores (CBPS) ensure resilience in propensity score modeling. To prevent model overfitting due to the inclusion of numerous parameters, we incorporate variable selection. All these components are unified under a penalized empirical likelihood framework. For confidence interval estimation, most existing approaches rely on asymptotic properties, which may be unreliable in finite samples. We derive an optimal finite-sample confidence interval for the average treatment effect using our proposed estimating equation, ensuring that the interval bounds remain unaffected by outliers. Through simulations and a real-world application involving hypertension data with outliers, we demonstrate that our method consistently outperforms existing approaches in both accuracy and robustness.
Authors: Wei-You Liao, Yuxuan Du, Xinbiao Wang, Tian-Ci Tian, Yong Luo, Bo Du, Dacheng Tao, He-Liang Huang
Abstract: The ongoing development of quantum processors is driving breakthroughs in scientific discovery. Despite this progress, the formidable cost of fabricating large-scale quantum processors means they will remain rare for the foreseeable future, limiting their widespread application. To address this bottleneck, we introduce the concept of predictive surrogates, which are classical learning models designed to emulate the mean-value behavior of a given quantum processor with provably computational efficiency. In particular, we propose two predictive surrogates that can substantially reduce the need for quantum processor access in diverse practical scenarios. To demonstrate their potential in advancing digital quantum simulation, we use these surrogates to emulate a quantum processor with up to 20 programmable superconducting qubits, enabling efficient pre-training of variational quantum eigensolvers for families of transverse-field Ising models and identification of non-equilibrium Floquet symmetry-protected topological phases. Experimental results reveal that the predictive surrogates not only reduce measurement overhead by orders of magnitude, but can also surpass the performance of conventional, quantum-resource-intensive approaches. Collectively, these findings establish predictive surrogates as a practical pathway to broadening the impact of advanced quantum processors.
Authors: Chuang Chen, Xiaolin Qin, Jing Hu, Wenyi Ge
Abstract: Upsampling LiDAR point clouds in autonomous driving scenarios remains a significant challenge due to the inherent sparsity and complex 3D structures of the data. Recent studies have attempted to address this problem by converting the complex 3D spatial scenes into 2D image super-resolution tasks. However, due to the sparse and blurry feature representation of range images, accurately reconstructing detailed and complex spatial topologies remains a major difficulty. To tackle this, we propose a novel sparse point cloud upsampling method named SRMambaV2, which enhances the upsampling accuracy in long-range sparse regions while preserving the overall geometric reconstruction quality. Specifically, inspired by human driver visual perception, we design a biomimetic 2D selective scanning self-attention (2DSSA) mechanism to model the feature distribution in distant sparse areas. Meanwhile, we introduce a dual-branch network architecture to enhance the representation of sparse features. In addition, we introduce a progressive adaptive loss (PAL) function to further refine the reconstruction of fine-grained details during the upsampling process. Experimental results demonstrate that SRMambaV2 achieves superior performance in both qualitative and quantitative evaluations, highlighting its effectiveness and practical value in automotive sparse point cloud upsampling tasks.
Authors: Rashika Raina, Nidhi Simmons, David E. Simmons, Michel Daoud Yacoub, Trung Q. Duong
Abstract: In next-generation communications and networks, machine learning (ML) models are expected to deliver not only accurate predictions but also well-calibrated confidence scores that reflect the true likelihood of correct decisions. This paper studies the calibration performance of an ML-based outage predictor within a single-user, multi-resource allocation framework. We first establish key theoretical properties of this system's outage probability (OP) under perfect calibration. Importantly, we show that as the number of resources grows, the OP of a perfectly calibrated predictor approaches the expected output conditioned on it being below the classification threshold. In contrast, when only one resource is available, the system's OP equals the model's overall expected output. We then derive the OP conditions for a perfectly calibrated predictor. These findings guide the choice of the classification threshold to achieve a desired OP, helping system designers meet specific reliability requirements. We also demonstrate that post-processing calibration cannot improve the system's minimum achievable OP, as it does not introduce new information about future channel states. Additionally, we show that well-calibrated models are part of a broader class of predictors that necessarily improve OP. In particular, we establish a monotonicity condition that the accuracy-confidence function must satisfy for such improvement to occur. To demonstrate these theoretical properties, we conduct a rigorous simulation-based analysis using post-processing calibration techniques: Platt scaling and isotonic regression. As part of this framework, the predictor is trained using an outage loss function specifically designed for this system. Furthermore, this analysis is performed on Rayleigh fading channels with temporal correlation captured by Clarke's 2D model, which accounts for receiver mobility.
Authors: Imad Bouhou, Stefano Fortunati, Leila Gharsalli, Alexandre Renaux
Abstract: This correspondence presents a power-aware cognitive radar framework for joint detection and tracking of multiple targets in a massive multiple-input multiple-output (MIMO) radar environment. Building on a previous single-target algorithm based on Partially Observable Monte Carlo Planning (POMCP), we extend it to the multi-target case by assigning each target an independent POMCP tree, enabling scalable and efficient planning. Departing from uniform power allocation-which is often suboptimal with varying signal-to-noise ratios (SNRs)-our approach predicts each target's future angular position and expected received power, based on its estimated range and radar cross-section (RCS). These predictions guide adaptive waveform design via a constrained optimization problem that allocates transmit energy to enhance the detectability of weaker or distant targets, while ensuring sufficient power for high-SNR targets. The reward function in the underlying partially observable Markov decision process (POMDP) is also modified to prioritize accurate spatial and power estimation. Simulations involving multiple targets with different SNRs confirm the effectiveness of our method. The proposed framework for the cognitive radar improves detection probability for low-SNR targets and achieves more accurate tracking compared to approaches using uniform or orthogonal waveforms. These results demonstrate the potential of the POMCP-based framework for adaptive, efficient multi-target radar systems.
Authors: V. Slavin, O. Kryvchikov, D. Laptev
Abstract: We present a graph-based deep learning framework for predicting the magnetic properties of quasi-one-dimensional Ising spin systems. The lattice geometry is encoded as a graph and processed by a graph neural network (GNN) followed by fully connected layers. The model is trained on Monte Carlo simulation data and accurately reproduces key features of the magnetization curve, including plateaus, critical transition points, and the effects of geometric frustration. It captures both local motifs and global symmetries, demonstrating that GNNs can infer magnetic behavior directly from structural connectivity. The proposed approach enables efficient prediction of magnetization without the need for additional Monte Carlo simulations.
Authors: Yu Li, Zhuoshi Pan, Honglin Lin, Mengyuan Sun, Conghui He, Lijun Wu
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a powerful paradigm for enhancing the reasoning capabilities of LLMs. Existing research has predominantly concentrated on isolated reasoning domains such as mathematical problem-solving, coding tasks, or logical reasoning. However, real world reasoning scenarios inherently demand an integrated application of multiple cognitive skills. Despite this, the interplay among these reasoning skills under reinforcement learning remains poorly understood. To bridge this gap, we present a systematic investigation of multi-domain reasoning within the RLVR framework, explicitly focusing on three primary domains: mathematical reasoning, code generation, and logical puzzle solving. We conduct a comprehensive study comprising four key components: (1) Leveraging the GRPO algorithm and the Qwen-2.5-7B model family, our study thoroughly evaluates the models' in-domain improvements and cross-domain generalization capabilities when trained on single-domain datasets. (2) Additionally, we examine the intricate interactions including mutual enhancements and conflicts that emerge during combined cross-domain training. (3) To further understand the influence of SFT on RL, we also analyze and compare performance differences between base and instruct models under identical RL configurations. (4) Furthermore, we delve into critical RL training details, systematically exploring the impacts of curriculum learning strategies, variations in reward design, and language-specific factors. Through extensive experiments, our results offer significant insights into the dynamics governing domain interactions, revealing key factors influencing both specialized and generalizable reasoning performance. These findings provide valuable guidance for optimizing RL methodologies to foster comprehensive, multi-domain reasoning capabilities in LLMs.
Authors: Leandro Von Krannichfeldt, Kristina Orehounig, Olga Fink
Abstract: Building energy modeling is a key tool for optimizing the performance of building energy systems. Historically, a wide spectrum of methods has been explored -- ranging from conventional physics-based models to purely data-driven techniques. Recently, hybrid approaches that combine the strengths of both paradigms have gained attention. These include strategies such as learning surrogates for physics-based models, modeling residuals between simulated and observed data, fine-tuning surrogates with real-world measurements, using physics-based outputs as additional inputs for data-driven models, and integrating the physics-based output into the loss function the data-driven model. Despite this progress, two significant research gaps remain. First, most hybrid methods focus on deterministic modeling, often neglecting the inherent uncertainties caused by factors like weather fluctuations and occupant behavior. Second, there has been little systematic comparison within a probabilistic modeling framework. This study addresses these gaps by evaluating five representative hybrid approaches for probabilistic building energy modeling, focusing on quantile predictions of building thermodynamics in a real-world case study. Our results highlight two main findings. First, the performance of hybrid approaches varies across different building room types, but residual learning with a Feedforward Neural Network performs best on average. Notably, the residual approach is the only model that produces physically intuitive predictions when applied to out-of-distribution test data. Second, Quantile Conformal Prediction is an effective procedure for calibrating quantile predictions in case of indoor temperature modeling.
Authors: Piotr Masztalski, Micha{\l} Romaniuk, Jakub \.Zak, Mateusz Matuszewski, Konrad Kowalczyk
Abstract: In speaker verification, contrastive learning is gaining popularity as an alternative to the traditionally used classification-based approaches. Contrastive methods can benefit from an effective use of hard negative pairs, which are different-class samples particularly challenging for a verification model due to their similarity. In this paper, we propose CHNS - a clustering-based hard negative sampling method, dedicated for supervised contrastive speaker representation learning. Our approach clusters embeddings of similar speakers, and adjusts batch composition to obtain an optimal ratio of hard and easy negatives during contrastive loss calculation. Experimental evaluation shows that CHNS outperforms a baseline supervised contrastive approach with and without loss-based hard negative sampling, as well as a state-of-the-art classification-based approach to speaker verification by as much as 18 % relative EER and minDCF on the VoxCeleb dataset using two lightweight model architectures.
Authors: Bonwoo Lee, Cheolwoo Park, Jeongyoun Ahn
Abstract: Differential privacy has become a cornerstone in the development of privacy-preserving learning algorithms. This work addresses optimizing differentially private kernel learning within the empirical risk minimization (ERM) framework. We propose a novel differentially private kernel ERM algorithm based on random projection in the reproducing kernel Hilbert space using Gaussian processes. Our method achieves minimax-optimal excess risk for both the squared loss and Lipschitz-smooth convex loss functions under a local strong convexity condition. We further show that existing approaches based on alternative dimension reduction techniques, such as random Fourier feature mappings or $\ell_2$ regularization, yield suboptimal generalization performance. Our key theoretical contribution also includes the derivation of dimension-free generalization bounds for objective perturbation-based private linear ERM -- marking the first such result that does not rely on noisy gradient-based mechanisms. Additionally, we obtain sharper generalization bounds for existing differentially private kernel ERM algorithms. Empirical evaluations support our theoretical claims, demonstrating that random projection enables statistically efficient and optimally private kernel learning. These findings provide new insights into the design of differentially private algorithms and highlight the central role of dimension reduction in balancing privacy and utility.
Authors: Sebastian Pokutta
Abstract: We consider the problem of minimizing a difference of (smooth) convex functions over a compact convex feasible region $P$, i.e., $\min_{x \in P} f(x) - g(x)$, with smooth $f$ and Lipschitz continuous $g$. This computational study builds upon and complements the framework of Maskan et al. [2025] by integrating advanced Frank-Wolfe variants to reduce computational overhead. We empirically show that constrained DC problems can be efficiently solved using a combination of the Blended Pairwise Conditional Gradients (BPCG) algorithm [Tsuji et al., 2022] with warm-starting and the adaptive error bound from Maskan et al. [2025]. The result is a highly efficient and scalable projection-free algorithm for constrained DC optimization.
Authors: Chen Ma, Xinjie Xu, Shuyu Cheng, Qi Xuan
Abstract: One of the most practical and challenging types of black-box adversarial attacks is the hard-label attack, where only the top-1 predicted label is available. One effective approach is to search for the optimal ray direction from the benign image that minimizes the $\ell_p$-norm distance to the adversarial region. The unique advantage of this approach is that it transforms the hard-label attack into a continuous optimization problem. The objective function value is the ray's radius, which can be obtained via binary search at a high query cost. Existing methods use a "sign trick" in gradient estimation to reduce the number of queries. In this paper, we theoretically analyze the quality of this gradient estimation and propose a novel prior-guided approach to improve ray search efficiency both theoretically and empirically. Specifically, we utilize the transfer-based priors from surrogate models, and our gradient estimators appropriately integrate them by approximating the projection of the true gradient onto the subspace spanned by these priors and random directions, in a query-efficient manner. We theoretically derive the expected cosine similarities between the obtained gradient estimators and the true gradient, and demonstrate the improvement achieved by incorporating priors. Extensive experiments on the ImageNet and CIFAR-10 datasets show that our approach significantly outperforms 11 state-of-the-art methods in terms of query efficiency.
Authors: Maciej K. Wozniak, Lianhang Liu, Yixi Cai, Patric Jensfelt
Abstract: While end-to-end autonomous driving models show promising results, their practical deployment is often hindered by large model sizes, a reliance on expensive LiDAR sensors and computationally intensive BEV feature representations. This limits their scalability, especially for mass-market vehicles equipped only with cameras. To address these challenges, we propose PRIX (Plan from Raw Pixels). Our novel and efficient end-to-end driving architecture operates using only camera data, without explicit BEV representation and forgoing the need for LiDAR. PRIX leverages a visual feature extractor coupled with a generative planning head to predict safe trajectories from raw pixel inputs directly. A core component of our architecture is the Context-aware Recalibration Transformer (CaRT), a novel module designed to effectively enhance multi-level visual features for more robust planning. We demonstrate through comprehensive experiments that PRIX achieves state-of-the-art performance on the NavSim and nuScenes benchmarks, matching the capabilities of larger, multimodal diffusion planners while being significantly more efficient in terms of inference speed and model size, making it a practical solution for real-world deployment. Our work is open-source and the code will be at https://maxiuw.github.io/prix.
Authors: Conor McNamara, Effirul Ramlan
Abstract: Recent advances in citation recommendation have improved accuracy by leveraging multi-view representation learning to integrate the various modalities present in scholarly documents. However, effectively combining multiple data views requires fusion techniques that can capture complementary information while preserving the unique characteristics of each modality. We propose a novel citation recommendation algorithm that improves upon linear Canonical Correlation Analysis (CCA) methods by applying Deep CCA (DCCA), a neural network extension capable of capturing complex, non-linear relationships between distributed textual and graph-based representations of scientific articles. Experiments on the large-scale DBLP (Digital Bibliography & Library Project) citation network dataset demonstrate that our approach outperforms state-of-the-art CCA-based methods, achieving relative improvements of over 11% in Mean Average Precision@10, 5% in Precision@10, and 7% in Recall@10. These gains reflect more relevant citation recommendations and enhanced ranking quality, suggesting that DCCA's non-linear transformations yield more expressive latent representations than CCA's linear projections.
