arXiv subjects

Yuan Huang

Publications and source records attributed to Yuan Huang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Bohrium + SciMaster: Building the Infrastructure and Ecosystem for Agentic Science at Scale

AI agents are emerging as a practical way to run multi-step scientific workflows that interleave reasoning with tool use and verification, pointing to a shift from isolated AI-assisted steps toward \emph{agentic science at scale}. This shift is increasingly feasible, as scientific tools and models can be invoked through stable interfaces and verified with recorded execution traces, and increasingly necessary, as AI accelerates scientific output and stresses the peer-review and publication pipeline, raising the bar for traceability and credible evaluation. However, scaling agentic science remains difficult: workflows are hard to observe and reproduce; many tools and laboratory systems are not agent-ready; execution is hard to trace and govern; and prototype AI Scientist systems are often bespoke, limiting reuse and systematic improvement from real workflow signals. We argue that scaling agentic science requires an infrastructure-and-ecosystem approach, instantiated in Bohrium+SciMaster. Bohrium acts as a managed, traceable hub for AI4S assets -- akin to a HuggingFace of AI for Science -- that turns diverse scientific data, software, compute, and laboratory systems into agent-ready capabilities. SciMaster orchestrates these capabilities into long-horizon scientific workflows, on which scientific agents can be composed and executed. Between infrastructure and orchestration, a \emph{scientific intelligence substrate} organizes reusable models, knowledge, and components into executable building blocks for workflow reasoning and action, enabling composition, auditability, and improvement through use. We demonstrate this stack with eleven representative master agents in real workflows, achieving orders-of-magnitude reductions in end-to-end scientific cycle time and generating execution-grounded signals from real workloads at multi-million scale.

cs.AI↗

arXiv2025

Integrating Prioritized and Non-Prioritized Structures in Win Statistics

Composite endpoints are frequently used as primary or secondary analyses in cardiovascular clinical trials to increase clinical relevance and statistical efficiency. Alternatively, the Win Ratio (WR) and other Win Statistics (WS) analyses rely on a strict hierarchical ordering of endpoints, assigning higher priority to clinically important endpoints. However, determining a definitive endpoint hierarchy can be challenging and may not adequately reflect situations where endpoints have comparable importance. In this study, we discuss the challenges of endpoint prioritization, underscore its critical role in WS analyses, and propose Rotation WR (RWR), a hybrid prioritization framework that integrates both prioritized and non-prioritized structures. By permitting blocks of equally-prioritized endpoints, RWR accommodates endpoints of equal or near equal clinical importance, recurrent events, and contexts requiring individualized shared decision making. Statistical inference for RWR is developed using U-statistics theory, including the hypothesis testing procedure and confidence interval construction. Extensions to two additional WS measures, Rotation Net Benefit and Rotation Win Odds, are also provided. Through extensive simulation studies involving multiple time-to-event endpoints, including recurrent events, we demonstrate that RWR achieves valid type I error control, desirable statistical power, and accurate confidence interval coverage. We illustrate both the methodological and practical insights of our work in a case study on endpoint prioritization with the SPRINT clinical trial, highlighting its implications for real-world clinical trial studies.

stat.ME↗

arXiv2025

A Non-Bipartite Matching Framework for Difference-in-Differences with General Treatment Types

Difference-in-differences (DID) is one of the most widely used causal inference frameworks in observational studies. However, most existing DID methods are designed for binary treatments and cannot be readily applied to non-binary treatment settings. Although recent work has begun to extend DID to non-binary (e.g., continuous) treatments, these approaches typically require strong additional assumptions, including parametric outcome models or the presence of idealized comparison units with (nearly) static treatment levels over time (commonly called ``stayers'' or ``quasi-stayers''). In this technical note, we introduce a new non-bipartite matching framework for DID that naturally accommodates general treatment types (e.g., binary, ordinal, or continuous). Our framework makes three main contributions. First, we develop an optimal non-bipartite matching design for DID that jointly balances baseline covariates across comparable units (reducing bias) and maximizes contrasts in treatment trajectories over time (improving efficiency). Second, we establish a post-matching randomization condition, the design-based counterpart to the traditional parallel-trends assumption, which enables valid design-based inference. Third, we introduce the sample average DID ratio, a finite-population-valid and fully nonparametric causal estimand applicable to arbitrary treatment types. Our design-based approach that preserves the full treatment-dose information, avoids parametric assumptions, does not rely on the existence of stayers or quasi-stayers, and operates entirely within a finite-population framework, without appealing to hypothetical super-populations or outcome distributions.

