arXiv subjects

Jian Wang

Publications and source records attributed to Jian Wang.

At least 307 records · Page 17Linked to original sources

arXiv2025

R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science

Recent advances in AI and ML have transformed data science, yet increasing complexity and expertise requirements continue to hinder progress. Although crowd-sourcing platforms alleviate some challenges, high-level machine learning engineering (MLE) tasks remain labor-intensive and iterative. We introduce R&D-Agent, a comprehensive, decoupled, and extensible framework that formalizes the MLE process. R&D-Agent defines the MLE workflow into two phases and six components, turning agent design for MLE from ad-hoc craftsmanship into a principled, testable process. Although several existing agents report promising gains on their chosen components, they can mostly be summarized as a partial optimization from our framework's simple baseline. Inspired by human experts, we designed efficient and effective agents within this framework that achieve state-of-the-art performance. Evaluated on MLE-Bench, the agent built on R&D-Agent ranks as the top-performing machine learning engineering agent, achieving 35.1% any medal rate, demonstrating the ability of the framework to speed up innovation and improve accuracy across a wide range of data science applications. We have open-sourced R&D-Agent on GitHub: https://github.com/microsoft/RD-Agent.

cs.AI↗

arXiv2025

Reinforced Strategy Optimization for Conversational Recommender Systems via Network-of-Experts

Conversational Recommender Systems (CRSs) aim to provide personalized recommendations through multi-turn natural language interactions with users. Given the strong interaction and reasoning skills of Large Language Models (LLMs), leveraging LLMs for CRSs has recently emerged as a promising direction. However, existing LLM-based methods often lack explicit optimization of interaction strategies, instead relying on unified prompts and the LLM's internal knowledge to decide how to interact, which can lead to suboptimal outcomes. In this paper, we propose a novel Reinforced Strategy Optimization (RSO) method for CRS, which decomposes the process of generating strategy-driven response decisions into the macro-level strategy planning and micro-level strategy adaptation through a network-of-experts architecture. At the macro level, a Planner expert selects macro-level interaction strategies (e.g., recommend, explain, encourage). At the micro level, an Actor expert generates detailed responses conditioned on the selected macro-level strategy, guided by auxiliary experts that provide complementary information such as user preferences and factual grounding. This hierarchical decomposition disentangles the optimization of different sub-tasks involved in CRS response generation, enabling more tractable learning at each level. To address the scarcity of high-quality multi-turn training data, we formulate strategy learning as a reinforcement learning problem, guided by an LLM-based reward model to achieve automatic strategy exploration. Extensive experiments show that RSO significantly improves interaction performance compared to state-of-the-art baselines, demonstrating the effectiveness of explicit hierarchical strategy optimization for CRS.

cs.CL↗

arXiv2025

Lyapunov exponents and growth indices for fractional stochastic heat equations with space-time Lévy white noise

We consider fractional stochastic heat equations with space-time Lévy white noise of the form $$\frac{\partial X}{\partial t}(t,x)={\cal L}_αX(t,x)+σ(X(t,x))\dotΛ(t,x).$$ Here, the principal part ${\cal L}_α=-(-Δ)^{α/2}$ is the $d$-dimensional fractional Laplacian with $α\in (0,2)$, the noise term $\dotΛ(t,x)$ denotes the space-time Lévy white noise, and the function $σ: \R\mapsto \R$ is Lipschitz continuous. Under suitable assumptions, we obtain bounds for the Lyapunov exponents and the growth indices of exponential type on $p$th moments of the mild solutions, which are connected with the weakly intermittency properties and the characterizations of the high peaks propagate away from the origin. Unlike the case of the Gaussian noise, the proofs heavily depend on the heavy tail property of heat kernel estimates for the fractional Laplacian. The results complement these in \cite{CD15-1,CK19} for fractional stochastic heat equations driven by space-time white noise and stochastic heat equations with Lévy noise, respectively.

math.PR↗

arXiv2025

T2Bs: Text-to-Character Blendshapes via Video Generation

We present T2Bs, a framework for generating high-quality, animatable character head morphable models from text by combining static text-to-3D generation with video diffusion. Text-to-3D models produce detailed static geometry but lack motion synthesis, while video diffusion models generate motion with temporal and multi-view geometric inconsistencies. T2Bs bridges this gap by leveraging deformable 3D Gaussian splatting to align static 3D assets with video outputs. By constraining motion with static geometry and employing a view-dependent deformation MLP, T2Bs (i) outperforms existing 4D generation methods in accuracy and expressiveness while reducing video artifacts and view inconsistencies, and (ii) reconstructs smooth, coherent, fully registered 3D geometries designed to scale for building morphable models with diverse, realistic facial motions. This enables synthesizing expressive, animatable character heads that surpass current 4D generation techniques.