Authors: Jasper Rou
Abstract: In this research, we explore neural network-based methods for pricing multidimensional American put options under the BlackScholes and Heston model, extending up to five dimensions. We focus on two approaches: the Time Deep Gradient Flow (TDGF) method and the Deep Galerkin Method (DGM). We extend the TDGF method to handle the free-boundary partial differential equation inherent in American options. We carefully design the sampling strategy during training to enhance performance. Both TDGF and DGM achieve high accuracy while outperforming conventional Monte Carlo methods in terms of computational speed. In particular, TDGF tends to be faster during training than DGM.
Authors: Miguel Carrasco, C\'esar Gonz\'alez-Mart\'in, Jos\'e Aranda, Luis Oliveros
Abstract: Visual attention mechanisms play a crucial role in human perception and aesthetic evaluation. Recent advances in Vision Transformers (ViTs) have demonstrated remarkable capabilities in computer vision tasks, yet their alignment with human visual attention patterns remains underexplored, particularly in aesthetic contexts. This study investigates the correlation between human visual attention and ViT attention mechanisms when evaluating handcrafted objects. We conducted an eye-tracking experiment with 30 participants (9 female, 21 male, mean age 24.6 years) who viewed 20 artisanal objects comprising basketry bags and ginger jars. Using a Pupil Labs eye-tracker, we recorded gaze patterns and generated heat maps representing human visual attention. Simultaneously, we analyzed the same objects using a pre-trained ViT model with DINO (Self-DIstillation with NO Labels), extracting attention maps from each of the 12 attention heads. We compared human and ViT attention distributions using Kullback-Leibler divergence across varying Gaussian parameters (sigma=0.1 to 3.0). Statistical analysis revealed optimal correlation at sigma=2.4 +-0.03, with attention head #12 showing the strongest alignment with human visual patterns. Significant differences were found between attention heads, with heads #7 and #9 demonstrating the greatest divergence from human attention (p< 0.05, Tukey HSD test). Results indicate that while ViTs exhibit more global attention patterns compared to human focal attention, certain attention heads can approximate human visual behavior, particularly for specific object features like buckles in basketry items. These findings suggest potential applications of ViT attention mechanisms in product design and aesthetic evaluation, while highlighting fundamental differences in attention strategies between human perception and current AI models.
Authors: Changxin Tian, Jiapeng Wang, Qian Zhao, Kunlong Chen, Jia Liu, Ziqi Liu, Jiaxin Mao, Wayne Xin Zhao, Zhiqiang Zhang, Jun Zhou
Abstract: Recent advances in learning rate (LR) scheduling have demonstrated the effectiveness of decay-free approaches that eliminate the traditional decay phase while maintaining competitive performance. Model merging techniques have emerged as particularly promising solutions in this domain. We present Warmup-Stable and Merge (WSM), a general framework that establishes a formal connection between learning rate decay and model merging. WSM provides a unified theoretical foundation for emulating various decay strategies-including cosine decay, linear decay and inverse square root decay-as principled model averaging schemes, while remaining fully compatible with diverse optimization methods. Through extensive experiments, we identify merge duration-the training window for checkpoint aggregation-as the most critical factor influencing model performance, surpassing the importance of both checkpoint interval and merge quantity. Our framework consistently outperforms the widely-adopted Warmup-Stable-Decay (WSD) approach across multiple benchmarks, achieving significant improvements of +3.5% on MATH, +2.9% on HumanEval, and +5.5% on MMLU-Pro. The performance advantages extend to supervised fine-tuning scenarios, highlighting WSM's potential for long-term model refinement.
Authors: Farnoush Bayatmakou, Reza Taleei, Nicole Simone, Arash Mohammadi
Abstract: Breast cancer (BC) remains one of the leading causes of cancer-related mortality among women, despite recent advances in Computer-Aided Diagnosis (CAD) systems. Accurate and efficient interpretation of multi-view mammograms is essential for early detection, driving a surge of interest in Artificial Intelligence (AI)-powered CAD models. While state-of-the-art multi-view mammogram classification models are largely based on Transformer architectures, their computational complexity scales quadratically with the number of image patches, highlighting the need for more efficient alternatives. To address this challenge, we propose Mammo-Mamba, a novel framework that integrates Selective State-Space Models (SSMs), transformer-based attention, and expert-driven feature refinement into a unified architecture. Mammo-Mamba extends the MambaVision backbone by introducing the Sequential Mixture of Experts (SeqMoE) mechanism through its customized SecMamba block. The SecMamba is a modified MambaVision block that enhances representation learning in high-resolution mammographic images by enabling content-adaptive feature refinement. These blocks are integrated into the deeper stages of MambaVision, allowing the model to progressively adjust feature emphasis through dynamic expert gating, effectively mitigating the limitations of traditional Transformer models. Evaluated on the CBIS-DDSM benchmark dataset, Mammo-Mamba achieves superior classification performance across all key metrics while maintaining computational efficiency.
Authors: Takashi Hayakawa, Satoshi Asai
Abstract: Previous studies have shown that hazard ratios between treatment groups estimated with the Cox model are uninterpretable because the indefinite baseline hazard of the model fails to identify temporal change in the risk set composition due to treatment assignment and unobserved factors among multiple, contradictory scenarios. To alleviate this problem, especially in studies based on observational data with uncontrolled dynamic treatment and real-time measurement of many covariates, we propose abandoning the baseline hazard and using machine learning to explicitly model the change in the risk set with or without latent variables. For this framework, we clarify the context in which hazard ratios can be causally interpreted, and then develop a method based on Neyman orthogonality to compute debiased maximum-likelihood estimators of hazard ratios. Computing the constructed estimators is more efficient than computing those based on weighted regression with marginal structural Cox models. Numerical simulations confirm that the proposed method identifies the ground truth with minimal bias. These results lay the foundation for developing a useful, alternative method for causal inference with uncontrolled, observational data in modern epidemiology.
Authors: Mohammad Nur Hossain Khan, David creswell, Jordan Albert, Patrick O'Connell, Shawn Fallon, Mathew Polowitz, Xuhai "orson" Xu, Bashima islam
Abstract: Mindfulness training is widely recognized for its benefits in reducing depression, anxiety, and loneliness. With the rise of smartphone-based mindfulness apps, digital meditation has become more accessible, but sustaining long-term user engagement remains a challenge. This paper explores whether respiration biosignal feedback and mindfulness skill estimation enhance system usability and skill development. We develop a smartphone's accelerometer-based respiration tracking algorithm, eliminating the need for additional wearables. Unlike existing methods, our approach accurately captures slow breathing patterns typical of mindfulness meditation. Additionally, we introduce the first quantitative framework to estimate mindfulness skills-concentration, sensory clarity, and equanimity-based on accelerometer-derived respiration data. We develop and test our algorithms on 261 mindfulness sessions in both controlled and real-world settings. A user study comparing an experimental group receiving biosignal feedback with a control group using a standard app shows that respiration feedback enhances system usability. Our respiration tracking model achieves a mean absolute error (MAE) of 1.6 breaths per minute, closely aligning with ground truth data, while our mindfulness skill estimation attains F1 scores of 80-84% in tracking skill progression. By integrating respiration tracking and mindfulness estimation into a commercial app, we demonstrate the potential of smartphone sensors to enhance digital mindfulness training.
Authors: Shan Jiang, Pranoy Kovuri, David Tao, Zhixun Tan
Abstract: Software obfuscation, particularly prevalent in JavaScript, hinders code comprehension and analysis, posing significant challenges to software testing, static analysis, and malware detection. This paper introduces CASCADE, a novel hybrid approach that integrates the advanced coding capabilities of Gemini with the deterministic transformation capabilities of a compiler Intermediate Representation (IR), specifically JavaScript IR (JSIR). By employing Gemini to identify critical prelude functions, the foundational components underlying the most prevalent obfuscation techniques, and leveraging JSIR for subsequent code transformations, CASCADE effectively recovers semantic elements like original strings and API names, and reveals original program behaviors. This method overcomes limitations of existing static and dynamic deobfuscation techniques, eliminating hundreds to thousands of hardcoded rules while achieving reliability and flexibility. CASCADE is already deployed in Google's production environment, demonstrating substantial improvements in JavaScript deobfuscation efficiency and reducing reverse engineering efforts.
Authors: Hongji Wang, Hongqiao Wang, Jinyong Ying, Qingping Zhou
Abstract: Inverse problems governed by partial differential equations (PDEs) play a crucial role in various fields, including computational science, image processing, and engineering. Particularly, Darcy flow equation is a fundamental equation in fluid mechanics, which plays a crucial role in understanding fluid flow through porous media. Bayesian methods provide an effective approach for solving PDEs inverse problems, while their numerical implementation requires numerous evaluations of computationally expensive forward solvers. Therefore, the adoption of surrogate models with lower computational costs is essential. However, constructing a globally accurate surrogate model for high-dimensional complex problems demands high model capacity and large amounts of data. To address this challenge, this study proposes an efficient locally accurate surrogate that focuses on the high-probability regions of the true likelihood in inverse problems, with relatively low model complexity and few training data requirements. Additionally, we introduce a sequential Bayesian design strategy to acquire the proposed surrogate since the high-probability region of the likelihood is unknown. The strategy treats the posterior evolution process of sequential Bayesian design as a Gaussian process, enabling algorithmic acceleration through one-step ahead prior. The complete algorithmic framework is referred to as Sequential Bayesian design for locally accurate surrogate (SBD-LAS). Finally, three experiments based the Darcy flow equation demonstrate the advantages of the proposed method in terms of both inversion accuracy and computational speed.
Authors: Arnab Neogi, Suryakant Mishra, Prasad P Iyer, Tzu-Ming Lu, Ezra Bussmann, Sergei Tretiak, Andrew Crandall Jones, Jian-Xin Zhu
Abstract: Increasingly large datasets of microscopic images with atomic resolution facilitate the development of machine learning methods to identify and analyze subtle physical phenomena embedded within the images. In this work, microscopic images of honeycomb lattice spin-ice samples serve as datasets from which we automate the calculation of net magnetic moments and directional orientations of spin-ice configurations. In the first stage of our workflow, machine learning models are trained to accurately predict magnetic moments and directions within spin-ice structures. Variational Autoencoders (VAEs), an emergent unsupervised deep learning technique, are employed to generate high-quality synthetic magnetic force microscopy (MFM) images and extract latent feature representations, thereby reducing experimental and segmentation errors. The second stage of proposed methodology enables precise identification and prediction of frustrated vertices and nanomagnetic segments, effectively correlating structural and functional aspects of microscopic images. This facilitates the design of optimized spin-ice configurations with controlled frustration patterns, enabling potential on-demand synthesis.
Authors: Yuelin Wang, Kai Yi, Xinliang Liu, Yu Guang Wang, Shi Jin
Abstract: Neural message passing is a basic feature extraction unit for graph-structured data considering neighboring node features in network propagation from one layer to the next. We model such process by an interacting particle system with attractive and repulsive forces and the Allen-Cahn force arising in the modeling of phase transition. The dynamics of the system is a reaction-diffusion process which can separate particles without blowing up. This induces an Allen-Cahn message passing (ACMP) for graph neural networks where the numerical iteration for the particle system solution constitutes the message passing propagation. ACMP which has a simple implementation with a neural ODE solver can propel the network depth up to one hundred of layers with theoretically proven strictly positive lower bound of the Dirichlet energy. It thus provides a deep model of GNNs circumventing the common GNN problem of oversmoothing. GNNs with ACMP achieve state of the art performance for real-world node classification tasks on both homophilic and heterophilic datasets. Codes are available at https://github.com/ykiiiiii/ACMP.
Authors: Matteo Gamba, Hossein Azizpour, M{\aa}rten Bj\"orkman
Abstract: Existing bounds on the generalization error of deep networks assume some form of smooth or bounded dependence on the input variable, falling short of investigating the mechanisms controlling such factors in practice. In this work, we present an extensive experimental study of the empirical Lipschitz constant of deep networks undergoing double descent, and highlight non-monotonic trends strongly correlating with the test error. Building a connection between parameter-space and input-space gradients for SGD around a critical point, we isolate two important factors -- namely loss landscape curvature and distance of parameters from initialization -- respectively controlling optimization dynamics around a critical point and bounding model function complexity, even beyond the training data. Our study presents novels insights on implicit regularization via overparameterization, and effective model complexity for networks trained in practice.
Authors: Pierre Wolinski
Abstract: When training large models, such as neural networks, the full derivatives of order 2 and beyond are usually inaccessible, due to their computational cost. Therefore, among the second-order optimization methods, it is common to bypass the computation of the Hessian by using first-order information, such as the gradient of the parameters (e.g., quasi-Newton methods) or the activations (e.g., K-FAC). In this paper, we focus on the exact and explicit computation of projections of the Hessian and higher-order derivatives on well-chosen subspaces relevant for optimization. Namely, for a given partition of the set of parameters, we compute tensors that can be seen as "higher-order derivatives according to the partition", at a reasonable cost as long as the number of subsets of the partition remains small. Then, we give some examples of how these tensors can be used. First, we show how to compute a learning rate per subset of parameters, which can be used for hyperparameter tuning. Second, we show how to use these tensors at order 2 to construct an optimization method that uses information contained in the Hessian. Third, we show how to use these tensors at order 3 (information contained in the third derivative of the loss) to regularize this optimization method. The resulting training step has several interesting properties, including: it takes into account long-range interactions between the layers of the trained neural network, which is usually not the case in similar methods (e.g., K-FAC); the trajectory of the optimization is invariant under affine layer-wise reparameterization.
Authors: Xupeng Miao, Gabriele Oliaro, Zhihao Zhang, Xinhao Cheng, Hongyi Jin, Tianqi Chen, Zhihao Jia
Abstract: In the rapidly evolving landscape of artificial intelligence (AI), generative large language models (LLMs) stand at the forefront, revolutionizing how we interact with our data. However, the computational intensity and memory consumption of deploying these models present substantial challenges in terms of serving efficiency, particularly in scenarios demanding low latency and high throughput. This survey addresses the imperative need for efficient LLM serving methodologies from a machine learning system (MLSys) research perspective, standing at the crux of advanced AI innovations and practical system optimizations. We provide in-depth analysis, covering a spectrum of solutions, ranging from cutting-edge algorithmic modifications to groundbreaking changes in system designs. The survey aims to provide a comprehensive understanding of the current state and future directions in efficient LLM serving, offering valuable insights for researchers and practitioners in overcoming the barriers of effective LLM deployment, thereby reshaping the future of AI.
