Search arXiv⌕ Search

arXiv subjects

Jin Huang

Publications and source records attributed to Jin Huang.

At least 19 recordsLinked to original sources
arXiv2026

Track finding and particle identification performance with particle physics foundation models

Modern high energy nuclear and particle physics experiments produce large amounts of sparse data that must be reduced significantly in size before performing physics analysis. Recently, there has been an effort to evaluate Foundation Models as a potential tool for performing data analysis. These models have been shown to learn features in high dimensional datasets based on the multi-modal data they are trained on. To show their effectiveness at performing complex pattern recognition tasks, such as reconstructing charged particles in a magnetic field, we evaluated the performance of a Foundation Model based reconstruction on several downstream tasks with a dataset composed of spacepoints in a Time Projection Chamber. We show that the performance of the Foundation Model based data analysis significantly improves upon that of traditional methods, and furthermore can be adapted to learn additional tasks that are not necessarily related to the initial training objective.

physics.data-an↗
arXiv2026

Task-State Governance: A Virtual Machine for Verifiable and Recoverable GUI Agent Execution

Graphical user interface (GUI) agents now execute workflows across real applications, but when execution drifts, users must determine what changed, what should have remained untouched, and what can still be recovered. Plan-centric interfaces leave users supervising procedure, but users live with changes to application state. We introduce task-state governance, an interaction paradigm that represents a task as an editable contract linking desired outcomes, facts that must be preserved, and observed application state. We instantiate this paradigm in TaskVM, which projects live state into editable task controls, executes user edits as genuine interface actions, verifies outcomes through independent observation, and marks each action as reversible, compensatable, or irreversible. Across an automated contract benchmark, a controlled user study, and an open live-model exploration, the paradigm strengthened contract compliance and users' recovery control and trust, while execution outcomes were held constant. Task contracts open a design space for governable agent interfaces.

cs.HC↗
arXiv2026

BehaviorBench: Benchmarking Foundation Models for Behavioral Science Tasks

Foundation models have been increasingly applied to behavioral science domains such as psychology, sociology, and economics. While these models show promise in tasks such as survey response prediction and human-subject experiment simulation, there remains no systematic understanding of how well they perform across diverse behavioral science tasks. We introduce BehaviorBench, a comprehensive benchmark that evaluates foundation models along four core capabilities: (1) behavior prediction and simulation, (2) strategic decision-making, (3) subject-trait inference, and (4) behavioral knowledge application. Crucially, BehaviorBench evaluates model outputs at both the individual and distributional levels, capturing not only per-subject accuracy but also population-level alignment, an essential requirement for behavioral validity. Our evaluation shows that BehaviorBench remains challenging for leading general-purpose LLMs and behavior foundation models that are specifically trained with behavioral data. We find that individual-level and distributional performance do not always align. General-purpose LLMs tend to underestimate the diversity of human responses, whereas behavior foundation models often lag behind at individual-level prediction. Our investigation further demonstrates how fine-tuning on diverse behavioral data can improve both individual-level prediction and distributional alignment, balancing these two objectives. Our results highlight the importance of evaluation at both individual and distributional levels, establishing BehaviorBench as a foundation for developing and assessing behaviorally aligned AI systems. Our BehaviorBench and models can be accessed via https://umich-foreseer.github.io/behaviorbench/

cs.CL↗
arXiv2026

HouseholdBench: Evaluating Large Language Models as Predictors of Household Economic Behavior

Large language models (LLMs) have the potential to meet a key goal in economics: a quantitative model of household decision making, across a variety of settings. Yet existing evaluations cover few surveys and outcomes, and do not study how households adjust to changing economic conditions. We introduce a new evaluation, HouseholdBench, which unites 6 U.S. household surveys and 32 prediction tasks spanning numeric, categorical and probabilistic outcomes, related to consumption, income, labor, expectations, and housing. Using past behavior, demographics and macroeconomic conditions, the tasks test whether LLMs predict behavior, including how households adjust to changes in various policies. We evaluate 13 proprietary and open-weight LLMs against a no-change baseline and a gradient-boosted tree model. Most LLMs outperform the no-change baseline, including for policy response tasks -- with the best model lowering error for numeric outcomes by 12.2%. Across most tasks, gradient-boosted trees rank first; leading proprietary LLMs approach their performance, but open-weight models lag. LLMs exhibit systematic over- and underprediction across different tasks. We identify methods that enable a 4 billion parameter open-weight model to match proprietary models' performance: fine-tuning and aggregating 16 predictions per observation. Improvements generalize to policy-response tasks, which are excluded from fine-tuning. We release our datasets, code, and leaderboard on our website: https://jn-huang.github.io/householdbench