stat.ME↗

arXiv2025

Correlation Matters! Streamlining the Sample Size Procedure with Composite Time-to-event Endpoints

Composite endpoints are widely used in cardiovascular clinical trials to improve statistical efficiency while preserving clinical relevance. The Win Ratio (WR) measure and more general frameworks of Win Statistics have emerged as increasingly popular alternatives to traditional time-to-first-event analyses. Although analytic sample size formulas for WR have been developed, they rely on design parameters that are often not straightforward to specify. Consequently, sample size determination in clinical trials with WR as the primary analysis is most often based on simulations, which can be computationally intensive. Moreover, these simulations commonly assume independence among component endpoints, an assumption that may not hold in practice and can lead to misleading power estimates. To address this challenge, we derive refined formulas to calculate the proportions of wins, losses, and ties for multiple prioritized time-to-event endpoints. These formulas rely on familiar design inputs and become directly applicable when integrated with existing sample size methods. We conduct a comprehensive assessment of how correlation among endpoints affects sample size requirements across varying design features. We further demonstrate the role of correlations through two case studies based on the landmark SPRINT and STICH clinical trials to generate further insights.

stat.ME↗

arXiv2025

Generalizing the Finkelstein-Schoenfeld Test to Incorporate Multiple Alternating Thresholds

Composite endpoints consisting of both terminal and non-terminal events, such as death and hospitalization, are frequently used in cardiovascular clinical trials. The Finkelstein-Schoenfeld (FS) test provides a way to employ a hierarchical structure to combine fatal and non-fatal events by giving death information an absolute priority, which may limit the contribution of clinically meaningful non-fatal events. To provide a more flexible alternative, we propose the Finkelstein-Schoenfeld with Multiple Thresholds (FS-MT) test, which extends the standard FS test by incorporating multiple thresholds applied sequentially and alternating across endpoints. A weighted adaptive approach is also developed to help determine the thresholds in FS-MT. The proposed approach retains the statistical properties of the FS test while allowing more flexible use of information from lower-priority events. We evaluate the operating characteristics of the proposed test through simulations that vary the follow-up time, the correlation between events, and the treatment effect sizes. A case study based on the Digitalis Investigation Group clinical trial data is presented to further illustrate our proposed method. An R package ``FSMT'' that implements the proposed methodology has been developed.

stat.ME↗

arXiv2025

Learning-at-Criticality in Large Language Models for Quantum Field Theory and Beyond

Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles. While artificial intelligence (AI) offers promise, its typical need for vast datasets to learn from hinders its use in these information-scarce frontiers. We introduce learning at criticality (LaC), a reinforcement learning (RL) scheme that tunes Large Language Models (LLMs) to a sharp learning transition, addressing this information scarcity. At this transition, LLMs achieve peak generalization from minimal data, exemplified by 7-digit base-7 addition -- a test of nontrivial arithmetic reasoning. To elucidate this peak, we analyze a minimal concept-network model (CoNet) designed to capture the essence of how LLMs might link tokens. Trained on a single exemplar, this model also undergoes a sharp learning transition. This transition exhibits hallmarks of a second-order phase transition, notably power-law distributed solution path lengths. At this critical point, the system maximizes a ``critical thinking pattern" crucial for generalization, enabled by the underlying scale-free exploration. This suggests LLMs reach peak performance by operating at criticality, where such explorative dynamics enable the extraction of underlying operational rules. We demonstrate LaC in quantum field theory: an 8B-parameter LLM, tuned to its critical point by LaC using a few exemplars of symbolic Matsubara sums, solves unseen, higher-order problems, significantly outperforming far larger models. LaC thus leverages critical phenomena, a physical principle, to empower AI for complex, data-sparse challenges in fundamental physics.