Authors: Dehua Peng, Zhipeng Gui, Wenzhang Wei, Fa Li, Jie Gui, Huayi Wu, Jianya Gong
Abstract: As a pivotal branch of machine learning, manifold learning uncovers the intrinsic low-dimensional structure within complex nonlinear manifolds in high-dimensional space for visualization, classification, clustering, and gaining key insights. Although existing techniques have achieved remarkable successes, they suffer from extensive distortions of cluster structure, which hinders the understanding of underlying patterns. Scalability issues also limit their applicability for handling large-scale data. We hence propose a sampling-based Scalable manifold learning technique that enables Uniform and Discriminative Embedding, namely SUDE, for large-scale and high-dimensional data. It starts by seeking a set of landmarks to construct the low-dimensional skeleton of the entire data, and then incorporates the non-landmarks into the learned space based on the constrained locally linear embedding (CLLE). We empirically validated the effectiveness of SUDE on synthetic datasets and real-world benchmarks, and applied it to analyze single-cell data and detect anomalies in electrocardiogram (ECG) signals. SUDE exhibits distinct advantage in scalability with respect to data size and embedding dimension, and has promising performance in cluster separation, integrity, and global structure preservation. The experiments also demonstrate notable robustness in embedding quality as the sampling rate decreases.
Authors: Yilin Zhang, Cai Xu, Han Jiang, Ziyu Guan, Wei Zhao, Xiaofei He, Murat Sensoy
Abstract: Multi-view learning methods often focus on improving decision accuracy while neglecting the decision uncertainty, which significantly restricts their applications in safety-critical scenarios. To address this, trusted multi-view learning methods estimate prediction uncertainties by learning class distributions from each instance. However, these methods heavily rely on high quality ground-truth labels. This motivates us to delve into a new problem: how to develop a reliable multi-view learning model under the guidance of noisy labels? We propose the Trusted Multi view Noise Refining (TMNR) method to address this challenge by modeling label noise arising from low-quality data features and easily-confused classes. TMNR employs evidential deep neural networks to construct view-specific opinions that capture both beliefs and uncertainty. These opinions are then transformed through noise correlation matrices to align with the noisy supervision, where matrix elements are constrained by sample uncertainty to reflect label reliability. Furthermore, considering the challenge of jointly optimizing the evidence network and noise correlation matrices under noisy supervision, we further propose Trusted Multi-view Noise Re-Refining (TMNR^2 ), which disentangles this complex co-training problem by establishing different training objectives for distinct modules. TMNR^2 identifies potentially mislabeled samples through evidence-label consistency and generates pseudo-labels from neighboring information. By assigning clean samples to optimize evidential networks and noisy samples to guide noise correlation matrices, respectively, TMNR^2 reduces mapping interference and achieves stabilizes training. Experimental results demonstrate that TMNR^2 significantly outperforms baseline methods, with average accuracy improvements of 7% on datasets with 50% label noise.
Authors: Dario Fenoglio, Gabriele Dominici, Pietro Barbiero, Alberto Tonda, Martin Gjoreski, Marc Langheinrich
Abstract: Federated Learning (FL), a privacy-aware approach in distributed deep learning environments, enables many clients to collaboratively train a model without sharing sensitive data, thereby reducing privacy risks. However, enabling human trust and control over FL systems requires understanding the evolving behaviour of clients, whether beneficial or detrimental for the training, which still represents a key challenge in the current literature. To address this challenge, we introduce Federated Behavioural Planes (FBPs), a novel method to analyse, visualise, and explain the dynamics of FL systems, showing how clients behave under two different lenses: predictive performance (error behavioural space) and decision-making processes (counterfactual behavioural space). Our experiments demonstrate that FBPs provide informative trajectories describing the evolving states of clients and their contributions to the global model, thereby enabling the identification of clusters of clients with similar behaviours. Leveraging the patterns identified by FBPs, we propose a robust aggregation technique named Federated Behavioural Shields to detect malicious or noisy client models, thereby enhancing security and surpassing the efficacy of existing state-of-the-art FL defense mechanisms. Our code is publicly available on GitHub.
Authors: Haibo Wang, Lutfu S. Sua, Bahram Alidaee
Abstract: The increasing scale and complexity of global supply chains have led to new challenges spanning various fields, such as supply chain disruptions due to long waiting lines at the ports, material shortages, and inflation. Coupled with the size of supply chains and the availability of vast amounts of data, efforts towards tackling such challenges have led to an increasing interest in applying machine learning methods in many aspects of supply chains. Unlike other solutions, ML techniques, including Random Forest, XGBoost, LightGBM, and Neural Networks, make predictions and approximate optimal solutions faster. This paper presents an automated ML framework to enhance supply chain security by detecting fraudulent activities, predicting maintenance needs, and forecasting material backorders. Using datasets of varying sizes, results show that fraud detection achieves an 88% accuracy rate using sampling methods, machine failure prediction reaches 93.4% accuracy, and material backorder prediction achieves 89.3% accuracy. Hyperparameter tuning significantly improved the performance of these models, with certain supervised techniques like XGBoost and LightGBM reaching up to 100% precision. This research contributes to supply chain security by streamlining data preprocessing, feature selection, model optimization, and inference deployment, addressing critical challenges and boosting operational efficiency.
Authors: Calvin McCarter
Abstract: Despite much work on advanced deep learning and generative modeling techniques for tabular data generation and imputation, traditional methods have continued to win on imputation benchmarks. We herein present UnmaskingTrees, a simple method for tabular imputation (and generation) employing gradient-boosted decision trees which are used to incrementally unmask individual features. On a benchmark for out-of-the-box performance on 27 small tabular datasets, UnmaskingTrees offers leading performance on imputation; state-of-the-art performance on generation given data with missingness; and competitive performance on vanilla generation given data without missingness. To solve the conditional generation subproblem, we propose a tabular probabilistic prediction method, BaltoBot, which fits a balanced tree of boosted tree classifiers. Unlike older methods, it requires no parametric assumption on the conditional distribution, accommodating features with multimodal distributions; unlike newer diffusion methods, it offers fast sampling, closed-form density estimation, and flexible handling of discrete variables. We finally consider our two approaches as meta-algorithms, demonstrating in-context learning-based generative modeling with TabPFN.
Authors: Helen Jin, Shreya Havaldar, Chaehyeon Kim, Anton Xue, Weiqiu You, Helen Qu, Marco Gatti, Daniel A Hashimoto, Bhuvnesh Jain, Amin Madani, Masao Sako, Lyle Ungar, Eric Wong
Abstract: Feature-based methods are commonly used to explain model predictions, but these methods often implicitly assume that interpretable features are readily available. However, this is often not the case for high-dimensional data, and it can be hard even for domain experts to mathematically specify which features are important. Can we instead automatically extract collections or groups of features that are aligned with expert knowledge? To address this gap, we present FIX (Features Interpretable to eXperts), a benchmark for measuring how well a collection of features aligns with expert knowledge. In collaboration with domain experts, we propose FIXScore, a unified expert alignment measure applicable to diverse real-world settings across cosmology, psychology, and medicine domains in vision, language, and time series data modalities. With FIXScore, we find that popular feature-based explanation methods have poor alignment with expert-specified knowledge, highlighting the need for new methods that can better identify features interpretable to experts.
Authors: Aakriti Agrawal, Mucong Ding, Zora Che, Chenghao Deng, Anirudh Satheesh, Bang An, Bayan Bruss, John Langford, Furong Huang
Abstract: With Large Language Models (LLMs) rapidly approaching and potentially surpassing human-level performance, it has become imperative to develop approaches capable of effectively supervising and enhancing these powerful models using smaller, human-level models exposed to only human-level data. We address this critical weak-to-strong (W2S) generalization challenge by proposing a novel method aimed at improving weak experts, by training on the same limited human-level data, enabling them to generalize to complex, super-human-level tasks. Our approach, called **EnsemW2S**, employs a token-level ensemble strategy that iteratively combines multiple weak experts, systematically addressing the shortcomings identified in preceding iterations. By continuously refining these weak models, we significantly enhance their collective ability to supervise stronger student models. We extensively evaluate the generalization performance of both the ensemble of weak experts and the subsequent strong student model across in-distribution (ID) and out-of-distribution (OOD) datasets. For OOD, we specifically introduce question difficulty as an additional dimension for defining distributional shifts. Our empirical results demonstrate notable improvements, achieving 4%, and 3.2% improvements on ID datasets and, upto 6% and 2.28% on OOD datasets for experts and student models respectively, underscoring the effectiveness of our proposed method in advancing W2S generalization.
Authors: Andy K Zhang, Kevin Klyman, Yifan Mai, Yoav Levine, Yian Zhang, Rishi Bommasani, Percy Liang
Abstract: Language models are extensively evaluated, but correctly interpreting evaluation results requires knowledge of train-test overlap which refers to the extent to which the language model is trained on the very data it is being tested on. The public currently lacks adequate information about train-test overlap: most models have no public train-test overlap statistics, and third parties cannot directly measure train-test overlap since they do not have access to the training data. To make this clear, we document the practices of 30 model developers, finding that just 9 developers report train-test overlap: 4 developers release training data under open-source licenses, enabling the community to directly measure train-test overlap, and 5 developers publish their train-test overlap methodology and statistics. By engaging with language model developers, we provide novel information about train-test overlap for three additional developers. Overall, we take the position that language model developers should publish train-test overlap statistics and/or training data whenever they report evaluation results on public test sets. We hope our work increases transparency into train-test overlap to increase the community-wide trust in model evaluations.
Authors: Nathan Phelps, Daniel J. Lizotte, Douglas G. Woolford
Abstract: Imbalanced binary classification problems arise in many fields of study. When using machine learning models for these problems, it is common to subsample the majority class (i.e., undersampling) to create a (more) balanced dataset for model training. This biases the model's predictions because the model learns from a dataset that does not follow the same data generating process as new data. One way of accounting for this bias is to analytically map the resulting predictions to new values based on the sampling rate for the majority class, which was used to create the training dataset. While this approach may work well for some machine learning models, we show that calibrating a random forest this way has unintended negative consequences, including prevalence estimates that can be upwardly biased. These prevalence estimates depend on both i) the number of predictors considered at each split in the random forest; and ii) the sampling rate used. We explain the former using known properties of random forests and analytical calibration. However, in investigating the latter issue, we made a surprising discovery - contrary to the widespread belief that decision trees are biased towards the majority class, they actually can be biased towards the minority class.
Authors: Farzan Moosavi, Bilal Farooq
Abstract: We introduce a multi-modal autonomous delivery optimization framework as a coalition game for a fleet of UAVs and ADRs operating in two overlaying networks to address last-mile delivery in urban environments, including high-density areas and time-critical applications. The problem is defined as multiple depot pickup and delivery with time windows constrained over operational restrictions, such as vehicle battery limitation, precedence time window, and building obstruction. Utilizing the coalition game theory, we investigate cooperation structures among the modes to capture how strategic collaboration can improve overall routing efficiency. To do so, a generalized reinforcement learning model is designed to evaluate the cost-sharing and allocation to different modes to learn the cooperative behaviour with respect to various realistic scenarios. Our methodology leverages an end-to-end deep multi-agent policy gradient method augmented by a novel spatio-temporal adjacency neighbourhood graph attention network using a heterogeneous edge-enhanced attention model and transformer architecture. Several numerical experiments on last-mile delivery applications have been conducted, showing the results from the case study in the city of Mississauga, which shows that despite the incorporation of an extensive network in the graph for two modes and a complex training structure, the model addresses realistic operational constraints and achieves high-quality solutions compared with the existing transformer-based and classical methods. It can perform well on non-homogeneous data distribution, generalizes well on different scales and configurations, and demonstrates a robust cooperative performance under stochastic scenarios across various tasks, which is effectively reflected by coalition analysis and cost allocation to signify the advantage of cooperation.
Authors: Seul Lee, Karsten Kreis, Srimukh Prasad Veccham, Meng Liu, Danny Reidenbach, Yuxing Peng, Saee Paliwal, Weili Nie, Arash Vahdat
Abstract: Drug discovery is a complex process that involves multiple stages and tasks. However, existing molecular generative models can only tackle some of these tasks. We present Generalist Molecular generative model (GenMol), a versatile framework that uses only a single discrete diffusion model to handle diverse drug discovery scenarios. GenMol generates Sequential Attachment-based Fragment Embedding (SAFE) sequences through non-autoregressive bidirectional parallel decoding, thereby allowing the utilization of a molecular context that does not rely on the specific token ordering while having better sampling efficiency. GenMol uses fragments as basic building blocks for molecules and introduces fragment remasking, a strategy that optimizes molecules by regenerating masked fragments, enabling effective exploration of chemical space. We further propose molecular context guidance (MCG), a guidance method tailored for masked discrete diffusion of GenMol. GenMol significantly outperforms the previous GPT-based model in de novo generation and fragment-constrained generation, and achieves state-of-the-art performance in goal-directed hit generation and lead optimization. These results demonstrate that GenMol can tackle a wide range of drug discovery tasks, providing a unified and versatile approach for molecular design. Our code is available at https://github.com/NVIDIA-Digital-Bio/genmol.
Authors: Angelo Rodio, Zheng Chen, Erik G. Larsson
Abstract: Decentralized learning enables distributed agents to collaboratively train a shared machine learning model without a central server, through local computation and peer-to-peer communication. Although each agent retains its dataset locally, sharing local models can still expose private information about the local training datasets to adversaries. To mitigate privacy attacks, a common strategy is to inject random artificial noise at each agent before exchanging local models between neighbors. However, this often leads to utility degradation due to the negative effects of cumulated artificial noise on the learning algorithm. In this work, we introduce CorN-DSGD, a novel covariance-based framework for generating correlated privacy noise across agents, which unifies several state-of-the-art methods as special cases. By leveraging network topology and mixing weights, CorN-DSGD optimizes the noise covariance to achieve network-wide noise cancellation. Experimental results show that CorN-DSGD cancels more noise than existing pairwise correlation schemes, improving model performance under formal privacy guarantees.
Authors: Fabian Schaipp, Alexander H\"agele, Adrien Taylor, Umut Simsekli, Francis Bach
Abstract: We show that learning-rate schedules for large model training behave surprisingly similar to a performance bound from non-smooth convex optimization theory. We provide a bound for the constant schedule with linear cooldown; in particular, the practical benefit of cooldown is reflected in the bound due to the absence of logarithmic terms. Further, we show that this surprisingly close match between optimization theory and practice can be exploited for learning-rate tuning: we achieve noticeable improvements for training 124M and 210M Llama-type models by (i) extending the schedule for continued training with optimal learning-rate, and (ii) transferring the optimal learning-rate across schedules.