cs.CL↗
arXiv2026

Benevolent Bias in Multi-Turn Human-Agent Dialogue

Bias in human-agent interaction can manifest not only through hostile language but also as benevolent bias, whereby unequal treatment hides behind a warm, positive tone. To make it detectable, we operationalise benevolent bias along two dimensions, tone and treatment, yielding three classes: neutral support, overt bias, and benevolent bias. Building on these definitions, we construct BENEVDIAL, a class-balanced corpus of 362,880 multi-turn support dialogues spanning user and agent demographics, roles, and generators, to support controlled evaluation. We then test two detector families on it: off-the-shelf safety detectors and prompted large language model (LLM) judges. Our findings reveal a notable detection gap: off-the-shelf detectors reliably flag overt bias yet largely fail to identify benevolent bias. LLM judges improve sensitivity when guided by explicit detection criteria, but this comes at the cost of increased misclassification of neutral supportive statements as benevolent bias, a tendency that is further exacerbated by the presence of demographic context. These findings suggest that fair monitoring of human-agent dialogue must look beyond surface cues to whether the agent's treatment is disparate.

cs.AI↗
arXiv2026

Complete-muE: Optimal Hyperparameter Transfer and Scaling for MoE Models

We propose Complete-muE, a framework which targets hyperparameter transfer across dense FFN and any Mixture-of-Experts (MoE) setups in transformer blocks. Existing tools such as $μ$P (requires fixed architectue) or SDE (requires fixed per-step token count) cannot directly solve the hyperparameter transfer problem in MoE setups because Dense to MoE transfer or MoE total experts scaling changes both architecture and tokens per expert. Complete-muE solves this challenge with a two-bridge system: Bridge~I maps between dense FFN and Dense MoE by active-width $μ$P with a normalized router scale. Bridge~II maps between Dense MoE and sparse MoE by activated-expert scaling, where the first-order SDE LR/WD correction cancels while a bounded residual $σ_0$ shift remains. The resulting transfer rule, which we term as Complete muE, covers changes in activated experts, total capacity, granularity, and shared/group-balanced hybrids for MoE models as well as network width/depth, batch size, and duration changes for general Transformer models. Extensive language model and diffusion model pretraining experiments confirm that complete-muE yields relatively stable hyperparameter optima across model architectures and parameter counts -- with only minor drift consistent with the non-strict SDE behavior of Bridge~II. In practice this drift is small enough that hyperparameters tuned on a single dense reference transfer near-optimally to all MoE configurations -- \emph{tune dense once, transfer to all} is the practical recipe at the core of Complete-muE. This enables MoE models to achieve accelerated convergence speedup over dense models when scaling model capacity without costly hyperparameter search.

cs.LG↗
arXiv2026

Robust and Generalizable Background Subtraction on Images of Calorimeter Jets using Unsupervised Generative Learning

Accurate separation of signal from background is one of the main challenges for precision measurements across high-energy and nuclear physics. Conventional supervised learning methods are insufficient here because the required paired signal and background examples are impossible to acquire in real experiments. Here, we introduce an unsupervised unpaired image-to-image translation neural network that learns to separate the signal and background from the input experimental data using cycle-consistency principles. We demonstrate the efficacy of this approach using images composed of simulated calorimeter data from the sPHENIX experiment, where physics signals (jets) are immersed in the extremely dense and fluctuating heavy-ion collision environment. Our method outperforms conventional subtraction algorithms in fidelity and overcomes the limitations of supervised methods. Furthermore, we evaluated the model's robustness in an out-of-distribution test scenario designed to emulate modified jets as in real experimental data. The model, trained on a simpler dataset, maintained its high fidelity on a more realistic, highly modified jet signal. This work represents the first use of unsupervised unpaired generative models for full detector jet background subtraction and offers a path for novel applications in real experimental data, enabling high-precision analyses across a wide range of imaging-based experiments.

nucl-ex↗
arXiv2026

ECP-Bench: Benchmarking and Learning Entertainment Content Promotion with Foundation Models