cs.LG↗

arXiv2025

Beyond Survival: Evaluating LLMs in Social Deduction Games with Human-Aligned Strategies

Social deduction games like Werewolf combine language, reasoning, and strategy, providing a testbed for studying natural language and social intelligence. However, most studies reduce the game to LLM-based self-play, yielding templated utterances and anecdotal cases that overlook the richness of social gameplay. Evaluation further relies on coarse metrics such as survival time or subjective scoring due to the lack of quality reference data. To address these gaps, we curate a high-quality, human-verified multimodal Werewolf dataset containing over 100 hours of video, 32.4M utterance tokens, and 15 rule variants. Based on this dataset, we propose a novel strategy-alignment evaluation that leverages the winning faction's strategies as ground truth in two stages: 1) Speech evaluation, formulated as multiple-choice-style tasks that assess whether the model can adopt appropriate stances across five dimensions of social ability; and 2) Decision evaluation, which assesses the model's voting choices and opponent-role inferences. This framework enables a fine-grained evaluation of models' linguistic and reasoning capabilities, while capturing their ability to generate strategically coherent gameplay. Our experiments show that state-of-the-art LLMs show diverse performance, with roughly half remain below 0.50, revealing clear gaps in deception and counterfactual reasoning. We hope our dataset further inspires research on language, reasoning, and strategy in multi-agent interaction.

cs.CL↗

arXiv2025

Fine-tuning of Large Language Models for Domain-Specific Cybersecurity Knowledge

Recent advancements in training paradigms for Large Language Models (LLMs) have unlocked their remarkable capabilities in natural language processing and cross-domain generalization. While LLMs excel in tasks like programming and mathematical problem-solving, their zero-shot performance in specialized domains requiring expert knowledge, such as cybersecurity, is often suboptimal. This limitation arises because foundational LLMs are designed for general-purpose applications, constraining their ability to encapsulate domain-specific expertise within their parameter space. To address this, we explore fine-tuning strategies to embed cybersecurity knowledge into LLMs, enhancing their performance in cybersecurity question-answering (Q\&A) tasks while prioritizing computational efficiency. Specifically, we investigate Supervised Fine-Tuning (SFT), Low-Rank Adaptation (LoRA), and Quantized Low-Rank Adaptation (QLoRA) using a cybersecurity Q\&A dataset. Our results demonstrate that these fine-tuning approaches significantly outperform the foundational model in cybersecurity Q\&A tasks. Moreover, LoRA and QLoRA achieve comparable performance to SFT with substantially lower computational costs, offering an efficient pathway for adapting LLMs to specialized domains. Our work highlights the potential of low-rank fine-tuning strategies to bridge the gap between general-purpose LLMs and domain-specific applications.

cs.LG↗

arXiv2025

ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data

Vision-Language Models (VLMs) have enabled computer use agents (CUAs) that operate GUIs autonomously, showing great potential, yet progress is limited by the lack of large-scale, open-source computer use data and foundation models. In this work, we introduce ScaleCUA, a step toward scaling open-source CUAs. It offers a large-scale dataset spanning 6 operating systems and 3 task domains, built via a closed-loop pipeline uniting automated agents with human experts. Trained on this scaled-up data, ScaleCUA can operate seamlessly across platforms. Specifically, it delivers strong gains over baselines (+26.6 on WebArena-Lite-v2, +10.7 on ScreenSpot-Pro) and sets new state-of-the-art results (94.4% on MMBench-GUI L1-Hard, 60.6% on OSWorld-G, 47.4% on WebArena-Lite-v2). These findings underscore the power of data-driven scaling for general-purpose computer use agents. We will release data, models, and code to advance future research: https://github.com/OpenGVLab/ScaleCUA.