cs.GR↗

arXiv2025

A Benchmark for Localizing Code and Non-Code Issues in Software Projects

Accurate project localization (e.g., files and functions) for issue resolution is a critical first step in software maintenance. However, existing benchmarks for issue localization, such as SWE-Bench and LocBench, are limited. They focus predominantly on pull-request issues and code locations, ignoring other evidence and non-code files such as commits, comments, configurations, and documentation. To address this gap, we introduce MULocBench, a comprehensive dataset of 1,100 issues from 46 popular GitHub Python projects. Comparing with existing benchmarks, MULocBench offers greater diversity in issue types, root causes, location scopes, and file types, providing a more realistic testbed for evaluation. Using this benchmark, we assess the performance of state-of-the-art localization methods and five LLM-based prompting strategies. Our results reveal significant limitations in current techniques: even at the file level, performance metrics (Acc@5, F1) remain below 40%. This underscores the challenge of generalizing to realistic, multi-faceted issue resolution. To enable future research on project localization for issue resolution, we publicly release MULocBench at https://huggingface.co/datasets/somethingone/MULocBench.

cs.SE↗

arXiv2025

Do Code Semantics Help? A Comprehensive Study on Execution Trace-Based Information for Code Large Language Models

Code Large Language Models (Code LLMs) have opened a new era in programming with their impressive capabilities. However, recent research has revealed critical limitations in their ability to reason about runtime behavior and understand the actual functionality of programs, which poses significant challenges for their post-training and practical deployment. Specifically, Code LLMs encounter two principal issues: (1) a lack of proficiency in reasoning about program execution behavior, as they struggle to interpret what programs actually do during runtime, and (2) the inconsistent and fragmented representation of semantic information, such as execution traces, across existing methods, which hinders their ability to generalize and reason effectively. These challenges underscore the necessity for more systematic approaches to enhance the reasoning capabilities of Code LLMs. To address these issues, we introduce a generic framework to support integrating semantic information~(e.g., execution trace) to code task-relevant prompts, and conduct a comprehensive study to explore the role of semantic information in enhancing the reasoning ability of Code LLMs accordingly. Specifically, we focus on investigating the usefulness of trace-based semantic information in boosting supervised fine-tuning~(SFT) and post-phase inference of Code LLMs. The experimental results surprisingly disagree with previous works and demonstrate that semantic information has limited usefulness for SFT and test time scaling of Code LLM.

cs.SE↗

arXiv2025

Spectral Torsion for Nonminimal de-Rham Hodge Operator

In this paper, we investigate some new spectral torsion which is the extension of spectral torsion for Dirac operators, and compute the spectral torsion associated with nonminimal de Rham-Hodge operators on manifolds with (or without) boundary. The novelty of this paper lies in the calculation related to the noncommutative residue for the inverse of the Laplacion with non-scalar principle symbols.

math-ph↗

arXiv2025

Evolution of internal waves in a 2D subcritical channel

We study the long time evolution of internal waves in two dimensional subcritical channels with flat horizontal ends. We show the leading profiles of solutions are the outgoing solutions to the stationary equations. This is done by showing a limiting absorption principle and the differentiability of the spectral measure for a zeroth order operator.

math.AP↗

arXiv2025

On the Matching Problem in Random Hypergraphs

We study a variant of the Erdős Matching Problem in random hypergraphs. Let $\mathcal{K}_p(n,k)$ denote the Erdős-Rényi random $k$-uniform hypergraph on $n$ vertices where each possible edge is included with probability $p$. We show that when $n\gg k^{2}s$ and $p$ is not too small, with high probability, the maximum number of edges in a sub-hypergraph of $\mathcal{K}_p(n,k)$ with matching number $s$ is obtained by the trivial sub-hypergraphs, i.e. the sub-hypergraph consisting of all edges containing at least one vertex in a fixed set of $s$ vertices.

