arXiv subjects

Yu Wang

Publications and source records attributed to Yu Wang.

At least 487 records · Page 27Linked to original sources

arXiv2025

Examining False Positives under Inference Scaling for Mathematical Reasoning

Recent advancements in language models have led to significant improvements in mathematical reasoning across various benchmarks. However, most of these benchmarks rely on automatic evaluation methods that only compare final answers using heuristics, without verifying the underlying reasoning steps. This limitation results in false positive solutions, where models may produce correct final answers but with flawed deduction paths. In this paper, we systematically examine the prevalence of false positive solutions in mathematical problem solving for language models. We analyze the characteristics and extent of this issue across different open-source models, datasets of varying difficulty levels, and decoding strategies. Specifically, we explore how false positives influence the inference time scaling behavior of language models. Our experimental results reveal that: (1) false positive solutions persist across different models, datasets, and decoding methods, (2) sampling-based inference time scaling methods do not alleviate the problem, and (3) the pass@N evaluation metric is more susceptible to false positives, suggesting a significantly lower scaling ceiling than what automatic evaluations indicate. Additionally, we analyze specific instances of false positives and discuss potential limitations in self-improvement techniques and synthetic data generation under such conditions. Our data and code are publicly available at https://github.com/Wloner0809/False-Positives-in-Math.

cs.CL↗

arXiv2025

Federated Cross-Training Learners for Robust Generalization under Data Heterogeneity

Federated learning benefits from cross-training strategies, which enables models to train on data from distinct sources to improve generalization capability. However, due to inherent differences in data distributions, the optimization goals of local models remain misaligned, and this mismatch continues to manifest as feature space heterogeneity even after cross-training. We argue that knowledge distillation from the personalized view preserves client-specific characteristics and expands the local knowledge base, while distillation from the global view provides consistent semantic anchors that facilitate feature alignment across clients. To achieve this goal, this paper presents a cross-training scheme, termed FedCT, includes three main modules, where the consistency-aware knowledge broadcasting module aims to optimize model assignment strategies, which enhances collaborative advantages between clients and achieves an efficient federated learning process. The multi-view knowledge-guided representation learning module leverages fused prototypical knowledge from both global and local views to enhance the preservation of local knowledge before and after model exchange, as well as to ensure consistency between local and global knowledge. The mixup-based feature augmentation module aggregates rich information to further increase the diversity of feature spaces, which enables the model to better discriminate complex samples. Extensive experiments were conducted on four datasets in terms of performance comparison, ablation study, in-depth analysis and case study. The results demonstrated that FedCT alleviates knowledge forgetting from both local and global views, which enables it outperform state-of-the-art methods.

cs.AI↗

arXiv2025

Discovering Mathematical Equations with Diffusion Language Model

Discovering valid and meaningful mathematical equations from observed data plays a crucial role in scientific discovery. While this task, symbolic regression, remains challenging due to the vast search space and the trade-off between accuracy and complexity. In this paper, we introduce DiffuSR, a pre-training framework for symbolic regression built upon a continuous-state diffusion language model. DiffuSR employs a trainable embedding layer within the diffusion process to map discrete mathematical symbols into a continuous latent space, modeling equation distributions effectively. Through iterative denoising, DiffuSR converts an initial noisy sequence into a symbolic equation, guided by numerical data injected via a cross-attention mechanism. We also design an effective inference strategy to enhance the accuracy of the diffusion-based equation generator, which injects logit priors into genetic programming. Experimental results on standard symbolic regression benchmarks demonstrate that DiffuSR achieves competitive performance with state-of-the-art autoregressive methods and generates more interpretable and diverse mathematical expressions.

cs.LG↗

arXiv2025

Fighting Fire with Fire (F3): A Training-free and Efficient Visual Adversarial Example Purification Method in LVLMs

Recent advances in large vision-language models (LVLMs) have showcased their remarkable capabilities across a wide range of multimodal vision-language tasks. However, these models remain vulnerable to visual adversarial attacks, which can substantially compromise their performance. In this paper, we introduce F3, a novel adversarial purification framework that employs a counterintuitive ``fighting fire with fire'' strategy: intentionally introducing simple perturbations to adversarial examples to mitigate their harmful effects. Specifically, F3 leverages cross-modal attentions derived from randomly perturbed adversary examples as reference targets. By injecting noise into these adversarial examples, F3 effectively refines their attention, resulting in cleaner and more reliable model outputs. Remarkably, this seemingly paradoxical approach of employing noise to counteract adversarial attacks yields impressive purification results. Furthermore, F3 offers several distinct advantages: it is training-free and straightforward to implement, and exhibits significant computational efficiency improvements compared to existing purification methods. These attributes render F3 particularly suitable for large-scale industrial applications where both robust performance and operational efficiency are critical priorities. The code is available at https://github.com/btzyd/F3.