cs.CV↗

arXiv2025

Testing Prioritized Composite Endpoint with Multiple Follow-up Time Examinations

Composite endpoints are widely used in cardiovascular clinical trials. In recent years, hierarchical composite endpoints-particularly the win ratio approach and its predecessor, the Finkelstein-Schoenfeld (FS) test, also known as the unmatched win ratio test-have gained popularity. These methods involve comparing individuals across multiple endpoints, ranked by priority, with mortality typically assigned the highest priority in many applications. However, these methods have not accounted for varying treatment effects, known as non-constant hazards over time in the context of survival analysis. To address this limitation, we propose an adaptation of the FS test that incorporates progressive follow-up time, which we will refer to as ProFS. This proposed test can jointly evaluate treatment effects at various follow-up time points by incorporating the maximum of several FS test statistics calculated at those specific times. Moreover, ProFS also supports clinical trials with group sequential monitoring strategies, providing flexibility in trial design. As demonstrated through extensive simulations, ProFS offers increased statistical power in scenarios where the treatment effect is mainly in the short term or when the second (non-fatal) layer might be concealed by a lack of effect or weak effect on the top (fatal) layer. We also apply ProFS to the SPRINT clinical trial, illustrating how our proposed method improves the performance of FS.

stat.ME↗

arXiv2025

MMBench-GUI: Hierarchical Multi-Platform Evaluation Framework for GUI Agents

We introduce MMBench-GUI, a hierarchical benchmark for evaluating GUI automation agents across Windows, macOS, Linux, iOS, Android, and Web platforms. It comprises four levels: GUI Content Understanding, Element Grounding, Task Automation, and Task Collaboration, covering essential skills for GUI agents. In addition, we propose a novel Efficiency-Quality Area (EQA) metric to assess GUI agent execution efficiency in online automation scenarios. Through MMBench-GUI, we identify accurate visual grounding as a critical determinant of overall task success, emphasizing the substantial benefits of modular frameworks that integrate specialized grounding modules. Furthermore, to achieve reliable GUI automation, an agent requires strong task planning and cross-platform generalization abilities, with long-context memory, a broad action space, and long-term reasoning playing a critical role. More important, task efficiency remains a critically underexplored dimension, and all models suffer from substantial inefficiencies, with excessive redundant steps even when tasks are ultimately completed. The integration of precise localization, effective planning, and early stopping strategies is indispensable to enable truly efficient and scalable GUI automation. Our benchmark code, evaluation data, and running environment will be publicly available at https://github.com/open-compass/MMBench-GUI.

cs.CV↗

arXiv2025

Empowering Multi-class Classification for Multivariate Functional Data with Simultaneous Feature Selection

The opportunity to utilize complex functional data types for conducting classification tasks is emerging with the growing availability of imaging data. However, the tools capable of effectively managing imaging data are limited, let alone those that can further leverage other one-dimensional functional data. Inspired by the extensive data provided by the Alzheimer's Disease Neuroimaging Initiative (ADNI), we introduce a novel classifier tailored for complex functional data. Each observation in this framework may be associated with numerous functional processes, varying in dimensions, such as curves and images. Each predictor is a random element in an infinite dimensional function space, and the number of functional predictors p can potentially be much greater than the sample size n. In this paper, we introduce a novel and scalable classifier termed functional BIC deep neural network. By adopting a sparse deep Rectified Linear Unit network architecture and incorporating the LassoNet algorithm, the proposed unified model performs feature selection and classification simultaneously, which is contrast to the existing functional data classifiers. The challenge arises from the complex inter-correlation structures among multiple functional processes, and at meanwhile without any assumptions on the distribution of these processes. Simulation study and real data application are carried out to demonstrate its favorable performance.