Content promotion spans a broad set of skills, from understanding content to forecasting its market reception. However, LLMs' ability to support such promotion decisions remains underexplored. Existing studies are often limited to a single task (e.g., popularity prediction) or a small set of tasks within a single domain (e.g., movies). As a result, there is a lack of understanding of LLMs' abilities in the full promotion process and how these abilities generalize across different tasks and domains. In this work, we introduce ECP-Bench, a benchmark containing 1.9M movie, game, and music items and 423,451 questions across 33 tasks in five content-promotion skill families. Our evaluation shows that frontier models achieve only 51.9\% overall accuracy and lose much of their advantage on post-cutoff content, with drops of up to 19.1 percentage points. In contrast, open-weight models fine-tuned on ECP-Bench achieve up to 60.3\%, remain substantially more stable across the knowledge cutoff, generalize to unseen content and tasks, and exhibit meaningful cross-domain generalization.

cs.IR↗
arXiv2026

Analytical Covariance Propagation for DVL-Aided Loosely Coupled SINS Under Attitude Uncertainty

In loosely coupled strapdown inertial navigation system/Doppler velocity log (SINS/DVL) integration, the body-frame DVL velocity is projected into the navigation frame using the SINS-computed attitude. Attitude uncertainty consequently affects both the projected velocity observation and its associated measurement covariance. This paper develops an analytical covariance propagation (ACP) method that treats these two effects consistently. The attitude-error-aware observation model represents perturbations in the projected DVL velocity, whereas the covariance construction propagates the body-frame DVL covariance at the matrix level and adds a closed-form term derived from the predicted attitude-error covariance, rather than directly rotating component-wise standard deviations. Simulation and a controlled surface-water field experiment compare ACP with conventional SINS/DVL and adaptive-covariance benchmarks. For the evaluated data sets and parameterizations, ACP yields the lowest position-error metrics among the compared methods. Relative to the variational Bayesian adaptive Kalman filter (VBAKF) in the field experiment, ACP reduces the northward, eastward, and downward position root-mean-square errors (RMSEs) by 48.5%, 55.0%, and 74.8%, respectively. Beyond DVL aiding, ACP provides a transferable covariance-construction principle for attitude-dependent vector-measurement fusion and a physically interpretable alternative to heuristic frame-dependent covariance assignment.

cs.RO↗
arXiv2026

Optimizing CUDA like a Human: Micro-Profiling Tools as Expert Surrogates for LLM-Based GPU Kernel Optimization

We present KernelPro, a closed-loop multi-agent system that automatically generates, profiles, and iteratively optimizes GPU kernel code by integrating large language model (LLM) code generation with hardware profiler feedback and pluggable bottleneck detection tools. KernelPro introduces four contributions: (1) a semantic feedback operator that encodes expert heuristics as pluggable micro-profiling tools, transforming raw hardware metrics into actionable natural language guidance; (2) a two-stage tool invocation architecture where roofline-based bottleneck classification filters which specialized analysis tools execute, combining kernel-level (ncu), instruction-level (SASS), and system-level (nsys) profiling; (3) a domain-adapted MCTS with progressive widening, asymmetric branching, log-reward calibration, dead-end pruning, and search memory for cross-iteration learning; and (4) direct CuTe source-level code generation via autonomous code search over the CUTLASS/CuTe codebase. On KernelBench, KernelPro achieves geometric mean speedups of 2.42x/4.69x/5.30x on Levels 1/2/3, establishing state-of-the-art performance across all difficulty levels. On VeOmni's expert-optimized MoE training kernels, KernelPro achieves 1.23x over hand-tuned Triton by generating a from-scratch raw-CUDA+CuTe Hopper WGMMA kernel. Ablation studies demonstrate that each design component independently and significantly improves optimization quality: micro-profiling tools (p < 0.0001 vs raw metrics), MCTS search (26% higher geometric mean vs greedy, p = 0.004), and proactive tool orchestration (23% improvement, p = 0.035). Finally, KernelPro is the first CUDA kernel coding agent to optimize energy efficiency beyond the speed-only focus of prior systems, demonstrating an 11.6% measured energy reduction at matched speed.

cs.LG↗
arXiv2026

Risk Occupancy: A New and Efficient Paradigm through Vehicle-Road-Cloud Collaboration