cs.CV↗

arXiv2025

Harmonic functions on minimal submanifold with cylindrical cone

In this paper, we study the relationship between the dimension of linear space of harmonic function with growth bounded by a fixed-degree polynomial on a minimal submanifold in Euclidean space and that on its one cylindrical tangent cone at infinity by using the method in [6, 7]. Specifically, if this degree does not coincide with any growth of polynomial-growth harmonic functions on the cone, then the corresponding dimensions are equal.

math.DG↗

arXiv2025

SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning

Vision-Language-Action (VLA) models have recently emerged as a powerful paradigm for robotic manipulation. Despite substantial progress enabled by large-scale pretraining and supervised fine-tuning (SFT), these models face two fundamental challenges: (i) the scarcity and high cost of large-scale human-operated robotic trajectories required for SFT scaling, and (ii) limited generalization to tasks involving distribution shift. Recent breakthroughs in Large Reasoning Models (LRMs) demonstrate that reinforcement learning (RL) can dramatically enhance step-by-step reasoning capabilities, raising a natural question: Can RL similarly improve the long-horizon step-by-step action planning of VLA? In this work, we introduce SimpleVLA-RL, an efficient RL framework tailored for VLA models. Building upon veRL, we introduce VLA-specific trajectory sampling, scalable parallelization, multi-environment rendering, and optimized loss computation. When applied to OpenVLA-OFT, SimpleVLA-RL achieves SoTA performance on LIBERO and even outperforms $π_0$ on RoboTwin 1.0\&2.0 with the exploration-enhancing strategies we introduce. SimpleVLA-RL not only reduces dependence on large-scale data and enables robust generalization, but also remarkably surpasses SFT in real-world tasks. Moreover, we identify a novel phenomenon ``pushcut'' during RL training, wherein the policy discovers previously unseen patterns beyond those seen in the previous training process. Github: https://github.com/PRIME-RL/SimpleVLA-RL

cs.RO↗

arXiv2025

FinZero: Launching Multi-modal Financial Time Series Forecast with Large Reasoning Model

Financial time series forecasting is both highly significant and challenging. Previous approaches typically standardized time series data before feeding it into forecasting models, but this encoding process inherently leads to a loss of important information. Moreover, past time series models generally require fixed numbers of variables or lookback window lengths, which further limits the scalability of time series forecasting. Besides, the interpretability and the uncertainty in forecasting remain areas requiring further research, as these factors directly impact the reliability and practical value of predictions. To address these issues, we first construct a diverse financial image-text dataset (FVLDB) and develop the Uncertainty-adjusted Group Relative Policy Optimization (UARPO) method to enable the model not only output predictions but also analyze the uncertainty of those predictions. We then proposed FinZero, a multimodal pre-trained model finetuned by UARPO to perform reasoning, prediction, and analytical understanding on the FVLDB financial time series. Extensive experiments validate that FinZero exhibits strong adaptability and scalability. After fine-tuning with UARPO, FinZero achieves an approximate 13.48\% improvement in prediction accuracy over GPT-4o in the high-confidence group, demonstrating the effectiveness of reinforcement learning fine-tuning in multimodal large model, including in financial time series forecasting tasks.

q-fin.CP↗

arXiv2025

Mixed Quantum-Classical Approaches to Spin Current and Polarization Dynamics in Chiral Molecular Junctions

Chiral molecular junctions offer a promising platform for realizing chiral-induced spin selectivity (CISS), where spin filtering occurs without external magnetic fields. Here, we investigate spin transport in such junctions by combining quantum master equation (QME) methods for purely electronic dynamics with surface hopping (SH) and mean-field Ehrenfest (MF) approaches to incorporate electron-phonon coupling. Our results show that transient spin polarization arises but ultimately decays to zero at long times. We find that bias voltage, molecular length, and spin-orbit coupling (SOC) strongly influence the spin current dynamics: higher bias enhances spin current but reduces polarization, while longer molecules and stronger SOC amplify transient polarization. Including electron-phonon coupling modifies current-voltage characteristics, enhancing spin currents at intermediate bias but suppressing them at high bias, while leaving the polarization dynamics largely unchanged. These findings highlight the interplay between electronic and vibrational effects in CISS and provide guidance for designing molecular spintronic devices.