Authors: Bibek Poudel, Xuan Wang, Weizi Li, Lei Zhu, Kevin Heaslip
Abstract: Reinforcement learning (RL) holds significant promise for adaptive traffic signal control. While existing RL-based methods demonstrate effectiveness in reducing vehicular congestion, their predominant focus on vehicle-centric optimization leaves pedestrian mobility needs and safety challenges unaddressed. In this paper, we present a deep RL framework for adaptive control of eight traffic signals along a real-world urban corridor, jointly optimizing both pedestrian and vehicular efficiency. Our single-agent policy is trained using real-world pedestrian and vehicle demand data derived from Wi-Fi logs and video analysis. The results demonstrate significant performance improvements over traditional fixed-time signals, reducing average wait times per pedestrian and per vehicle by up to 67% and 52% respectively, while simultaneously decreasing total wait times for both groups by up to 67% and 53%. Additionally, our results demonstrate generalization capabilities across varying traffic demands, including conditions entirely unseen during training, validating RL's potential for developing transportation systems that serve all road users.
Authors: Guangqiang Li, M. Amine Atoui, Xiangshun Li
Abstract: Fault diagnosis in multimode processes plays a critical role in ensuring the safe operation of industrial systems across multiple modes. It faces a great challenge yet to be addressed - that is, the significant distributional differences among monitoring data from multiple modes make it difficult for the models to extract shared feature representations related to system health conditions. In response to this problem, this paper introduces a novel method called attention-based multiscale temporal fusion network. The multiscale depthwise convolution and gated recurrent unit are employed to extract multiscale contextual local features and long-short-term features. Instance normalization is applied to suppress mode-specific information. Furthermore, a temporal attention mechanism is designed to focus on critical time points with higher cross-mode shared information, thereby enhancing the accuracy of fault diagnosis. The proposed model is applied to Tennessee Eastman process dataset and three-phase flow facility dataset. The experiments demonstrate that the proposed model achieves superior diagnostic performance and maintains a small model size. The source code will be available on GitHub at https://github.com/GuangqiangLi/AMTFNet.
Authors: Yuzhou Zhu
Abstract: Deep neural networks employ specialized architectures for vision, sequential and language tasks, yet this proliferation obscures their underlying commonalities. We introduce a unified matrix-order framework that casts convolutional, recurrent and self-attention operations as sparse matrix multiplications. Convolution is realized via an upper-triangular weight matrix performing first-order transformations; recurrence emerges from a lower-triangular matrix encoding stepwise updates; attention arises naturally as a third-order tensor factorization. We prove algebraic isomorphism with standard CNN, RNN and Transformer layers under mild assumptions. Empirical evaluations on image classification (MNIST, CIFAR-10/100, Tiny ImageNet), time-series forecasting (ETTh1, Electricity Load Diagrams) and language modeling/classification (AG News, WikiText-2, Penn Treebank) confirm that sparse-matrix formulations match or exceed native model performance while converging in comparable or fewer epochs. By reducing architecture design to sparse pattern selection, our matrix perspective aligns with GPU parallelism and leverages mature algebraic optimization tools. This work establishes a mathematically rigorous substrate for diverse neural architectures and opens avenues for principled, hardware-aware network design.
Authors: Hao Li, Bowen Deng, Chang Xu, Zhiyuan Feng, Viktor Schlegel, Yu-Hao Huang, Yizheng Sun, Jingyuan Sun, Kailai Yang, Yiyao Yu, Jiang Bian
Abstract: A unified foundation model for medical time series -- pretrained on open access and ethics board-approved medical corpora -- offers the potential to reduce annotation burdens, minimize model customization, and enable robust transfer across clinical institutions, modalities, and tasks, particularly in data-scarce or privacy-constrained environments. However, existing generalist time series foundation models struggle to handle medical time series data due to their inherent challenges, including irregular intervals, heterogeneous sampling rates, and frequent missing values. To address these challenges, we introduce MIRA, a unified foundation model specifically designed for medical time series forecasting. MIRA incorporates a Continuous-Time Rotary Positional Encoding that enables fine-grained modeling of variable time intervals, a frequency-specific mixture-of-experts layer that routes computation across latent frequency regimes to further promote temporal specialization, and a Continuous Dynamics Extrapolation Block based on Neural ODE that models the continuous trajectory of latent states, enabling accurate forecasting at arbitrary target timestamps. Pretrained on a large-scale and diverse medical corpus comprising over 454 billion time points collect from publicly available datasets, MIRA achieves reductions in forecasting errors by an average of 10% and 7% in out-of-distribution and in-distribution scenarios, respectively, when compared to other zero-shot and fine-tuned baselines. We also introduce a comprehensive benchmark spanning multiple downstream clinical tasks, establishing a foundation for future research in medical time series modeling.
Authors: Jo\~ao Manoel Herrera Pinheiro, Suzana Vilas Boas de Oliveira, Thiago Henrique Segreto Silva, Pedro Antonio Rabelo Saraiva, Enzo Ferreira de Souza, Ricardo V. Godoy, Leonardo Andr\'e Ambrosio, Marcelo Becker
Abstract: This research addresses the critical lack of comprehensive studies on feature scaling by systematically evaluating 12 scaling techniques - including several less common transformations - across 14 different Machine Learning algorithms and 16 datasets for classification and regression tasks. We meticulously analyzed impacts on predictive performance (using metrics such as accuracy, MAE, MSE, and $R^2$) and computational costs (training time, inference time, and memory usage). Key findings reveal that while ensemble methods (such as Random Forest and gradient boosting models like XGBoost, CatBoost and LightGBM) demonstrate robust performance largely independent of scaling, other widely used models such as Logistic Regression, SVMs, TabNet, and MLPs show significant performance variations highly dependent on the chosen scaler. This extensive empirical analysis, with all source code, experimental results, and model parameters made publicly available to ensure complete transparency and reproducibility, offers model-specific crucial guidance to practitioners on the need for an optimal selection of feature scaling techniques.
Authors: Tushar Talukder Showrav, Soyabul Islam Lincoln, Md. Kamrul Hasan
Abstract: Deep learning has significantly propelled the performance of ECG arrhythmia classification, yet its clinical adoption remains hindered by challenges in interpretability and deployment on resource-constrained edge devices. To bridge this gap, we propose EXGnet, a novel and reliable ECG arrhythmia classification network tailored for single-lead signals, specifically designed to balance high accuracy, explainability, and edge compatibility. EXGnet integrates XAI supervision during training via a normalized cross-correlation based loss, directing the model's attention to clinically relevant ECG regions, similar to a cardiologist's focus. This supervision is driven by automatically generated ground truth, derived through an innovative heart rate variability-based approach, without the need for manual annotation. To enhance classification accuracy without compromising deployment simplicity, we incorporate quantitative ECG features during training. These enrich the model with multi-domain knowledge but are excluded during inference, keeping the model lightweight for edge deployment. Additionally, we introduce an innovative multiresolution block to efficiently capture both short and long-term signal features while maintaining computational efficiency. Rigorous evaluation on the Chapman and Ningbo benchmark datasets validates the supremacy of EXGnet, which achieves average five-fold accuracies of 98.762% and 96.932%, and F1-scores of 97.910% and 95.527%, respectively. Comprehensive ablation studies and both quantitative and qualitative interpretability assessment confirm that the XAI guidance is pivotal, demonstrably enhancing the model's focus and trustworthiness. Overall, EXGnet sets a new benchmark by combining high-performance arrhythmia classification with interpretability, paving the way for more trustworthy and accessible portable ECG based health monitoring systems.
Authors: Gabriel J. Perin, Runjin Chen, Xuxi Chen, Nina S. T. Hirata, Zhangyang Wang, Junyuan Hong
Abstract: Large Language Models (LLMs) have become indispensable in real-world applications. However, their widespread adoption raises significant safety concerns, particularly in responding to socially harmful questions. Despite substantial efforts to improve model safety through alignment, aligned models can still have their safety protections undermined by subsequent fine-tuning - even when the additional training data appears benign. In this paper, we empirically demonstrate that this vulnerability stems from the sensitivity of safety-critical low-rank subspaces in LLM parameters to fine-tuning. Building on this insight, we propose a novel training-free method, termed Low-Rank Extrapolation (LoX), to enhance safety robustness by extrapolating the safety subspace of an aligned LLM. Our experimental results confirm the effectiveness of LoX, demonstrating significant improvements in robustness against both benign and malicious fine-tuning attacks while preserving the model's adaptability to new tasks. For instance, LoX leads to 11% to 54% absolute reductions in attack success rates (ASR) facing benign or malicious fine-tuning attacks. By investigating the ASR landscape of parameters, we attribute the success of LoX to that the extrapolation moves LLM parameters to a flatter zone, thereby less sensitive to perturbations. The code is available at github.com/VITA-Group/LoX.
Authors: Prajwal Koirala, Cody Fleming
Abstract: Generative models such as diffusion and flow-matching offer expressive policies for offline reinforcement learning (RL) by capturing rich, multimodal action distributions, but their iterative sampling introduces high inference costs and training instability due to gradient propagation across sampling steps. We propose the \textit{Single-Step Completion Policy} (SSCP), a generative policy trained with an augmented flow-matching objective to predict direct completion vectors from intermediate flow samples, enabling accurate, one-shot action generation. In an off-policy actor-critic framework, SSCP combines the expressiveness of generative models with the training and inference efficiency of unimodal policies, without requiring long backpropagation chains. Our method scales effectively to offline, offline-to-online, and online RL settings, offering substantial gains in speed and adaptability over diffusion-based baselines. We further extend SSCP to goal-conditioned RL, enabling flat policies to exploit subgoal structures without explicit hierarchical inference. SSCP achieves strong results across standard offline RL and behavior cloning benchmarks, positioning it as a versatile, expressive, and efficient framework for deep RL and sequential decision-making.
Authors: Yilie Huang, Xun Yu Zhou
Abstract: We study reinforcement learning (RL) for the same class of continuous-time stochastic linear--quadratic (LQ) control problems as in \cite{huang2024sublinear}, where volatilities depend on both states and controls while states are scalar-valued and running control rewards are absent. We propose a model-free, data-driven exploration mechanism that adaptively adjusts entropy regularization by the critic and policy variance by the actor. Unlike the constant or deterministic exploration schedules employed in \cite{huang2024sublinear}, which require extensive tuning for implementations and ignore learning progresses during iterations, our adaptive exploratory approach boosts learning efficiency with minimal tuning. Despite its flexibility, our method achieves a sublinear regret bound that matches the best-known model-free results for this class of LQ problems, which were previously derived only with fixed exploration schedules. Numerical experiments demonstrate that adaptive explorations accelerate convergence and improve regret performance compared to the non-adaptive model-free and model-based counterparts.
Authors: Yue Ding, Conor McCarthy, Kevin O'Shea, Mingming Liu
Abstract: With the rise of smart mobility and shared e-mobility services, numerous advanced technologies have been applied to this field. Cloud-based traffic simulation solutions have flourished, offering increasingly realistic representations of the evolving mobility landscape. LLMs have emerged as pioneering tools, providing robust support for various applications, including intelligent decision-making, user interaction, and real-time traffic analysis. As user demand for e-mobility continues to grow, delivering comprehensive end-to-end solutions has become crucial. In this paper, we present a cloud-based, LLM-powered shared e-mobility platform, integrated with a mobile application for personalized route recommendations. The optimization module is evaluated based on travel time and cost across different traffic scenarios. Additionally, the LLM-powered RAG framework is evaluated at the schema level for different users, using various evaluation methods. Schema-level RAG with XiYanSQL achieves an average execution accuracy of 0.81 on system operator queries and 0.98 on user queries.
Authors: Agata Kaczmarek, Dawid P{\l}udowski, Piotr Wilczy\'nski, Krzysztof Kotowski, Ramez Shendy, Evridiki Ntagiou, Jakub Nalepa, Artur Janicki, Przemys{\l}aw Biecek
Abstract: The "Fake or Real" competition hosted on Kaggle (https://www.kaggle.com/competitions/fake-or-real-the-impostor-hunt ) is the second part of a series of follow-up competitions and hackathons related to the "Assurance for Space Domain AI Applications" project funded by the European Space Agency (https://assurance-ai.space-codev.org/ ). The competition idea is based on two real-life AI security threats identified within the project -- data poisoning and overreliance in Large Language Models. The task is to distinguish between the proper output from LLM and the output generated under malicious modification of the LLM. As this problem was not extensively researched, participants are required to develop new techniques to address this issue or adjust already existing ones to this problem's statement.
URLs: https://www.kaggle.com/competitions/fake-or-real-the-impostor-hunt, https://assurance-ai.space-codev.org/
Authors: Obumneme Zimuzor Nwafor, Mohammed Abdul Majeed Al Hooti
Abstract: As nations seek sustainable alternatives to fossil fuels, green hydrogen has emerged as a promising strategic pathway toward decarbonisation, particularly in solar-rich arid regions. However, identifying optimal locations for hydrogen production requires the integration of complex environmental, atmospheric, and infrastructural factors, often compounded by limited availability of direct hydrogen yield data. This study presents a novel Artificial Intelligence (AI) framework for computing green hydrogen yield and site suitability index using mean absolute SHAP (SHapley Additive exPlanations) values. This framework consists of a multi-stage pipeline of unsupervised multi-variable clustering, supervised machine learning classifier and SHAP algorithm. The pipeline trains on an integrated meteorological, topographic and temporal dataset and the results revealed distinct spatial patterns of suitability and relative influence of the variables. With model predictive accuracy of 98%, the result also showed that water proximity, elevation and seasonal variation are the most influential factors determining green hydrogen site suitability in Oman with mean absolute shap values of 2.470891, 2.376296 and 1.273216 respectively. Given limited or absence of ground-truth yield data in many countries that have green hydrogen prospects and ambitions, this study offers an objective and reproducible alternative to subjective expert weightings, thus allowing the data to speak for itself and potentially discover novel latent groupings without pre-imposed assumptions. This study offers industry stakeholders and policymakers a replicable and scalable tool for green hydrogen infrastructure planning and other decision making in data-scarce regions.
Authors: Defeng Liu, Ying Liu, Carson Eisenach
Abstract: In this work, we study how to efficiently apply reinforcement learning (RL) for solving large-scale stochastic optimization problems by leveraging intervention models. The key of the proposed methodology is to better explore the solution space by simulating and composing the stochastic processes using pre-trained deep learning (DL) models. We demonstrate our approach on a challenging real-world application, the multi-sourcing multi-period inventory management problem in supply chain optimization. In particular, we employ deep RL models for learning and forecasting the stochastic supply chain processes under a range of assumptions. Moreover, we also introduce a constraint coordination mechanism, designed to forecast dual costs given the cross-products constraints in the inventory network. We highlight that instead of directly modeling the complex physical constraints into the RL optimization problem and solving the stochastic problem as a whole, our approach breaks down those supply chain processes into scalable and composable DL modules, leading to improved performance on large real-world datasets. We also outline open problems for future research to further investigate the efficacy of such models.