This paper proposes a novel 4D risk occupancy (RiskOcc) perception paradigm under the Vehicle-Road-Cloud integrated architecture, which unifies object detection and local mapping into a single representation spanning four dimensions: road-surface x and y coordinates, risk, and time. Distinct from conventional grid occupancy and risk field methods, this paradigm adopts an anchor-node-based perception approach with a concise yet reliable risk quantification scheme, enabling flexible and accurate capture of static and dynamic object occupancy states at current and future time steps, with per-occupancy-unit risk quantification. Compared with 3D-Occ, the proposed RiskOcc requires only one layer of data volume and achieves higher perceptual computational efficiency. Visualizations of risk occupancy perception results are presented based on the DAIR-V2X dataset, and a path planning method is developed to verify the usability of the risk occupancy map. Experimental results show that at an initial braking speed of 8 m/s, the model improves safety redundancy by 12.5\% and reduces average deceleration by 5.41\%, enhancing both safety and comfort. In summary, this study introduces a novel perception paradigm for intelligent connected vehicle (ICV), providing a compact and interpretable risk representation for cooperative perception and downstream planning in ICV systems.

cs.RO↗
arXiv2026

Automatic programming via large language models with population self-evolution for dynamic fuzzy job shop scheduling problem

Heuristic dispatching rules (HDRs) are widely used for solving the dynamic fuzzy job shop scheduling problem (DFJSSP). However, their performance is highly sensitive to specific scenarios and often necessitates expert customization. To overcome this, automated design methods like genetic programming (GP) and gene expression programming (GEP) have been proposed. Despite their success, these methods face challenges, such as high randomness in the search process. Recently, the combination of large language models (LLMs) with evolutionary algorithms has opened new possibilities for prompt engineering and automated algorithm design. To improve the ability of LLMs in automatic HDR design, this paper introduces a novel population self-evolutionary (SeEvo) framework, which draws inspiration from the self-reflective design strategies employed by human experts. Notably, this framework employs a novel teacher-student learning mechanism, allowing the LLM (student) to generate robust HDRs. Guided by a teacher model with complete knowledge of actual processing times, the student learns to infer fuzzy uncertainties from historical deviations, enabling it to effectively anticipate and adapt to fuzzy impacts. Experimental results demonstrate that SeEvo significantly outperforms GP, GEP, deep reinforcement learning (DRL) methods, and more than ten commonly used HDRs from the literature, particularly in previously unseen and dynamic scenarios.

cs.NE↗
arXiv2026

DSevolve: Enabling Real-Time Adaptive Scheduling on Dynamic Flexible Job Shop with LLM-Evolved Heuristic Portfolios

In dynamic flexible job shops, order arrivals, machine breakdowns, and processing-time deviations continually reshape the scheduling state and the priority trade-offs behind dispatching decisions. Dispatching rules are well suited to this setting because they are fast, interpretable, and easy to deploy, and recent LLM-assisted automatic heuristic design further expands their expressiveness by evolving composite priority functions. The key challenge is to make these evolved rule behaviors state-adaptive without losing the rapid response needed for online rescheduling. This paper proposes a dynamic self-evolutionary framework DSevolve, which separates offline rule-library construction from online state-conditioned rule selection. Offline, an LLM-guided quality-diversity search combines multi-persona seeding, a MAP-Elites behavioral archive, and behavior-guided variation to evolve a library of complementary rules rather than a single elite, and event-level simulation then trains a neural selector to rank the rules by state. Online, after each rescheduling event, a neural selector maps a 22-dimensional state fingerprint to rule scores and dispatches the top-ranked rule within about a second, meeting the response-time requirement after each disruption. Experiments on dynamic instances derived from standard flexible job shop benchmarks show that DSevolve achieves lower mean makespan than individual LLM-evolved rules, classical dispatching rules, and learning-based baselines under a one-active-rule deployment protocol. Trained only on small instances, the selector transfers zero-shot to substantially larger dynamic shops. These results show that state-conditioned selection preserves the speed and interpretability of dispatching rules while improving adaptability.

cs.AI↗
arXiv2026

StARS: Socially Appropriate Robot Actions via a Recommender System-Driven Approach

Social appropriateness in human-robot interaction (HRI) is not universal: different people can judge the same robot action differently in the same situation. To capture this inter-subject variability, we reformulate socially appropriate action generation as a preference modelling problem inspired by recommender systems, treating annotators as users, contexts/scenes as items, and appropriateness scores over a set of candidate robot actions as targets. We propose StARS, a novel model-agnostic framework that integrates collaborative filtering with learnable scene representations to generate user-specific appropriateness scores over candidate robot actions. StARS is model-agnostic: it can be integrated with various scene encoders and backbones, enabling personalisation without redesigning the underlying model. We evaluate StARS on two socially aware robotics datasets, MannersDB+ and SocNav1, and analyse robustness under sparse preference feedback. Across datasets and backbones, StARS consistently improves performance and agreement with annotators, supporting personalised action selection aligned with user norms. Our code is publicly available at https://github.com/Cambridge-AFAR/StARS.git.