math.CO↗

arXiv2025

Self-Rewarding Rubric-Based Reinforcement Learning for Open-Ended Reasoning

Open-ended evaluation is essential for deploying large language models in real-world settings. In studying HealthBench, we observe that using the model itself as a grader and generating rubric-based reward signals substantially improves reasoning performance. Remarkably, the trained model also becomes a stronger grader. Motivated by this, we introduce Self-Rewarding Rubric-Based Reinforcement Learning for Open-Ended Reasoning, a lightweight framework that enables faster and more resource-efficient training while surpassing baselines. Remarkably, on Qwen3-32B, training with just the 4000-sample HealthBench Easy subset is sufficient to obtain a model that exceeds GPT-5 on HealthBench Hard. Incorporating a small amount of teacher-graded data further enhances performance for less capable models.

cs.CL↗

arXiv2025

An Empirical Study of Federated Prompt Learning for Vision Language Model

The Vision Language Model (VLM) excels in aligning vision and language representations, and prompt learning has emerged as a key technique for adapting such models to downstream tasks. However, the application of prompt learning with VLM in federated learning (FL) scenarios remains underexplored. This paper systematically investigates the behavioral differences between language prompt learning (LPT) and vision prompt learning (VPT) under data heterogeneity challenges, including label skew and domain shift. We conduct extensive experiments to evaluate the impact of various FL and prompt configurations, such as client scale, aggregation strategies, and prompt length, to assess the robustness of Federated Prompt Learning (FPL). Furthermore, we explore strategies for enhancing prompt learning in complex scenarios where label skew and domain shift coexist, including leveraging both prompt types when computational resources allow. Our findings offer practical insights into optimizing prompt learning in federated settings, contributing to the broader deployment of VLMs in privacy-preserving environments.

cs.LG↗

arXiv2025

Analytic Result of Higgs Boson Decay to Gluons with Full Quark Mass Dependence

We report the first analytic calculation of the Higgs boson decay width to gluons up to next-to-leading order in quantum chromodynamics including full dependence on the bottom and charm quark masses. The interference between top- and bottom/charm-quark-induced amplitudes provides unexpectedly large contributions because the large logarithms of the bottom (charm) quark masses over the Higgs boson mass compensate for the suppression of the Yukawa coupling. These logarithms exhibit a unique structure distinct from Sudakov resummation and more complex than the Higgs-gluon-gluon form factor. Our results demonstrate that the previous calculation with massless bottom and charm quarks overestimates the decay width by nearly 20%. The prediction of the partial decay width with massive quarks provides essential theory input needed in the precise determination of Higgs couplings at future colliders.

hep-ph↗

arXiv2025

Signed Graph Learning with Hidden Nodes

Signed graphs, which are characterized by both positive and negative edge weights, have recently attracted significant attention in the field of graph signal processing (GSP). Existing works on signed graph learning typically assume that all graph nodes are available. However, in some specific applications, only a subset of nodes can be observed while the remaining nodes stay hidden. To address this challenge, we propose a novel method for identifying signed graph that accounts for hidden nodes, termed \textit{signed graph learning with hidden nodes under column-sparsity regularization} (SGL-HNCS). Our method is based on the assumption that graph signals are smooth over signed graphs, i.e., signal values of two nodes connected by positive (negative) edges are similar (dissimilar). Rooted in this prior assumption, the topology inference of a signed graph is formulated as a constrained optimization problem with column-sparsity regularization, where the goal is to reconstruct the signed graph Laplacian matrix without disregarding the influence of hidden nodes. We solve the constrained optimization problem using a tailored block coordinate descent (BCD) approach. Experimental results using synthetic data and real-world data demonstrate the efficiency of the proposed SGL-HNCS method.

eess.SP↗

arXiv2025

HIRAG: Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation

Retrieval-augmented generation (RAG) has become a fundamental paradigm for addressing the challenges faced by large language models in handling real-time information and domain-specific problems. Traditional RAG systems primarily rely on the in-context learning (ICL) capabilities of the large language model itself. Still, in-depth research on the specific capabilities needed by the RAG generation model is lacking, leading to challenges with inconsistent document quality and retrieval system imperfections. Even the limited studies that fine-tune RAG generative models often \textit{lack a granular focus on RAG task} or \textit{a deeper utilization of chain-of-thought processes}. To address this, we propose that RAG models should possess three progressively hierarchical abilities (1) Filtering: the ability to select relevant information; (2) Combination: the ability to combine semantic information across paragraphs; and (3) RAG-specific reasoning: the ability to further process external knowledge using internal knowledge. Thus, we introduce our new RAG instruction fine-tuning method, Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation (HIRAG) incorporates a "think before answering" strategy. This method enhances the model's open-book examination capability by utilizing multi-level progressive chain-of-thought. Experiments show that the HIRAG training strategy significantly improves the model's performance on datasets such as RGB, PopQA, MuSiQue, HotpotQA, and PubmedQA.