Authors: Adrian Kaiser, Claudiu Leoveanu-Condrei, Ryan Gold, Marius-Constantin Dinu, Markus Hofmarcher
Abstract: The synergy between symbolic knowledge, often represented by Knowledge Graphs (KGs), and the generative capabilities of neural networks is central to advancing neurosymbolic AI. A primary bottleneck in realizing this potential is the difficulty of automating KG construction, which faces challenges related to output reliability, consistency, and verifiability. These issues can manifest as structural inconsistencies within the generated graphs, such as the formation of disconnected $\textit{isolated islands}$ of data or the inaccurate conflation of abstract classes with specific instances. To address these challenges, we propose HyDRA, a $\textbf{Hy}$brid-$\textbf{D}$riven $\textbf{R}$easoning $\textbf{A}$rchitecture designed for verifiable KG automation. Given a domain or an initial set of documents, HyDRA first constructs an ontology via a panel of collaborative neurosymbolic agents. These agents collaboratively agree on a set of competency questions (CQs) that define the scope and requirements the ontology must be able to answer. Given these CQs, we build an ontology graph that subsequently guides the automated extraction of triplets for KG generation from arbitrary documents. Inspired by design-by-contracts (DbC) principles, our method leverages verifiable contracts as the primary control mechanism to steer the generative process of Large Language Models (LLMs). To verify the output of our approach, we extend beyond standard benchmarks and propose an evaluation framework that assesses the functional correctness of the resulting KG by leveraging symbolic verifications as described by the neurosymbolic AI framework, $\textit{SymbolicAI}$. This work contributes a hybrid-driven architecture for improving the reliability of automated KG construction and the exploration of evaluation methods for measuring the functional integrity of its output. The code is publicly available.
Authors: Tom\'as H\"uttebr\"aucker, Mario Edoardo Pandolfo, Simone Fiorellino, Emilio Calvanese Strinati, Paolo Di Lorenzo
Abstract: Semantic communication systems introduce a new paradigm in wireless communications, focusing on transmitting the intended meaning rather than ensuring strict bit-level accuracy. These systems often rely on Deep Neural Networks (DNNs) to learn and encode meaning directly from data, enabling more efficient communication. However, in multi-user settings where interacting agents are trained independently-without shared context or joint optimization-divergent latent representations across AI-native devices can lead to semantic mismatches, impeding mutual understanding even in the absence of traditional transmission errors. In this work, we address semantic mismatch in Multiple-Input Multiple-Output (MIMO) channels by proposing a joint physical and semantic channel equalization framework that leverages the presence of Reconfigurable Intelligent Surfaces (RIS). The semantic equalization is implemented as a sequence of transformations: (i) a pre-equalization stage at the transmitter; (ii) propagation through the RIS-aided channel; and (iii) a post-equalization stage at the receiver. We formulate the problem as a constrained Minimum Mean Squared Error (MMSE) optimization and propose two solutions: (i) a linear semantic equalization chain, and (ii) a non-linear DNN-based semantic equalizer. Both methods are designed to operate under semantic compression in the latent space and adhere to transmit power constraints. Through extensive evaluations, we show that the proposed joint equalization strategies consistently outperform conventional, disjoint approaches to physical and semantic channel equalization across a broad range of scenarios and wireless channel conditions.
Authors: Yang Bai, Kuntara Pukthuanthong
Abstract: We use multi-class machine learning classifiers to identify the stocks that outperform or underperform other stocks. The resulting long-short portfolios achieve annual Sharpe ratios of 1.67 (value-weighted) and 3.35 (equal-weighted), with annual alphas ranging from 29\% to 48\%. These results persist after controlling for machine learning regressions and remain robust among large-cap stocks. Machine uncertainty, as measured by predicted probabilities, impairs the prediction performance. Stocks with higher machine uncertainty experience lower returns, particularly when human proxies of information uncertainty align with machine uncertainty. Consistent with the literature, such an effect is driven by the past underperformers.
Authors: El Mahdi Chayti, Nikita Doikov, Martin Jaggi
Abstract: We study stochastic Cubic Newton methods for solving general possibly non-convex minimization problems. We propose a new framework, which we call the helper framework, that provides a unified view of the stochastic and variance-reduced second-order algorithms equipped with global complexity guarantees. It can also be applied to learning with auxiliary information. Our helper framework offers the algorithm designer high flexibility for constructing and analyzing the stochastic Cubic Newton methods, allowing arbitrary size batches, and the use of noisy and possibly biased estimates of the gradients and Hessians, incorporating both the variance reduction and the lazy Hessian updates. We recover the best-known complexities for the stochastic and variance-reduced Cubic Newton, under weak assumptions on the noise. A direct consequence of our theory is the new lazy stochastic second-order method, which significantly improves the arithmetic complexity for large dimension problems. We also establish complexity bounds for the classes of gradient-dominated objectives, that include convex and strongly convex problems. For Auxiliary Learning, we show that using a helper (auxiliary function) can outperform training alone if a given similarity measure is small.
Authors: Chengkun Li, Gr\'egoire Clart\'e, Martin J{\o}rgensen, Luigi Acerbi
Abstract: In applied Bayesian inference scenarios, users may have access to a large number of pre-existing model evaluations, for example from maximum-a-posteriori (MAP) optimization runs. However, traditional approximate inference techniques make little to no use of this available information. We propose the framework of post-process Bayesian inference as a means to obtain a quick posterior approximation from existing target density evaluations, with no further model calls. Within this framework, we introduce Variational Sparse Bayesian Quadrature (VSBQ), a method for post-process approximate inference for models with black-box and potentially noisy likelihoods. VSBQ reuses existing target density evaluations to build a sparse Gaussian process (GP) surrogate model of the log posterior density function. Subsequently, we leverage sparse-GP Bayesian quadrature combined with variational inference to achieve fast approximate posterior inference over the surrogate. We validate our method on challenging synthetic scenarios and real-world applications from computational neuroscience. The experiments show that VSBQ builds high-quality posterior approximations by post-processing existing optimization traces, with no further model evaluations.
Authors: Yingtai Xiao, Guanlin He, Levent Toksoz, Zeyu Ding, Danfeng Zhang, Daniel Kifer
Abstract: Noisy marginals are a common form of confidentiality protecting data release and are useful for many downstream tasks such as contingency table analysis, construction of Bayesian networks, and even synthetic data generation. Privacy mechanisms that provide unbiased noisy answers to linear queries (such as marginals) are known as matrix mechanisms. We propose ResidualPlanner and ResidualPlanner+, two highly scalable matrix mechanisms. ResidualPlanner is both optimal and scalable for answering marginal queries with Gaussian noise, while ResidualPlanner+ provides support for more general workloads, such as combinations of marginals and range queries or prefix-sum queries. ResidualPlanner can optimize for many loss functions that can be written as a convex function of marginal variances (prior work was restricted to just one predefined objective function). ResidualPlanner can optimize the accuracy of marginals in large scale settings in seconds, even when the previous state of the art (HDMM) runs out of memory. It even runs on datasets with 100 attributes in a couple of minutes. Furthermore, ResidualPlanner can efficiently compute variance/covariance values for each marginal (prior methods quickly run out of memory, even for relatively small datasets). ResidualPlanner+ provides support for more complex workloads that combine marginal and range/prefix-sum queries (e.g., a marginal on race, a range query on age, and a combined race/age tabulation that answers age range queries for each race). It even supports custom user-defined workloads on different attributes. With this added flexibility, ResidualPlanner+ is not necessarily optimal, however it is still extremely scalable and outperforms the prior state-of-the-art (HDMM) on prefix-sum queries both in terms of accuracy and speed.
Authors: Matthew B. Webster, Joonnyong Lee
Abstract: The task of blind source separation (BSS) involves separating sources from a mixture without prior knowledge of the sources or the mixing system. Single-channel mixtures and non-linear mixtures are a particularly challenging problem in BSS. In this paper, we propose a novel method for addressing BSS with single-channel non-linear mixtures by leveraging the natural feature subspace specialization ability of multi-encoder autoencoders. During the training phase, our method unmixes the input into the separate encoding spaces of the multi-encoder network and then remixes these representations within the decoder for a reconstruction of the input. Then to perform source inference, we introduce a novel encoding masking technique whereby masking out all but one of the encodings enables the decoder to estimate a source signal. To this end, we also introduce a sparse mixing loss that encourages sparse remixing of source encodings throughout the decoder and a so-called zero reconstruction loss on the decoder for coherent source estimations. To analyze and evaluate our method, we conduct experiments on a toy dataset, designed to demonstrate this property of feature subspace specialization, and with real-world biosignal recordings from a polysomnography sleep study for extracting respiration from electrocardiogram and photoplethysmography signals.
Authors: David Winderl, Nicola Franco, Jeanette Miriam Lorenz
Abstract: With the rapid advancement of Quantum Machine Learning (QML), the critical need to enhance security measures against adversarial attacks and protect QML models becomes increasingly evident. In this work, we outline the connection between quantum noise channels and differential privacy (DP), by constructing a family of noise channels which are inherently $\epsilon$-DP: $(\alpha, \gamma)$-channels. Through this approach, we successfully replicate the $\epsilon$-DP bounds observed for depolarizing and random rotation channels, thereby affirming the broad generality of our framework. Additionally, we use a semi-definite program to construct an optimally robust channel. In a small-scale experimental evaluation, we demonstrate the benefits of using our optimal noise channel over depolarizing noise, particularly in enhancing adversarial accuracy. Moreover, we assess how the variables $\alpha$ and $\gamma$ affect the certifiable robustness and investigate how different encoding methods impact the classifier's robustness.
Authors: Robert McCarthy, Daniel C. H. Tan, Dominik Schmidt, Fernando Acero, Nathan Herr, Yilun Du, Thomas G. Thuruthel, Zhibin Li
Abstract: Scaling deep learning to massive and diverse internet data has driven remarkable breakthroughs in domains such as video generation and natural language processing. Robot learning, however, has thus far failed to replicate this success and remains constrained by a scarcity of available data. Learning from videos (LfV) methods aim to address this data bottleneck by augmenting traditional robot data with large-scale internet video. This video data provides foundational information regarding physical dynamics, behaviours, and tasks, and can be highly informative for general-purpose robots. This survey systematically examines the emerging field of LfV. We first outline essential concepts, including detailing fundamental LfV challenges such as distribution shift and missing action labels in video data. Next, we comprehensively review current methods for extracting knowledge from large-scale internet video, overcoming LfV challenges, and improving robot learning through video-informed training. The survey concludes with a critical discussion of future opportunities. Here, we emphasize the need for scalable foundation model approaches that can leverage the full range of available internet video and enhance the learning of robot policies and dynamics models. Overall, the survey aims to inform and catalyse future LfV research, driving progress towards general-purpose robots.
Authors: Pascal K\"undig, Fabio Sigrist
Abstract: We introduce a novel machine learning model for credit risk by combining tree-boosting with a latent spatio-temporal Gaussian process model accounting for frailty correlation. This allows for modeling non-linearities and interactions among predictor variables in a flexible data-driven manner and for accounting for spatio-temporal variation that is not explained by observable predictor variables. We also show how estimation and prediction can be done in a computationally efficient manner. In an application to a large U.S. mortgage credit risk data set, we find that both predictive default probabilities for individual loans and predictive loan portfolio loss distributions obtained with our novel approach are more accurate compared to conventional independent linear hazard models and also linear spatio-temporal models. Using interpretability tools for machine learning models, we find that the likely reasons for this outperformance are strong interaction and non-linear effects in the predictor variables and the presence of spatio-temporal frailty effects.
Authors: Alexander Davydov, Franck Djeumou, Marcus Greiff, Makoto Suminaka, Michael Thompson, John Subosits, Thomas Lew
Abstract: Combining data-driven models that adapt online and model predictive control (MPC) has enabled effective control of nonlinear systems. However, when deployed on unstable systems, online adaptation may not be fast enough to ensure reliable simultaneous learning and control. For example, a controller on a vehicle executing highly dynamic maneuvers--such as drifting to avoid an obstacle--may push the vehicle's tires to their friction limits, destabilizing the vehicle and allowing modeling errors to quickly compound and cause a loss of control. To address this challenge, we present an active information gathering framework for identifying vehicle dynamics as quickly as possible. We propose an expressive vehicle dynamics model that leverages Bayesian last-layer meta-learning to enable rapid online adaptation. The model's uncertainty estimates are used to guide informative data collection and quickly improve the model prior to deployment. Dynamic drifting experiments on a Toyota Supra show that (i) the framework enables reliable control of a vehicle at the edge of stability, (ii) online adaptation alone may not suffice for zero-shot control and can lead to undesirable transient errors or spin-outs, and (iii) active data collection helps achieve reliable performance.
Authors: Shuo Tan, Rui Liu, Xuesong Han, XianLei Long, Kai Wan, Linqi Song, Yong Li
Abstract: Deploying Convolutional Neural Networks (CNNs) on resource-constrained devices necessitates efficient management of computational resources, often via distributed environments susceptible to latency from straggler nodes. This paper introduces the Flexible Coded Distributed Convolution Computing (FCDCC) framework to enhance straggler resilience and numerical stability in distributed CNNs. We extend Coded Distributed Computing (CDC) with Circulant and Rotation Matrix Embedding (CRME) which was originally proposed for matrix multiplication to high-dimensional tensor convolution. For the proposed scheme, referred to as the Numerically Stable Coded Tensor Convolution (NSCTC) scheme, we also propose two new coded partitioning schemes: Adaptive-Padding Coded Partitioning (APCP) for the input tensor and Kernel-Channel Coded Partitioning (KCCP) for the filter tensor. These strategies enable linear decomposition of tensor convolutions and encoding them into CDC subtasks, combining model parallelism with coded redundancy for robust and efficient execution. Theoretical analysis identifies an optimal trade-off between communication and storage costs. Empirical results validate the framework's effectiveness in computational efficiency, straggler resilience, and scalability across various CNN architectures.