cond-mat.mes-hall↗

arXiv2025

Short GRB 090510: a magnetized neutron star binary merger leading to a black hole

We model the short gamma-ray bursts (GRB) 090510 as the product of a magnetized neutron star (NS) binary merger. Accounting for the NS critical mass constraint given by the mass of PSR J0952--0607, we infer that GRB 090510 was a highly-magnetized NS-NS merger that left as remnant a Kerr black hole (BH) of $2.4 M_\odot$ with a low-mass accretion disk. The gamma-ray precursor is powered by the magnetic energy released during the merger of the NSs. The prompt emission originates at the transparency of an ultra-relativistic $e^+e^-$ pair-plasma produced by the overcritical electric field induced by the rotating strong magnetic field around the merged object before it reaches the critical mass, the GeV emission by the extractable energy of the newborn BH, and the X-ray afterglow by accretion onto it. We derive the masses of the merging NSs, their magnetic fields, the BH mass, spin, and irreducible mass, the strength of the magnetic field, the disk mass, and obtain an estimate of the gravitational-wave emission during the merger phase preceding the prompt short GRB emission. The inferred parameters agree with up-to-date numerical relativity simulations, confirming that strong magnetic fields above $10^{14}$ G develop in NS-NS mergers and that mergers leading to a central BH remnant have low-mass disks of $\sim 10^{-2} M_\odot$. We also advance the possibility that quasi-period oscillations of tens of Hz of frequency due to Lense-Thirring precession of the matter surrounding the merged object before BH formation can explain the successive spikes following the prompt emission peak.

astro-ph.HE↗

arXiv2025

CoMMIT: Coordinated Multimodal Instruction Tuning

Instruction tuning in multimodal large language models (MLLMs) generally involves cooperative learning between a backbone LLM and a feature encoder of non-text input modalities. The major challenge is how to efficiently find the synergy between the two modules so that LLMs can adapt their reasoning abilities to downstream tasks while feature encoders can adjust to provide more task-specific information about its modality. In this paper, we analyze the MLLM instruction tuning from both theoretical and empirical perspectives, where we find the unbalanced learning between the feature encoder and the LLM can cause problems of oscillation and biased learning that lead to sub-optimal convergence. Inspired by our findings, we propose a Multimodal Balance Coefficient that enables quantitative measurement of the balance of learning. Based on this, we further design a dynamic learning scheduler that better coordinates the learning between the LLM and feature encoder, alleviating the problems of oscillation and biased learning. In addition, we introduce an auxiliary regularization on the gradient to promote updating with larger step sizes, which potentially allows for a more accurate estimation of the proposed MultiModal Balance Coefficient and further improves the training sufficiency. Our proposed approach is agnostic to the architecture of LLM and feature encoder, so it can be generically integrated with various MLLMs. We conduct experiments on multiple downstream tasks with various MLLMs, demonstrating that the proposed method is more effective than the baselines in MLLM instruction tuning.

cs.LG↗

arXiv2025

Robust Analysis for Resilient AI System

Operational hazards in Manufacturing Industrial Internet (MII) systems generate severe data outliers that cripple traditional statistical analysis. This paper proposes a novel robust regression method, DPD-Lasso, which integrates Density Power Divergence with Lasso regularization to analyze contaminated data from AI resilience experiments. We develop an efficient iterative algorithm to overcome previous computational bottlenecks. Applied to an MII testbed for Aerosol Jet Printing, DPD-Lasso provides reliable, stable performance on both clean and outlier-contaminated data, accurately quantifying hazard impacts. This work establishes robust regression as an essential tool for developing and validating resilient industrial AI systems.

stat.AP↗

arXiv2025

TemporalFlowViz: Parameter-Aware Visual Analytics for Interpreting Scramjet Combustion Evolution