stat.ME↗

arXiv2025

Mediation Analysis for Sparse and Irregularly Spaced Longitudinal Outcomes with Application to the MrOS Sleep Study

Mediation analysis has become a widely used method for identifying the pathways through which an independent variable influences a dependent variable via intermediate mediators. However, limited research addresses the case where mediators are high-dimensional and the outcome is represented by sparse, irregularly spaced longitudinal data. To address these challenges, we propose a mediation analysis approach for scalar exposures, high-dimensional mediators, and sparse longitudinal outcomes. This approach effectively identifies significant mediators by addressing two key issues: (i) the underlying correlation structure within the sparse and irregular cognitive measurements, and (ii) adjusting mediation effects to handle the high-dimensional set of candidate mediators. In the MrOS Sleep study, our primary objective is to explore lipid pathways that may mediate the relationship between rest-activity rhythms and longitudinal cognitive decline in older men. Our findings suggest a potential mechanism involving rest-activity rhythms, lipid metabolites, and cognitive decline, and highlight significant mediators identified through multiple testing procedures.

stat.ME↗

arXiv2025

UrbanCraft: Urban View Extrapolation via Hierarchical Sem-Geometric Priors

Existing neural rendering-based urban scene reconstruction methods mainly focus on the Interpolated View Synthesis (IVS) setting that synthesizes from views close to training camera trajectory. However, IVS can not guarantee the on-par performance of the novel view outside the training camera distribution (\textit{e.g.}, looking left, right, or downwards), which limits the generalizability of the urban reconstruction application. Previous methods have optimized it via image diffusion, but they fail to handle text-ambiguous or large unseen view angles due to coarse-grained control of text-only diffusion. In this paper, we design UrbanCraft, which surmounts the Extrapolated View Synthesis (EVS) problem using hierarchical sem-geometric representations serving as additional priors. Specifically, we leverage the partially observable scene to reconstruct coarse semantic and geometric primitives, establishing a coarse scene-level prior through an occupancy grid as the base representation. Additionally, we incorporate fine instance-level priors from 3D bounding boxes to enhance object-level details and spatial relationships. Building on this, we propose the \textbf{H}ierarchical \textbf{S}emantic-Geometric-\textbf{G}uided Variational Score Distillation (HSG-VSD), which integrates semantic and geometric constraints from pretrained UrbanCraft2D into the score distillation sampling process, forcing the distribution to be consistent with the observable scene. Qualitative and quantitative comparisons demonstrate the effectiveness of our methods on EVS problem.

cs.CV↗

arXiv2025

MMAC-Copilot: Multi-modal Agent Collaboration Operating Copilot

Large language model agents that interact with PC applications often face limitations due to their singular mode of interaction with real-world environments, leading to restricted versatility and frequent hallucinations. To address this, we propose the Multi-Modal Agent Collaboration framework (MMAC-Copilot), a framework utilizes the collective expertise of diverse agents to enhance interaction ability with application. The framework introduces a team collaboration chain, enabling each participating agent to contribute insights based on their specific domain knowledge, effectively reducing the hallucination associated with knowledge domain gaps. We evaluate MMAC-Copilot using the GAIA benchmark and our newly introduced Visual Interaction Benchmark (VIBench). MMAC-Copilot achieved exceptional performance on GAIA, with an average improvement of 6.8\% over existing leading systems. VIBench focuses on non-API-interactable applications across various domains, including 3D gaming, recreation, and office scenarios. It also demonstrated remarkable capability on VIBench. We hope this work can inspire in this field and provide a more comprehensive assessment of Autonomous agents. The anonymous Github is available at \href{https://anonymous.4open.science/r/ComputerAgentWithVision-3C12}{Anonymous Github}

cs.AI↗

arXiv2025

A New Segment Routing method with Swap Node Selection Strategy Based on Deep Reinforcement Learning for Software Defined Network