Authors: Raphael Ruschel, Md Awsafur Rahman, Hardik Prajapati, Suya You, B. S. Manjuanth
Abstract: Understanding video content is pivotal for advancing real-world applications like activity recognition, autonomous systems, and human-computer interaction. While scene graphs are adept at capturing spatial relationships between objects in individual frames, extending these representations to capture dynamic interactions across video sequences remains a significant challenge. To address this, we present TCDSG, Temporally Consistent Dynamic Scene Graphs, an innovative end-to-end framework that detects, tracks, and links subject-object relationships across time, generating action tracklets, temporally consistent sequences of entities and their interactions. Our approach leverages a novel bipartite matching mechanism, enhanced by adaptive decoder queries and feedback loops, ensuring temporal coherence and robust tracking over extended sequences. This method not only establishes a new benchmark by achieving over 60% improvement in temporal recall@k on the Action Genome, OpenPVSG, and MEVA datasets but also pioneers the augmentation of MEVA with persistent object ID annotations for comprehensive tracklet generation. By seamlessly integrating spatial and temporal dynamics, our work sets a new standard in multi-frame video analysis, opening new avenues for high-impact applications in surveillance, autonomous navigation, and beyond.
Authors: Junyang Cai, Serdar Kadioglu, Bistra Dilkina
Abstract: Mixed-integer programming (MIP) is a powerful paradigm for modeling and solving various important combinatorial optimization problems. Recently, learning-based approaches have shown a potential to speed up MIP solving via offline training that then guides important design decisions during the search. However, a significant drawback of these methods is their heavy reliance on offline training, which requires collecting training datasets and computationally costly training epochs yet offering only limited generalization to unseen (larger) instances. In this paper, we propose Balans, an adaptive meta-solver for MIPs with online learning capability that does not require any supervision or apriori training. At its core, Balans is based on adaptive large-neighborhood search, operating on top of an MIP solver by successive applications of destroy and repair neighborhood operators. During the search, the selection among different neighborhood definitions is guided on the fly for the instance at hand via multi-armed bandit algorithms. Our extensive experiments on hard optimization instances show that Balans offers significant performance gains over the default MIP solver, is better than committing to any single best neighborhood, and improves over the state-of-the-art large-neighborhood search for MIPs. Finally, we release Balans as a highly configurable, MIP solver agnostic, open-source software.
Authors: Maida Wang, Jinyang Jiang, Peter V. Coveney
Abstract: Quantum machine learning has gained attention for its potential to address computational challenges. However, whether those algorithms can effectively solve practical problems and outperform their classical counterparts, especially on current quantum hardware, remains a critical question. In this work, we propose a novel quantum machine learning method, called Parameter-Efficient Quantum Anomaly Detection (PEQAD), for practical image anomaly detection, which aims to achieve both parameter efficiency and superior accuracy compared to classical models. Emulation results indicate that PEQAD demonstrates favourable recognition capabilities compared to classical baselines, achieving an average accuracy of over 90% on benchmarks with significantly fewer trainable parameters. Theoretical analysis confirms that PEQAD has a comparable expressivity to classical counterparts while requiring only a fraction of the parameters. Furthermore, we demonstrate the first implementation of a quantum anomaly detection method for general image datasets on a superconducting quantum processor. Specifically, we achieve an accuracy of over 80% with only 16 parameters on the device, providing initial evidence of PEQAD's practical viability in the noisy intermediate-scale quantum era and highlighting its significant reduction in parameter requirements.
Authors: Shupeng Ning, Hanqing Zhu, Chenghao Feng, Jiaqi Gu, David Z. Pan, Ray T. Chen
Abstract: The rapid growth in computing demands, particularly driven by artificial intelligence applications, has begun to exceed the capabilities of traditional electronic hardware. Optical computing offers a promising alternative due to its parallelism, high computational speed, and low power consumption. However, existing photonic integrated circuits are constrained by large footprints, costly electro-optical interfaces, and complex control mechanisms, limiting the practical scalability of optical neural networks (ONNs). To address these limitations, we introduce a block-circulant photonic tensor core for a structure-compressed optical neural network (StrC-ONN) architecture. The structured compression technique substantially reduces both model complexity and hardware resources without sacrificing the versatility of neural networks, and achieves accuracy comparable to uncompressed models. Additionally, we propose a hardware-aware training framework to compensate for on-chip nonidealities to improve model robustness and accuracy. Experimental validation through image processing and classification tasks demonstrates that our StrC-ONN achieves a reduction in trainable parameters of up to 74.91%,while still maintaining competitive accuracy levels. Performance analyses further indicate that this hardware-software co-design approach is expected to yield a 3.56 times improvement in power efficiency. By reducing both hardware requirements and control complexity across multiple dimensions, this work explores a new pathway toward practical and scalable ONNs, highlighting a promising route to address future computational efficiency challenges.
Authors: Yaroslav Balytskyi, Inna Hubenko, Alina Balytska, Christopher V. Kelly
Abstract: Accurate identification of druggable pockets and their features is essential for structure-based drug design and effective downstream docking. Here, we present RAPID-Net, a deep learning-based algorithm designed for the accurate prediction of binding pockets and seamless integration with docking pipelines. On the PoseBusters benchmark, RAPID-Net-guided AutoDock Vina achieves 54.9% of Top-1 poses with RMSD < 2 A and satisfying the PoseBusters chemical-validity criterion, compared to 49.1% for DiffBindFR. On the most challenging time split of PoseBusters aiming to assess generalization ability (structures submitted after September 30, 2021), RAPID-Net-guided AutoDock Vina achieves 53.1% of Top-1 poses with RMSD < 2 A and PB-valid, versus 59.5% for AlphaFold 3. Notably, in 92.2% of cases, RAPID-Net-guided Vina samples at least one pose with RMSD < 2 A (regardless of its rank), indicating that pose ranking, rather than sampling, is the primary accuracy bottleneck. The lightweight inference, scalability, and competitive accuracy of RAPID-Net position it as a viable option for large-scale virtual screening campaigns. Across diverse benchmark datasets, RAPID-Net outperforms other pocket prediction tools, including PUResNet and Kalasanty, in both docking accuracy and pocket-ligand intersection rates. Furthermore, we demonstrate the potential of RAPID-Net to accelerate the development of novel therapeutics by highlighting its performance on pharmacologically relevant targets. RAPID-Net accurately identifies distal functional sites, offering new opportunities for allosteric inhibitor design. In the case of the RNA-dependent RNA polymerase of SARS-CoV-2, RAPID-Net uncovers a wider array of potential binding pockets than existing predictors, which typically annotate only the orthosteric pocket and overlook secondary cavities.
Authors: Shehzeen Hussain, Paarth Neekhara, Xuesong Yang, Edresson Casanova, Subhankar Ghosh, Mikyas T. Desta, Roy Fejgin, Rafael Valle, Jason Li
Abstract: While autoregressive speech token generation models produce speech with remarkable variety and naturalness, their inherent lack of controllability often results in issues such as hallucinations and undesired vocalizations that do not conform to conditioning inputs. We introduce Koel-TTS, a suite of enhanced encoder-decoder Transformer TTS models that address these challenges by incorporating preference alignment techniques guided by automatic speech recognition and speaker verification models. Additionally, we incorporate classifier-free guidance to further improve synthesis adherence to the transcript and reference speaker audio. Our experiments demonstrate that these optimizations significantly enhance target speaker similarity, intelligibility, and naturalness of synthesized speech. Notably, Koel-TTS directly maps text and context audio to acoustic tokens, and on the aforementioned metrics, outperforms state-of-the-art TTS models, despite being trained on a significantly smaller dataset. Audio samples and demos are available on our website.
Authors: Songming Zhang, Xue Zhang, Tong Zhang, Bojie Hu, Yufeng Chen, Jinan Xu
Abstract: In modern large language models (LLMs), LLM alignment is of crucial importance and is typically achieved through methods such as reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO). However, in most existing methods for LLM alignment, all tokens in the response are optimized using a sparse, response-level reward or preference annotation. The ignorance of token-level rewards may erroneously punish high-quality tokens or encourage low-quality tokens, resulting in suboptimal performance and slow convergence speed. To address this issue, we propose AlignDistil, an RLHF-equivalent distillation method for token-level reward optimization. Specifically, we introduce the reward learned by DPO into the RLHF objective and theoretically prove the equivalence between this objective and a token-level distillation process, where the teacher distribution linearly combines the logits from the DPO model and a reference model. On this basis, we further bridge the accuracy gap between the reward from the DPO model and the pure reward model, by building a contrastive DPO reward with a normal and a reverse DPO model. Moreover, to avoid under- and over-optimization on different tokens, we design a token adaptive logit extrapolation mechanism to construct an appropriate teacher distribution for each token. Experimental results demonstrate the superiority of our AlignDistil over existing methods and showcase fast convergence due to its token-level distributional reward optimization.
Authors: Pranshav Gajjar, Vijay K. Shah
Abstract: Despite the transformative impact of Large Language Models (LLMs) across critical domains such as healthcare, customer service, and business marketing, their integration into Open Radio Access Networks (O-RAN) remains limited. This gap is primarily due to the absence of domain-specific foundational models, with existing solutions often relying on general-purpose LLMs that fail to address the unique challenges and technical intricacies of O-RAN. To bridge this gap, we introduce ORANSight-2.0 (O-RAN Insights), a pioneering initiative to develop specialized foundational LLMs tailored for O-RAN. Built on 18 models spanning five open-source LLM frameworks -- Mistral, Qwen, Llama, Phi, and Gemma -- ORANSight-2.0 fine-tunes models ranging from 1B to 70B parameters, significantly reducing reliance on proprietary, closed-source models while enhancing performance in O-RAN-specific tasks. At the core of ORANSight-2.0 is RANSTRUCT, a novel Retrieval-Augmented Generation (RAG)-based instruction-tuning framework that employs two LLM agents -- a Mistral-based Question Generator and a Qwen-based Answer Generator -- to create high-quality instruction-tuning datasets. The generated dataset is then used to fine-tune the 18 pre-trained open-source LLMs via QLoRA. To evaluate ORANSight-2.0, we introduce srsRANBench, a novel benchmark designed for code generation and codebase understanding in the context of srsRAN, a widely used 5G O-RAN stack.
Authors: Amir Mohammad Izadi, Seyed Mohammad Hadi Hosseini, Soroush Vafaie Tabar, Ali Abdollahi, Armin Saghafian, Mahdieh Soleymani Baghshah
Abstract: Text-to-image generative models have made significant advancements in recent years; however, accurately capturing intricate details in textual prompts-such as entity missing, attribute binding errors, and incorrect relationships remains a formidable challenge. In response, we present an innovative, training-free method that directly addresses these challenges by incorporating tailored objectives to account for textual constraints. Unlike layout-based approaches that enforce rigid structures and limit diversity, our proposed approach offers a more flexible arrangement of the scene by imposing just the extracted constraints from the text, without any unnecessary additions. These constraints are formulated as losses-entity missing, entity mixing, attribute binding, and spatial relationships-integrated into a unified loss that is applied in the first generation stage. Furthermore, we introduce a feedback-driven system for fine-grained initial noise refinement. This system integrates a verifier that evaluates the generated image, identifies inconsistencies, and provides corrective feedback. Leveraging this feedback, our refinement method first targets the unmet constraints by refining the faulty attention maps caused by initial noise, through the optimization of selective losses associated with these constraints. Subsequently, our unified loss function is reapplied to proceed the second generation phase. Experimental results demonstrate that our method, relying solely on our proposed objective functions, significantly enhances compositionality, achieving a 24% improvement in human evaluation and a 25% gain in spatial relationships. Furthermore, our fine-grained noise refinement proves effective, boosting performance by up to 5%. Code is available at \href{https://github.com/hadi-hosseini/noise-refinement}{https://github.com/hadi-hosseini/noise-refinement}.
URLs: https://github.com/hadi-hosseini/noise-refinement, https://github.com/hadi-hosseini/noise-refinement
Authors: Elyas Meguellati, Stefano Civelli, Pietro Bernardelle, Shazia Sadiq, Irwin King, Gianluca Demartini
Abstract: Political advertising plays a pivotal role in shaping public opinion and influencing electoral outcomes, often through subtle persuasive techniques embedded in broader propaganda strategies. Detecting these persuasive elements is crucial for enhancing voter awareness and ensuring transparency in democratic processes. This paper presents an integrated approach that bridges model development and real-world application through two interconnected studies. First, we introduce a lightweight model for persuasive text detection that achieves state-of-the-art performance in Subtask 3 of SemEval 2023 Task 3 while requiring significantly fewer computational resources and training data than existing methods. Second, we demonstrate the model's practical utility by collecting the Australian Federal Election 2022 Facebook Ads (APA22) dataset, partially annotating a subset for persuasion, and fine-tuning the model to adapt from mainstream news to social media content. We then apply the fine-tuned model to label the remainder of the APA22 dataset, revealing distinct patterns in how political campaigns leverage persuasion through different funding strategies, word choices, demographic targeting, and temporal shifts in persuasion intensity as election day approaches. Our findings not only underscore the necessity of domain-specific modeling for analyzing persuasion on social media but also show how uncovering these strategies can enhance transparency, inform voters, and promote accountability in digital campaigns.
Authors: Ekaterina Dmitrieva, Maksim Kaledin
Abstract: Speech Enhancement techniques have become core technologies in mobile devices and voice software. Still, modern deep learning solutions often require high amount of computational resources what makes their usage on low-resource devices challenging. We present HiFi-Stream, an optimized version of recently published HiFi++ model. Our experiments demonstrate that HiFi-Stream saves most of the qualities of the original model despite its size and computational complexity improved in comparison to the original HiFi++ making it one of the smallest and fastest models available. The model is evaluated in streaming setting where it demonstrates its superior performance in comparison to modern baselines.
Authors: Yu Inatsu
Abstract: Bayesian optimization based on the Gaussian process upper confidence bound (GP-UCB) offers a theoretical guarantee for optimizing black-box functions. In practice, however, black-box functions often involve input uncertainty. To handle such cases, GP-UCB can be extended to optimize evaluation criteria known as robustness measures. However, GP-UCB-based methods for robustness measures require a trade-off parameter, $\beta$, which, as in the original GP-UCB, must be set sufficiently large to ensure theoretical validity. In this study, we propose randomized robustness measure GP-UCB (RRGP-UCB), a novel method that samples $\beta$ from a chi-squared-based probability distribution. This approach eliminates the need to explicitly specify $\beta$. Notably, the expected value of $\beta$ under this distribution is not excessively large. Furthermore, we show that RRGP-UCB provides tight bounds on the expected regret between the optimal and estimated solutions. Numerical experiments demonstrate the effectiveness of the proposed method.