Understanding the complex combustion dynamics within scramjet engines is critical for advancing high-speed propulsion technologies. However, the large scale and high dimensionality of simulation-generated temporal flow field data present significant challenges for visual interpretation, feature differentiation, and cross-case comparison. In this paper, we present TemporalFlowViz, a parameter-aware visual analytics workflow and system designed to support expert-driven clustering, visualization, and interpretation of temporal flow fields from scramjet combustion simulations. Our approach leverages hundreds of simulated combustion cases with varying initial conditions, each producing time-sequenced flow field images. We use pretrained Vision Transformers to extract high-dimensional embeddings from these frames, apply dimensionality reduction and density-based clustering to uncover latent combustion modes, and construct temporal trajectories in the embedding space to track the evolution of each simulation over time. To bridge the gap between latent representations and expert reasoning, domain specialists annotate representative cluster centroids with descriptive labels. These annotations are used as contextual prompts for a vision-language model, which generates natural-language summaries for individual frames and full simulation cases. The system also supports parameter-based filtering, similarity-based case retrieval, and coordinated multi-view exploration to facilitate in-depth analysis. We demonstrate the effectiveness of TemporalFlowViz through two expert-informed case studies and expert feedback, showing TemporalFlowViz enhances hypothesis generation, supports interpretable pattern discovery, and enhances knowledge discovery in large-scale scramjet combustion analysis.

cs.CV↗

arXiv2025

Efficient Entanglement Routing for Satellite-Aerial-Terrestrial Quantum Networks

In the era of 6G and beyond, space-aerial-terrestrial quantum networks (SATQNs) are shaping the future of the global-scale quantum Internet. This paper investigates the collaboration among satellite, aerial, and terrestrial quantum networks to efficiently transmit high-fidelity quantum entanglements over long distances. We begin with a comprehensive overview of existing satellite-, aerial-, and terrestrial-based quantum networks. Subsequently, we address the entanglement routing problem with the objective of maximizing quantum network throughput by jointly optimizing path selection and entanglement generation rates (PS-EGR). Given that the original problem is formulated as a mixed-integer linear programming (MILP) problem, which is inherently intractable, we propose a Benders' decomposition (BD)-based algorithm to solve the problem efficiently. Numerical results validate the effectiveness of the proposed PS-EGR scheme, offering valuable insights into various optimizable factors within the system. Finally, we discuss the current challenges and propose promising avenues for future research in SATQNs.

quant-ph↗

arXiv2025

DynaSaur: Large Language Agents Beyond Predefined Actions

Existing LLM agent systems typically select actions from a fixed and predefined set at every step. While this approach is effective in closed, narrowly scoped environments, it presents two major challenges for real-world, open-ended scenarios: (1) it significantly restricts the planning and acting capabilities of LLM agents, and (2) it requires substantial human effort to enumerate and implement all possible actions, which is impractical in complex environments with a vast number of potential actions. To address these limitations, we propose an LLM agent framework that can dynamically create and compose actions as needed. In this framework, the agent interacts with its environment by generating and executing programs written in a general-purpose programming language. Moreover, generated actions are accumulated over time for future reuse. Our extensive experiments across multiple benchmarks show that this framework significantly improves flexibility and outperforms prior methods that rely on a fixed action set. Notably, it enables LLM agents to adapt and recover in scenarios where predefined actions are insufficient or fail due to unforeseen edge cases. Our code can be found in https://github.com/adobe-research/dynasaur.

cs.CL↗

arXiv2025

Delayed Interactions in Active Agents: Stability and Formations

Active agents with time-delayed interactions arise naturally in various real-world systems, such as biological systems, transportation networks and robotic swarms. Such systems are typically modeled as Delay Differential Equations (DDEs) that incorporate inertial effects. In this paper, we investigate the stability of pattern formation of active agents with inertia and time delays, in both uncoupled and coupled scenarios. We derive and analyze a high-dimensional linear DDE model that characterizes the stability of such formations. Starting with the uncoupled scenario, where agents are driven only by a virtual leader, we describe the stability spectrum and provide conditions for the delay-independent (absolute) stability of the formations, as well as delay-dependent stability and unstable hyperbolic behavior. Different cases correspond to distinct universality classes of the corresponding spectrum. For the coupled scenario, where agents are driven by both the virtual leader and inter-agent interactions, we consider both symmetric and non-symmetric coupling topologies. Here we also provide an explicit spectrum classification, including the absolute stability criterion. Additionally, we investigate interactions in the large-delay limit, where delays affect inter-agent coupling, while local feedback remains instantaneous. In this limit, we prove rigorously that the stability region in the complex plane of the eigenvalues of the Laplacian matrix converges to a circle centered at the origin, a phenomenon previously observed in delay-coupled networks. Our findings provide a universal framework for understanding stable formations and motions of active agents with delayed interactions.

math.DS↗

arXiv2025

MedFoundationHub: A Lightweight and Secure Toolkit for Deploying Medical Vision Language Foundation Models