cs.RO↗
arXiv2026

Idea2Plan: Exploring AI-Powered Research Planning

Large language models (LLMs) have demonstrated significant potential to accelerate scientific discovery as valuable tools for analyzing data, generating hypotheses, and supporting innovative approaches in various scientific fields. In this work, we investigate how LLMs can handle the transition from conceptual research ideas to well-structured research plans. Effective research planning not only supports scientists in advancing their research but also represents a crucial capability for the development of autonomous research agents. Despite its importance, the field lacks a systematic understanding of LLMs' research planning capability. To rigorously measure this capability, we introduce the Idea2Plan task and Idea2Plan Bench, a set of benchmarks built from ICML 2025 and Nature Mental Health papers released after major LLM training cutoffs. Each benchmark instance includes a research idea and a grading rubric capturing the key components of valid plans. We further propose Idea2Plan JudgeEval, a complementary benchmark to assess the reliability of LLM-based judges against expert annotations. Experimental results show that GPT-5 achieves the strongest performance on the benchmark, though substantial headroom remains for improvement. Our study provides new insights into LLMs' capability for research planning and lays the groundwork for future progress.

cs.CL↗
arXiv2026

Aligning Clinical Needs and AI Capabilities: A Survey on LLMs for Medical Reasoning

Large language models (LLMs) have emerged as important tools in healthcare, showing growing potential for clinical reasoning and patient care. This survey examines recent progress in medical LLMs, focusing on reasoning applications and requirements. We present a dual-view approach that connects clinical practice with computational methods. On the clinical side, we establish a five-level competency scheme following Miller's Pyramid, progressing from knowledge recall to dynamic case management. On the computational side, we link deductive, inductive, and abductive reasoning patterns to common medical goals and tasks. We also introduce a benchmark dataset spanning five levels of medical reasoning capability and report results on 18 state-of-the-art models, revealing that medical specialist models excel in diagnosis-centric tasks while general models lead in decision support and dialogue. We conclude by discussing current progress and open challenges, including data limitations, hallucination, and grounding issues, and outline directions toward safer, more reliable, and workflow-ready systems.

cs.AI↗
arXiv2026

Toward Personalized Social Robots for Child Well-being: Data Requirement Principles from a Recommender-System Perspective

Social robots are increasingly deployed in clinical settings to support the well-being of children, where effective support must be personalized to each child. Personalization, choosing the robot action best suited to each child, can be framed as a recommendation problem, and a recently proposed recommender-system framework for social robots offers a principled approach through user profiling, ranking, and responsible computing. Instantiating it, however, is blocked not by the model but by the data, which is hard to gather. A child's state shifts within and across visits, so no fixed description of the user holds. Within a session, the few signals of whether the robot's actions helped are weak and indirect. Across sessions, children are rarely seen more than once, and anonymization breaks the identity needed to link visits. Because care cannot be randomized, existing data is observational, biased toward whatever was already done. Each is a familiar recommender-system problem, and we propose four data principles in response: an integrated profile, effectiveness signals, linkable coverage, and an exposure record logged at collection time. We identify which of these principles each capability requires, and frame them as concrete guidelines for data collection.

cs.HC↗
arXiv2026

Natural Language Camera Movement Understanding

Understanding camera movement in natural language is critical for training and evaluating video generation models, among other applications. However, we demonstrate that existing vision-language models (VLMs) fail this task in surprising ways, frequently confusing translation with rotation, left with right, and object movement with camera movement. To address these limitations, we establish natural language camera movement understanding as a standalone research task. We introduce a two-level cinematographic taxonomy and an extensive, atomic benchmark featuring both real and synthetic videos. Furthermore, we curate a large-scale, multi-source training set enhanced by targeted camera movement augmentation. Our fine-tuned VLM-8B outperforms Gemini 3.1 Pro by 10% and 11% on our benchmark's real and synthetic videos, respectively. Despite these gains, a significant gap remains relative to human performance, underscoring the need to promote and facilitate future research on natural language camera movement understanding.

cs.CV↗