The existing segment routing (SR) methods need to determine the routing first and then use path segmentation approaches to select swap nodes to form a segment routing path (SRP). They require re-segmentation of the path when the routing changes. Furthermore, they do not consider the flow table issuance time, which cannot maximize the speed of issuance flow table. To address these issues, this paper establishes an optimization model that can simultaneously form routing strategies and path segmentation strategies for selecting the appropriate swap nodes to reduce flow table issuance time. It also designs an intelligent segment routing algorithm based on deep reinforcement learning (DRL-SR) to solve the proposed model. First, a traffic matrix is designed as the state space for the deep reinforcement learning agent; this matrix includes multiple QoS performance indicators, flow table issuance time overhead and SR label stack depth. Second, the action selection strategy and corresponding reward function are designed, where the agent selects the next node considering the routing; in addition, the action selection strategy whether the newly added node is selected as the swap node and the corresponding reward function are designed considering the time cost factor for the controller to issue the flow table to the swap node. Finally, a series of experiments and their results show that, compared with the existing methods, the designed segmented route optimization model and the intelligent solution algorithm (DRL-SR) can reduce the time overhead required to complete the segmented route establishment task while optimizing performance metrics such as throughput, delays and packet losses.

cs.AI↗

arXiv2025

Range and Bird's Eye View Fused Cross-Modal Visual Place Recognition

Image-to-point cloud cross-modal Visual Place Recognition (VPR) is a challenging task where the query is an RGB image, and the database samples are LiDAR point clouds. Compared to single-modal VPR, this approach benefits from the widespread availability of RGB cameras and the robustness of point clouds in providing accurate spatial geometry and distance information. However, current methods rely on intermediate modalities that capture either the vertical or horizontal field of view, limiting their ability to fully exploit the complementary information from both sensors. In this work, we propose an innovative initial retrieval + re-rank method that effectively combines information from range (or RGB) images and Bird's Eye View (BEV) images. Our approach relies solely on a computationally efficient global descriptor similarity search process to achieve re-ranking. Additionally, we introduce a novel similarity label supervision technique to maximize the utility of limited training data. Specifically, we employ points average distance to approximate appearance similarity and incorporate an adaptive margin, based on similarity differences, into the vanilla triplet loss. Experimental results on the KITTI dataset demonstrate that our method significantly outperforms state-of-the-art approaches.

cs.CV↗

arXiv2025

Vevo: Controllable Zero-Shot Voice Imitation with Self-Supervised Disentanglement

The imitation of voice, targeted on specific speech attributes such as timbre and speaking style, is crucial in speech generation. However, existing methods rely heavily on annotated data, and struggle with effectively disentangling timbre and style, leading to challenges in achieving controllable generation, especially in zero-shot scenarios. To address these issues, we propose Vevo, a versatile zero-shot voice imitation framework with controllable timbre and style. Vevo operates in two core stages: (1) Content-Style Modeling: Given either text or speech's content tokens as input, we utilize an autoregressive transformer to generate the content-style tokens, which is prompted by a style reference; (2) Acoustic Modeling: Given the content-style tokens as input, we employ a flow-matching transformer to produce acoustic representations, which is prompted by a timbre reference. To obtain the content and content-style tokens of speech, we design a fully self-supervised approach that progressively decouples the timbre, style, and linguistic content of speech. Specifically, we adopt VQ-VAE as the tokenizer for the continuous hidden features of HuBERT. We treat the vocabulary size of the VQ-VAE codebook as the information bottleneck, and adjust it carefully to obtain the disentangled speech representations. Solely self-supervised trained on 60K hours of audiobook speech data, without any fine-tuning on style-specific corpora, Vevo matches or surpasses existing methods in accent and emotion conversion tasks. Additionally, Vevo's effectiveness in zero-shot voice conversion and text-to-speech tasks further demonstrates its strong generalization and versatility. Audio samples are available at https://versavoice.github.io.

cs.SD↗