Authors: Minshuo Chen, Renyuan Xu, Yumin Xu, Ruixun Zhang
Abstract: Financial scenario simulation is essential for risk management and portfolio optimization, yet it remains challenging especially in high-dimensional and small data settings common in finance. We propose a diffusion factor model that integrates latent factor structure into generative diffusion processes, bridging econometrics with modern generative AI to address the challenges of the curse of dimensionality and data scarcity in financial simulation. By exploiting the low-dimensional factor structure inherent in asset returns, we decompose the score function--a key component in diffusion models--using time-varying orthogonal projections, and this decomposition is incorporated into the design of neural network architectures. We derive rigorous statistical guarantees, establishing nonasymptotic error bounds for both score estimation at O(d^{5/2} n^{-2/(k+5)}) and generated distribution at O(d^{5/4} n^{-1/2(k+5)}), primarily driven by the intrinsic factor dimension k rather than the number of assets d, surpassing the dimension-dependent limits in the classical nonparametric statistics literature and making the framework viable for markets with thousands of assets. Numerical studies confirm superior performance in latent subspace recovery under small data regimes. Empirical analysis demonstrates the economic significance of our framework in constructing mean-variance optimal portfolios and factor portfolios. This work presents the first theoretical integration of factor structure with diffusion models, offering a principled approach for high-dimensional financial simulation with limited data. Our code is available at https://github.com/xymmmm00/diffusion_factor_model.
Authors: Ioannis Kontogiorgakis, Iason Tsardanidis, Dimitrios Bormpoudakis, Ilias Tsoumas, Dimitra A. Loka, Christos Noulas, Alexandros Tsitouras, Charalampos Kontoes
Abstract: Effective weed management is crucial for improving agricultural productivity, as weeds compete with crops for vital resources like nutrients and water. Accurate maps of weed management methods are essential for policymakers to assess farmer practices, evaluate impacts on vegetation health, biodiversity, and climate, as well as ensure compliance with policies and subsidies. However, monitoring weed management methods is challenging as they commonly rely on ground-based field surveys, which are often costly, time-consuming and subject to delays. In order to tackle this problem, we leverage earth observation data and Machine Learning (ML). Specifically, we developed separate ML models using Sentinel-2 and PlanetScope satellite time series data, respectively, to classify four distinct weed management methods (Mowing, Tillage, Chemical-spraying, and No practice) in orchards. The findings demonstrate the potential of ML-driven remote sensing to enhance the efficiency and accuracy of weed management mapping in orchards.
Authors: Andreas Kalogeras, Dimitrios Bormpoudakis, Iason Tsardanidis, Dimitra A. Loka, Charalampos Kontoes
Abstract: The widespread use of Exogenous Organic Matter in agriculture necessitates monitoring to assess its effects on soil and crop health. This study evaluates optical Sentinel-2 satellite imagery for detecting digestate application, a practice that enhances soil fertility but poses environmental risks like microplastic contamination and nitrogen losses. In the first instance, Sentinel-2 satellite image time series (SITS) analysis of specific indices (EOMI, NDVI, EVI) was used to characterize EOM's spectral behavior after application on the soils of four different crop types in Thessaly, Greece. Furthermore, Machine Learning (ML) models (namely Random Forest, k-NN, Gradient Boosting and a Feed-Forward Neural Network), were used to investigate digestate presence detection, achieving F1-scores up to 0.85. The findings highlight the potential of combining remote sensing and ML for scalable and cost-effective monitoring of EOM applications, supporting precision agriculture and sustainability.
Authors: Arun M. Saranathan, Mahmoud Saeedimoghaddam, Brandon Smith, Deepthi Raghunandan, Grey Nearing, Craig Pelissier
Abstract: Snow is an essential input for various land surface models. Seasonal snow estimates are available as snow water equivalent (SWE) from process-based reanalysis products or locally from in situ measurements. While the reanalysis products are computationally expensive and available at only fixed spatial and temporal resolutions, the in situ measurements are highly localized and sparse. To address these issues and enable the analysis of the effect of a large suite of physical, morphological, and geological conditions on the presence and amount of snow, we build a Long Short-Term Memory (LSTM) network, which is able to estimate the SWE based on time series input of the various physical/meteorological factors as well static spatial/morphological factors. Specifically, this model breaks down the SWE estimation into two separate tasks: (i) a classification task that indicates the presence/absence of snow on a specific day and (ii) a regression task that indicates the height of the SWE on a specific day in the case of snow presence. The model is trained using physical/in situ SWE measurements from the SNOw TELemetry (SNOTEL) snow pillows in the western United States. We will show that trained LSTM models have a classification accuracy of $\geq 93\%$ for the presence of snow and a coefficient of correlation of $\sim 0.9$ concerning their SWE estimates. We will also demonstrate that the models can generalize both spatially and temporally to previously unseen data.
Authors: Eric Tillmann Bill, Enis Simsar, Thomas Hofmann
Abstract: We introduce JEDI, a test-time adaptation method that enhances subject separation and compositional alignment in diffusion models without requiring retraining or external supervision. JEDI operates by minimizing semantic entanglement in attention maps using a novel Jensen-Shannon divergence based objective. To improve efficiency, we leverage adversarial optimization, reducing the number of updating steps required. JEDI is model-agnostic and applicable to architectures such as Stable Diffusion 1.5 and 3.5, consistently improving prompt alignment and disentanglement in complex scenes. Additionally, JEDI provides a lightweight, CLIP-free disentanglement score derived from internal attention distributions, offering a principled benchmark for compositional alignment under test-time conditions. Code and results are available at https://ericbill21.github.io/JEDI/.
Authors: Daniel Barta, Darya Martyniuk, Johannes Jung, Adrian Paschke
Abstract: Quantum computing holds immense potential, yet its practical success depends on multiple factors, including advances in quantum circuit design. In this paper, we introduce a generative approach based on denoising diffusion models (DMs) to synthesize parameterized quantum circuits (PQCs). Extending the recent diffusion model pipeline of F\"urrutter et al. [1], our model effectively conditions the synthesis process, enabling the simultaneous generation of circuit architectures and their continuous gate parameters. We demonstrate our approach in synthesizing PQCs optimized for generating high-fidelity Greenberger-Horne-Zeilinger (GHZ) states and achieving high accuracy in quantum machine learning (QML) classification tasks. Our results indicate a strong generalization across varying gate sets and scaling qubit counts, highlighting the versatility and computational efficiency of diffusion-based methods. This work illustrates the potential of generative models as a powerful tool for accelerating and optimizing the design of PQCs, supporting the development of more practical and scalable quantum applications.
Authors: Xiaoyan Li, Shixin Xu, Faisal Habib, Arvind Gupta, Huaxiong Huang
Abstract: Reconstructing ECG from PPG is a promising yet challenging task. While recent advancements in generative models have significantly improved ECG reconstruction, accurately capturing fine-grained waveform features remains a key challenge. To address this, we propose a novel PPG-to-ECG reconstruction method that leverages a Vision Transformer (ViT) as the core network. Unlike conventional approaches that rely on single-channel PPG, our method employs a four-channel signal image representation, incorporating the original PPG, its first-order difference, second-order difference, and area under the curve. This multi-channel design enriches feature extraction by preserving both temporal and physiological variations within the PPG. By leveraging the self-attention mechanism in ViT, our approach effectively captures both inter-beat and intra-beat dependencies, leading to more robust and accurate ECG reconstruction. Experimental results demonstrate that our method consistently outperforms existing 1D convolution-based approaches, achieving up to 29% reduction in PRD and 15% reduction in RMSE. The proposed approach also produces improvements in other evaluation metrics, highlighting its robustness and effectiveness in reconstructing ECG signals. Furthermore, to ensure a clinically relevant evaluation, we introduce new performance metrics, including QRS area error, PR interval error, RT interval error, and RT amplitude difference error. Our findings suggest that integrating a four-channel signal image representation with the self-attention mechanism of ViT enables more effective extraction of informative PPG features and improved modeling of beat-to-beat variations for PPG-to-ECG mapping. Beyond demonstrating the potential of PPG as a viable alternative for heart activity monitoring, our approach opens new avenues for cyclic signal analysis and prediction.
Authors: Lai Wei, Yuting Li, Kaipeng Zheng, Chen Wang, Yue Wang, Linghe Kong, Lichao Sun, Weiran Huang
Abstract: Recent advancements in large language models (LLMs) have demonstrated impressive chain-of-thought reasoning capabilities, with reinforcement learning (RL) playing a crucial role in this progress. While "aha moment" patterns--where models exhibit self-correction through reflection--are often attributed to emergent properties from RL, we first demonstrate that these patterns exist in multimodal LLMs (MLLMs) prior to RL training but may not necessarily correlate with improved reasoning performance. Building on these insights, we present a comprehensive study on enhancing multimodal reasoning through a two-stage approach: (1) supervised fine-tuning (SFT) as a cold start with structured chain-of-thought reasoning patterns, followed by (2) reinforcement learning via GRPO to further refine these capabilities. Our extensive experiments show that this combined approach consistently outperforms both SFT-only and RL-only methods across challenging multimodal reasoning benchmarks. The resulting models achieve state-of-the-art performance among open-source MLLMs at both 3B and 7B scales, with our 7B model showing substantial improvements over base models (e.g., 66.3 %$\rightarrow$73.4 % on MathVista, 62.9 %$\rightarrow$70.4 % on We-Math) and our 3B model achieving performance competitive with several 7B models. Overall, this work provides practical guidance for building advanced multimodal reasoning models. Our code is available at https://github.com/waltonfuture/RL-with-Cold-Start.
Authors: Yang Wang, Yin Xu, Cixiao Zhang, Zhiyong Chen, Mingzeng Dai, Haiming Wang, Bingchao Liu, Dazhi He, Meixia Tao
Abstract: Reconfigurable intelligent surface (RIS) has been recognized as a promising technology for next-generation wireless communications. However, the performance of RIS-assisted systems critically depends on accurate channel state information (CSI). To address this challenge, this letter proposes a novel channel estimation method for RIS-aided millimeter-wave (mmWave) systems based on diffusion models (DMs). Specifically, the forward diffusion process of the original signal is formulated to model the received signal as a noisy observation within the framework of DMs. Subsequently, the channel estimation task is formulated as the reverse diffusion process, and a sampling algorithm based on denoising diffusion implicit models (DDIMs) is developed to enable effective inference. Furthermore, a lightweight neural network, termed BRCNet, is introduced to replace the conventional U-Net, significantly reducing the number of parameters and computational complexity. Extensive experiments conducted under various scenarios demonstrate that the proposed method consistently outperforms existing baselines.
Authors: Haoyu Dong, Yuwen Chen, Hanxue Gu, Nicholas Konz, Yaqian Chen, Qihang Li, Maciej A. Mazurowski
Abstract: The widespread use of Magnetic Resonance Imaging (MRI) in combination with deep learning shows promise for many high-impact automated diagnostic and prognostic tools. However, training new models requires large amounts of labeled data, a challenge due to high cost of precise annotations and data privacy. To address this issue, we introduce the MRI-CORE, a vision foundation model trained using more than 6 million slices from over 110 thousand MRI volumes across 18 body locations. Our experiments show notable improvements in performance over state-of-the-art methods in 13 data-restricted segmentation tasks, as well as in image classification, and zero-shot segmentation, showing the strong potential of MRI-CORE to enable data-efficient development of artificial intelligence models. We also present data on which strategies yield most useful foundation models and a novel analysis relating similarity between pre-training and downstream task data with transfer learning performance. Our model is publicly available with a permissive license.
Authors: Xin Chen, Jason M. Klusowski, Yan Shuo Tan, Chang Yu
Abstract: Combining randomized estimators in an ensemble, such as via random forests, has become a fundamental technique in modern data science, but can be computationally expensive. Furthermore, the mechanism by which this improves predictive performance is poorly understood. We address these issues in the context of sparse linear regression by proposing and analyzing an ensemble of greedy forward selection estimators that are randomized by feature subsampling -- at each iteration, the best feature is selected from within a random subset. We design a novel implementation based on dynamic programming that greatly improves its computational efficiency. Furthermore, we show via careful numerical experiments that our method can outperform popular methods such as lasso and elastic net across a wide range of settings. Next, contrary to prevailing belief that randomized ensembling is analogous to shrinkage, we show via numerical experiments that it can simultaneously reduce training error and degrees of freedom, thereby shifting the entire bias-variance trade-off curve of the base estimator. We prove this fact rigorously in the setting of orthogonal features, in which case, the ensemble estimator rescales the ordinary least squares coefficients with a two-parameter family of logistic weights, thereby enlarging the model search space. These results enhance our understanding of random forests and suggest that implicit regularization in general may have more complicated effects than explicit regularization.
Authors: Xiaosheng Zhao, Yang Huang, Guirong Xue, Xiao Kong, Jifeng Liu, Xiaoyu Tang, Timothy C. Beers, Yuan-Sen Ting, A-Li Luo
Abstract: In recent years, large language models (LLMs) have transformed natural language understanding through vast datasets and large-scale parameterization. Inspired by this success, we present SpecCLIP, a foundation model framework that extends LLM-inspired methodologies to stellar spectral analysis. Stellar spectra, akin to structured language, encode rich physical and chemical information about stars. By training foundation models on large-scale spectral datasets, our goal is to learn robust and informative embeddings that support diverse downstream applications. As a proof of concept, SpecCLIP involves pre-training on two spectral types--LAMOST low-resolution and Gaia XP--followed by contrastive alignment using the CLIP (Contrastive Language-Image Pre-training) framework, adapted to associate spectra from different instruments. This alignment is complemented by auxiliary decoders that preserve spectrum-specific information and enable translation (prediction) between spectral types, with the former achieved by maximizing mutual information between embeddings and input spectra. The result is a cross-spectrum framework enabling intrinsic calibration and flexible applications across instruments. We demonstrate that fine-tuning these models on moderate-sized labeled datasets improves adaptability to tasks such as stellar-parameter estimation and chemical-abundance determination. SpecCLIP also enhances the accuracy and precision of parameter estimates benchmarked against external survey data. Additionally, its similarity search and cross-spectrum prediction capabilities offer potential for anomaly detection. Our results suggest that contrastively trained foundation models enriched with spectrum-aware decoders can advance precision stellar spectroscopy.
Authors: Rahul Ramachandran, Ali Garjani, Roman Bachmann, Andrei Atanov, O\u{g}uzhan Fatih Kar, Amir Zamir
Abstract: Multimodal foundation models, such as GPT-4o, have recently made remarkable progress, but it is not clear where exactly these models stand in terms of understanding vision. In this paper, we benchmark the performance of popular multimodal foundation models (GPT-4o, o4-mini, Gemini 1.5 Pro and Gemini 2.0 Flash, Claude 3.5 Sonnet, Qwen2-VL, Llama 3.2) on standard computer vision tasks (semantic segmentation, object detection, image classification, depth and surface normal prediction) using established datasets (e.g., COCO, ImageNet and its variants, etc). The main challenges to performing this are: 1) most models are trained to output text and cannot natively express versatile domains, such as segments or 3D geometry, and 2) many leading models are proprietary and accessible only at an API level, i.e., there is no weight access to adapt them. We address these challenges by translating standard vision tasks into equivalent text-promptable and API-compatible tasks via prompt chaining to create a standardized benchmarking framework. We observe that 1) the models are not close to the state-of-the-art specialist models at any task. However, 2) they are respectable generalists; this is remarkable as they are presumably trained on primarily image-text-based tasks. 3) They perform semantic tasks notably better than geometric ones. 4) While the prompt-chaining techniques affect performance, better models exhibit less sensitivity to prompt variations. 5) GPT-4o performs the best among non-reasoning models, securing the top position in 4 out of 6 tasks, 6) reasoning models, e.g. o3, show improvements in geometric tasks, and 7) a preliminary analysis of models with native image generation, like the latest GPT-4o, shows they exhibit quirks like hallucinations and spatial misalignments.