cs.CL↗

arXiv2025

Quenched and annealed heat kernel estimates for Brox's diffusion

Brox's diffusion is a typical one-dimensional singular diffusion, which was introduced by Brox (1986) as a continuous analogue of Sinai's random walk. In this paper, we will establish quenched heat kernel estimates for short time and annealed heat kernel estimates for large time of Brox's diffusion. The proofs are based on Brox's construction via the scale-transformation and the time-change arguments as well as the theory of resistance forms for symmetric strongly recurrent Markov processes. We emphasize that, since the reference measure of Brox's diffusion does not satisfy the so-called volume doubling conditions neither for the small scale nor the large scale, the existing methods for heat kernel estimates of diffusions in ergodic media do not work, and new techniques will be introduced to establish both quenched and annealed heat kernel estimates of Brox's diffusions, which take into account different oscillation properties for one-dimensional Brownian motion in random environments.

math.PR↗

arXiv2025

Benchmarking for Domain-Specific LLMs: A Case Study on Academia and Beyond

The increasing demand for domain-specific evaluation of large language models (LLMs) has led to the development of numerous benchmarks. These efforts often adhere to the principle of data scaling, relying on large corpora or extensive question-answer (QA) sets to ensure broad coverage. However, the impact of corpus and QA set design on the precision and recall of domain-specific LLM performance remains poorly understood. In this paper, we argue that data scaling is not always the optimal principle for domain-specific benchmark construction. Instead, we introduce Comp-Comp, an iterative benchmarking framework grounded in the principle of comprehensiveness and compactness. Comprehensiveness ensures semantic recall by covering the full breadth of the domain, while compactness improves precision by reducing redundancy and noise. To demonstrate the effectiveness of our approach, we present a case study conducted at a well-renowned university, resulting in the creation of PolyBench, a large-scale, high-quality academic benchmark. Although this study focuses on academia, the Comp-Comp framework is domain-agnostic and readily adaptable to a wide range of specialized fields. The source code and datasets can be accessed at https://github.com/Anya-RB-Chen/COMP-COMP.

cs.AI↗

arXiv2025

HANRAG: Heuristic Accurate Noise-resistant Retrieval-Augmented Generation for Multi-hop Question Answering

The Retrieval-Augmented Generation (RAG) approach enhances question-answering systems and dialogue generation tasks by integrating information retrieval (IR) technologies with large language models (LLMs). This strategy, which retrieves information from external knowledge bases to bolster the response capabilities of generative models, has achieved certain successes. However, current RAG methods still face numerous challenges when dealing with multi-hop queries. For instance, some approaches overly rely on iterative retrieval, wasting too many retrieval steps on compound queries. Additionally, using the original complex query for retrieval may fail to capture content relevant to specific sub-queries, resulting in noisy retrieved content. If the noise is not managed, it can lead to the problem of noise accumulation. To address these issues, we introduce HANRAG, a novel heuristic-based framework designed to efficiently tackle problems of varying complexity. Driven by a powerful revelator, HANRAG routes queries, decomposes them into sub-queries, and filters noise from retrieved documents. This enhances the system's adaptability and noise resistance, making it highly capable of handling diverse queries. We compare the proposed framework against other leading industry methods across various benchmarks. The results demonstrate that our framework obtains superior performance in both single-hop and multi-hop question-answering tasks.

cs.CL↗

arXiv2025

KABB: Knowledge-Aware Bayesian Bandits for Dynamic Expert Coordination in Multi-Agent Systems

As scaling large language models faces prohibitive costs, multi-agent systems emerge as a promising alternative, though challenged by static knowledge assumptions and coordination inefficiencies. We introduces Knowledge-Aware Bayesian Bandits (KABB), a novel framework that enhances multi-agent system coordination through semantic understanding and dynamic adaptation. The framework features three key innovations: a three-dimensional knowledge distance model for deep semantic understanding, a dual-adaptation mechanism for continuous expert optimization, and a knowledge-aware Thompson Sampling strategy for efficient expert selection. Extensive evaluation demonstrates KABB achieves an optimal cost-performance balance, maintaining high performance while keeping computational demands relatively low in multi-agent coordination.

cs.AI↗