Recent advances in medical vision-language models (VLMs) open up remarkable opportunities for clinical applications such as automated report generation, copilots for physicians, and uncertainty quantification. However, despite their promise, medical VLMs introduce serious security concerns, most notably risks of Protected Health Information (PHI) exposure, data leakage, and vulnerability to cyberthreats - which are especially critical in hospital environments. Even when adopted for research or non-clinical purposes, healthcare organizations must exercise caution and implement safeguards. To address these challenges, we present MedFoundationHub, a graphical user interface (GUI) toolkit that: (1) enables physicians to manually select and use different models without programming expertise, (2) supports engineers in efficiently deploying medical VLMs in a plug-and-play fashion, with seamless integration of Hugging Face open-source models, and (3) ensures privacy-preserving inference through Docker-orchestrated, operating system agnostic deployment. MedFoundationHub requires only an offline local workstation equipped with a single NVIDIA A6000 GPU, making it both secure and accessible within the typical resources of academic research labs. To evaluate current capabilities, we engaged board-certified pathologists to deploy and assess five state-of-the-art VLMs (Google-MedGemma3-4B, Qwen2-VL-7B-Instruct, Qwen2.5-VL-7B-Instruct, and LLaVA-1.5-7B/13B). Expert evaluation covered colon cases and renal cases, yielding 1015 clinician-model scoring events. These assessments revealed recurring limitations, including off-target answers, vague reasoning, and inconsistent pathology terminology.

cs.CV↗

arXiv2025

Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications

Multi-sequence Magnetic Resonance Imaging (MRI) offers remarkable versatility, enabling the distinct visualization of different tissue types. Nevertheless, the inherent heterogeneity among MRI sequences poses significant challenges to the generalization capability of deep learning models. These challenges undermine model performance when faced with varying acquisition parameters, thereby severely restricting their clinical utility. In this study, we present PRISM, a foundation model PRe-trained with large-scale multI-Sequence MRI. We collected a total of 64 datasets from both public and private sources, encompassing a wide range of whole-body anatomical structures, with scans spanning diverse MRI sequences. Among them, 336,476 volumetric MRI scans from 34 datasets (8 public and 26 private) were curated to construct the largest multi-organ multi-sequence MRI pretraining corpus to date. We propose a novel pretraining paradigm that disentangles anatomically invariant features from sequence-specific variations in MRI, while preserving high-level semantic representations. We established a benchmark comprising 44 downstream tasks, including disease diagnosis, image segmentation, registration, progression prediction, and report generation. These tasks were evaluated on 32 public datasets and 5 private cohorts. PRISM consistently outperformed both non-pretrained models and existing foundation models, achieving first-rank results in 39 out of 44 downstream benchmarks with statistical significance improvements. These results underscore its ability to learn robust and generalizable representations across unseen data acquired under diverse MRI protocols. PRISM provides a scalable framework for multi-sequence MRI analysis, thereby enhancing the translational potential of AI in radiology. It delivers consistent performance across diverse imaging protocols, reinforcing its clinical applicability.

eess.IV↗

arXiv2025

360Brew: A Decoder-only Foundation Model for Personalized Ranking and Recommendation

Ranking and recommendation systems are the foundation for numerous online experiences, ranging from search results to personalized content delivery. These systems have evolved into complex, multilayered architectures that leverage vast datasets and often incorporate thousands of predictive models. The maintenance and enhancement of these models is a labor intensive process that requires extensive feature engineering. This approach not only exacerbates technical debt but also hampers innovation in extending these systems to emerging problem domains. In this report, we present our research to address these challenges by utilizing a large foundation model with a textual interface for ranking and recommendation tasks. We illustrate several key advantages of our approach: (1) a single model can manage multiple predictive tasks involved in ranking and recommendation, (2) decoder models with textual interface due to their comprehension of reasoning capabilities, can generalize to new recommendation surfaces and out-of-domain problems, and (3) by employing natural language interfaces for task definitions and verbalizing member behaviors and their social connections, we eliminate the need for feature engineering and the maintenance of complex directed acyclic graphs of model dependencies. We introduce our research pre-production model, 360Brew V1.0, a 150B parameter, decoder-only model that has been trained and fine-tuned on LinkedIn's data and tasks. This model is capable of solving over 30 predictive tasks across various segments of the LinkedIn platform, achieving performance levels comparable to or exceeding those of current production systems based on offline metrics, without task-specific fine-tuning. Notably, each of these tasks is conventionally addressed by dedicated models that have been developed and maintained over multiple years by teams of a similar or larger size than our own.

cs.IR↗