Authors: Yizhou Wang, Lingzhi Zhang, Yue Bai, Mang Tik Chiu, Zhengmian Hu, Mingyuan Zhang, Qihua Dong, Yu Yin, Sohrab Amirghodsi, Yun Fu
Abstract: Next token prediction paradigm has been prevailing for autoregressive models in the era of LLMs. The current default sampling choice for popular LLMs is temperature scaling together with nucleus sampling to balance diversity and coherence. Nevertheless, such approach leads to inferior performance in various NLP tasks when the model is not certain about testing questions. To this end, we propose a brand new training-free decoding strategy, dubbed as Cautious Next Token Prediction (CNTP). In the decoding process, if the model has comparatively high prediction entropy at a certain step, we sample multiple trials starting from the step independently and stop when encountering any punctuation. Then we select the trial with the lowest perplexity score viewed as the most probable and reliable trial path given the model's capacity. The trial number is negatively correlated with the prediction confidence, i.e., the less confident the model is, the more trials it should sample. This is consistent with human beings' behaviour: when feeling uncertain or unconfident, one tends to think more creatively, exploring multiple thinking paths, to cautiously select the path one feels most confident about. Extensive experiments on both LLMs and MLLMs show that our proposed CNTP approach outperforms existing standard decoding strategies consistently by a clear margin. Moreover, the integration of CNTP with self consistency can further improve over vanilla self consistency. We believe our proposed CNTP has the potential to become one of the default choices for LLM decoding. Code is available at https://github.com/wyzjack/CNTP.
Authors: Juan B. Guti\'errez
Abstract: Large language models (LLMs) turn writing into a live exchange between humans and software. We characterize this new medium as a discursive network that treats people and LLMs as equal nodes and tracks how their statements circulate. We define the generation of erroneous information as invalidation (any factual, logical, or structural breach) and show it follows four hazards: drift from truth, self-repair, fresh fabrication, and external detection. We develop a general mathematical model of discursive networks that shows that a network governed only by drift and self-repair stabilizes at a modest error rate. Giving each false claim even a small chance of peer review shifts the system to a truth-dominant state. We operationalize peer review with the open-source Flaws-of-Others (FOO) algorithm: a configurable loop in which any set of agents critique one another while a harmonizer merges their verdicts. We identify an ethical transgression, epithesis, that occurs when humans fail to engage in the discursive network. The takeaway is practical and cultural: reliability in this new medium comes not from perfecting single models but from connecting imperfect ones into networks that enforce mutual accountability.
Authors: Dell Zhang, Xiangyu Chen, Jixiang Luo, Mengxi Jia, Changzhi Sun, Ruilong Ren, Jingren Liu, Hao Sun, Xuelong Li
Abstract: The rapid advancements in Large Language Models (LLMs) and their multimodal extensions (MLLMs) have ushered in remarkable progress in video understanding. However, a fundamental challenge persists: effectively processing and comprehending video content that extends beyond minutes or hours. While recent efforts like Video-XL-2 have demonstrated novel architectural solutions for extreme efficiency, and advancements in positional encoding such as HoPE and VideoRoPE++ aim to improve spatio-temporal understanding over extensive contexts, current state-of-the-art models still encounter significant computational and memory constraints when faced with the sheer volume of visual tokens from lengthy sequences. Furthermore, maintaining temporal coherence, tracking complex events, and preserving fine-grained details over extended periods remain formidable hurdles, despite progress in agentic reasoning systems like Deep Video Discovery. This position paper posits that a logical, albeit ambitious, next frontier for multimedia research is Infinite Video Understanding -- the capability for models to continuously process, understand, and reason about video data of arbitrary, potentially never-ending duration. We argue that framing Infinite Video Understanding as a blue-sky research objective provides a vital north star for the multimedia, and the wider AI, research communities, driving innovation in areas such as streaming architectures, persistent memory mechanisms, hierarchical and adaptive representations, event-centric reasoning, and novel evaluation paradigms. Drawing inspiration from recent work on long/ultra-long video understanding and several closely related fields, we outline the core challenges and key research directions towards achieving this transformative capability.
Authors: Alireza Golkarieh, Kiana Kiashemshaki, Sajjad Rezvani Boroujeni, Nasibeh Asadi Isakan
Abstract: This study investigates the effectiveness of U-Net architectures integrated with various convolutional neural network (CNN) backbones for automated lung cancer detection and segmentation in chest CT images, addressing the critical need for accurate diagnostic tools in clinical settings. A balanced dataset of 832 chest CT images (416 cancerous and 416 non-cancerous) was preprocessed using Contrast Limited Adaptive Histogram Equalization (CLAHE) and resized to 128x128 pixels. U-Net models were developed with three CNN backbones: ResNet50, VGG16, and Xception, to segment lung regions. After segmentation, CNN-based classifiers and hybrid models combining CNN feature extraction with traditional machine learning classifiers (Support Vector Machine, Random Forest, and Gradient Boosting) were evaluated using 5-fold cross-validation. Metrics included accuracy, precision, recall, F1-score, Dice coefficient, and ROC-AUC. U-Net with ResNet50 achieved the best performance for cancerous lungs (Dice: 0.9495, Accuracy: 0.9735), while U-Net with VGG16 performed best for non-cancerous segmentation (Dice: 0.9532, Accuracy: 0.9513). For classification, the CNN model using U-Net with Xception achieved 99.1 percent accuracy, 99.74 percent recall, and 99.42 percent F1-score. The hybrid CNN-SVM-Xception model achieved 96.7 percent accuracy and 97.88 percent F1-score. Compared to prior methods, our framework consistently outperformed existing models. In conclusion, combining U-Net with advanced CNN backbones provides a powerful method for both segmentation and classification of lung cancer in CT scans, supporting early diagnosis and clinical decision-making.
Authors: Suyuan Zhao, Yizhen Luo, Ganbo Yang, Yan Zhong, Hao Zhou, Zaiqing Nie
Abstract: Spatial Transcriptomics (ST) technologies provide biologists with rich insights into single-cell biology by preserving spatial context of cells. Building foundational models for ST can significantly enhance the analysis of vast and complex data sources, unlocking new perspectives on the intricacies of biological tissues. However, modeling ST data is inherently challenging due to the need to extract multi-scale information from tissue slices containing vast numbers of cells. This process requires integrating macro-scale tissue morphology, micro-scale cellular microenvironment, and gene-scale gene expression profile. To address this challenge, we propose SToFM, a multi-scale Spatial Transcriptomics Foundation Model. SToFM first performs multi-scale information extraction on each ST slice, to construct a set of ST sub-slices that aggregate macro-, micro- and gene-scale information. Then an SE(2) Transformer is used to obtain high-quality cell representations from the sub-slices. Additionally, we construct \textbf{SToCorpus-88M}, the largest high-resolution spatial transcriptomics corpus for pretraining. SToFM achieves outstanding performance on a variety of downstream tasks, such as tissue region semantic segmentation and cell type annotation, demonstrating its comprehensive understanding of ST data through capturing and integrating multi-scale information.
Authors: Binh Huy Nguyen, Matti Schneider
Abstract: Solving cell problems in homogenization is hard, and available deep-learning frameworks fail to match the speed and generality of traditional computational frameworks. More to the point, it is generally unclear what to expect of machine-learning approaches, let alone single out which approaches are promising. In the work at hand, we advocate Fourier Neural Operators (FNOs) for micromechanics, empowering them by insights from computational micromechanics methods based on the fast Fourier transform (FFT). We construct an FNO surrogate mimicking the basic scheme foundational for FFT-based methods and show that the resulting operator predicts solutions to cell problems with arbitrary stiffness distribution only subject to a material-contrast constraint up to a desired accuracy. In particular, there are no restrictions on the material symmetry like isotropy, on the number of phases and on the geometry of the interfaces between materials. Also, the provided fidelity is sharp and uniform, providing explicit guarantees leveraging our physical empowerment of FNOs. To show the desired universal approximation property, we construct an FNO explicitly that requires no training to begin with. Still, the obtained neural operator complies with the same memory requirements as the basic scheme and comes with runtimes proportional to classical FFT solvers. In particular, large-scale problems with more than 100 million voxels are readily handled. The goal of this work is to underline the potential of FNOs for solving micromechanical problems, linking FFT-based methods to FNOs. This connection is expected to provide a fruitful exchange between both worlds.
Authors: ZhengXiao He, Jinghao Wen, Huayu Li, Siyuan Tian, Ao Li
Abstract: We present a novel and interpretable framework for electrocardiogram (ECG)-based disease detection that combines hyperdimensional computing (HDC) with learnable neural encoding. Unlike conventional HDC approaches that rely on static, random projections, our method introduces a rhythm-aware and trainable encoding pipeline based on RR intervals, a physiological signal segmentation strategy that aligns with cardiac cycles. The core of our design is a neural-distilled HDC architecture, featuring a learnable RR-block encoder and a BinaryLinear hyperdimensional projection layer, optimized jointly with cross-entropy and proxy-based metric loss. This hybrid framework preserves the symbolic interpretability of HDC while enabling task-adaptive representation learning. Experiments on Apnea-ECG and PTB-XL demonstrate that our model significantly outperforms traditional HDC and classical ML baselines, achieving 73.09\% precision and an F1 score of 0.626 on Apnea-ECG, with comparable robustness on PTB-XL. Our framework offers an efficient and scalable solution for edge-compatible ECG classification, with strong potential for interpretable and personalized health monitoring.
Authors: Ravin Kumar
Abstract: We propose the APTx Neuron, a novel, unified neural computation unit that integrates non-linear activation and linear transformation into a single trainable expression. The APTx Neuron is derived from the APTx activation function, thereby eliminating the need for separate activation layers and making the architecture both computationally efficient and elegant. The proposed neuron follows the functional form $y = \sum_{i=1}^{n} ((\alpha_i + \tanh(\beta_i x_i)) \cdot \gamma_i x_i) + \delta$, where all parameters $\alpha_i$, $\beta_i$, $\gamma_i$, and $\delta$ are trainable. We validate our APTx Neuron-based architecture on the MNIST dataset, achieving up to 96.69% test accuracy in just 20 epochs using approximately 332K trainable parameters. The results highlight the superior expressiveness and computational efficiency of the APTx Neuron compared to traditional neurons, pointing toward a new paradigm in unified neuron design and the architectures built upon it.
Authors: Mehdi Khalaj, Shahrzad Golestani Najafabadi, Julita Vassileva
Abstract: Personalized News Recommendation systems (PNR) have emerged as a solution to information overload by predicting and suggesting news items tailored to individual user interests. However, traditional PNR systems face several challenges, including an overreliance on textual content, common neglect of short-term user interests, and significant privacy concerns due to centralized data storage. This paper addresses these issues by introducing a novel multimodal federated learning-based approach for news recommendation. First, it integrates both textual and visual features of news items using a multimodal model, enabling a more comprehensive representation of content. Second, it employs a time-aware model that balances users' long-term and short-term interests through multi-head self-attention networks, improving recommendation accuracy. Finally, to enhance privacy, a federated learning framework is implemented, enabling collaborative model training without sharing user data. The framework divides the recommendation model into a large server-maintained news model and a lightweight user model shared between the server and clients. The client requests news representations (vectors) and a user model from the central server, then computes gradients with user local data, and finally sends their locally computed gradients to the server for aggregation. The central server aggregates gradients to update the global user model and news model. The updated news model is further used to infer news representation by the server. To further safeguard user privacy, a secure aggregation algorithm based on Shamir's secret sharing is employed. Experiments on a real-world news dataset demonstrate strong performance compared to existing systems, representing a significant advancement in privacy-preserving personalized news recommendation.
Authors: Peng Chen, Hailiang Zhao, Jiaji Zhang, Xueyan Tang, Yixuan Wang, Shuiguang Deng
Abstract: The online caching problem aims to minimize cache misses when serving a sequence of requests under a limited cache size. While naive learning-augmented caching algorithms achieve ideal $1$-consistency, they lack robustness guarantees. Existing robustification methods either sacrifice $1$-consistency or introduce significant computational overhead. In this paper, we introduce Guard, a lightweight robustification framework that enhances the robustness of a broad class of learning-augmented caching algorithms to $2H_k + 2$, while preserving their $1$-consistency. Guard achieves the current best-known trade-off between consistency and robustness, with only $O(1)$ additional per-request overhead, thereby maintaining the original time complexity of the base algorithm. Extensive experiments across multiple real-world datasets and prediction models validate the effectiveness of Guard in practice.
Authors: Yanjun Zheng, Xiyang Du, Longfei Liao, Xiaoke Zhao, Zhaowen Zhou, Jingze Song, Bo Zhang, Jiawei Liu, Xiang Qi, Zhe Li, Zhiqiang Zhang, Wei Wang, Peng Zhang
Abstract: Large Language Models (LLMs) exhibit considerable promise in financial applications; however, prevailing models frequently demonstrate limitations when confronted with scenarios that necessitate sophisticated reasoning capabilities, stringent trustworthiness criteria, and efficient adaptation to domain-specific requirements. We introduce the Agentar-Fin-R1 series of financial large language models (8B and 32B parameters), specifically engineered based on the Qwen3 foundation model to enhance reasoning capabilities, reliability, and domain specialization for financial applications. Our optimization approach integrates a high-quality, systematic financial task label system with a comprehensive multi-layered trustworthiness assurance framework. This framework encompasses high-quality trustworthy knowledge engineering, multi-agent trustworthy data synthesis, and rigorous data validation governance. Through label-guided automated difficulty-aware optimization, tow-stage training pipeline, and dynamic attribution systems, we achieve substantial improvements in training efficiency. Our models undergo comprehensive evaluation on mainstream financial benchmarks including Fineva, FinEval, and FinanceIQ, as well as general reasoning datasets such as MATH-500 and GPQA-diamond. To thoroughly assess real-world deployment capabilities, we innovatively propose the Finova evaluation benchmark, which focuses on agent-level financial reasoning and compliance verification. Experimental results demonstrate that Agentar-Fin-R1 not only achieves state-of-the-art performance on financial tasks but also exhibits exceptional general reasoning capabilities, validating its effectiveness as a trustworthy solution for high-stakes financial applications. The Finova bench is available at https://github.com/antgroup/Finova.