arXiv subjects

Xinyu Hu

Publications and source records attributed to Xinyu Hu.

At least 19 recordsLinked to original sources

arXiv2026

TAPO: Transition-Aware Policy Optimization for LLM Agents

Recently, Reinforcement Learning (RL) has emerged as a crucial paradigm for the post-training of Large Language Model (LLM) agents. However, existing methods predominantly rely on sparse task rewards for policy optimization, failing to fully exploit another class of inherently dense supervisory signals naturally present during online interaction: environmental feedback following action execution. Recent theoretical studies suggest that generalization in multi-step, goal-oriented tasks hinges on predictive knowledge of environmental consequences. Inspired by this, we propose TAPO: Transition-Aware Policy Optimization for LLM Agents, a unified training framework that alternates between policy optimization and transition supervision. Beyond standard RL updates, TAPO repurposes rollout data to apply action-conditioned next-observation prediction supervision on a shared backbone model. This approach enhances the model's sensitivity to environmental transition dynamics and action consequences while concurrently optimizing the policy. It serves as a computationally lightweight, plug-and-play enhancement module for existing agent RL algorithms, requiring no additional expert data, extra sampling costs, or inference-time overhead. We conduct systematic experiments on WebShop and ALFWorld, integrating foundation models of various scales with different policy optimization algorithms. Empirical results demonstrate that TAPO consistently improves task performance over pure policy optimization baselines.

cs.LG↗

arXiv2026

New Tower-Type Lower Bounds for Hypergraph Ramsey Numbers

The Ramsey number $r_k(s,m)$ is the smallest $N$ such that any red/blue coloring of the $k$-subsets of $[N]$ contains a red $s$-set or a blue $m$-set. For fixed $k$ and $s$, and for sufficiently large $m$, the tower growth rate is determined by the stepping-up lemma, but for $s=m=k+1$ the available stepping-up lemmas do not apply. Fox asked for estimates of $r_k(k+1,k+1)$. Pudl\'ak, R\"odl, and Wesley gave the first tower-type bound: $r_k(k+1,k+1)\ge s_3(\lfloor k/4\rfloor)\ge 4\operatorname{twr}_{\lfloor k/4\rfloor-4}(2)$, where $s_3(k)$ is the $3$-color shift number and $\operatorname{twr}_1(2)=2$, $\operatorname{twr}_{i+1}(2)=2^{\operatorname{twr}_i(2)}$. In this paper, for $k\ge 6$, we improve the lower bound to $r_k(k+1,k+1)> s_3\bigl(\lfloor k/2\rfloor-2\bigr)$ by overcoming an obstruction in their construction. In addition, we give an exact characterization of $s_3(k)$ and, for $k\ge 5$, obtain a new explicit lower bound $s_3(k)\ge(\operatorname{twr}_{k-2}(2))^2$, which improves the result of Pudl\'ak and R\"odl. Consequently, for $k\ge 14$, $r_k(k+1,k+1)>(\operatorname{twr}_{\lfloor k/2\rfloor-4}(2))^2$.

math.CO↗

arXiv2026

SentTrack: Sentiment-Driven Bottleneck Detection in GitHub Issue Repositories

Software engineering teams increasingly depend on GitHub issue threads to coordinate work, report bugs, and negotiate technical decisions, yet most repository health tools focus on code metrics and ignore the conversational dynamics that drive or stall development. This paper presents SentTrack, a dual-lens framework for detecting socio-technical bottlenecks from GitHub issue discussions. Applied to the AvaloniaUI open-source repository across approximately 9,000 issue threads, the framework addresses three questions: how to automate workflow-inefficiency detection from real-time conversational data, whether sentiment signals can surface risk earlier than traditional label-based methods, and how to isolate human narrative from machine-generated noise in mixed-media issue text. SentTrack combines two complementary pipelines. A horizontal pipeline translates raw issue reports into clean summaries using a large language model, extracts mid-level concern phrases, and clusters them through UMAP and HDBSCAN, producing 613 semantic clusters from the first 3,608 issues processed. A vertical pipeline applies the ABCDE collaborative interaction framework to classify each comment and infer thread-level outcomes. Across the full corpus, 49\% of threads ended in stagnation and only 13\% reached resolution, with the resolution gap identified as the dominant bottleneck signal. A weighted scoring engine that combines negativity, stagnation, resolution gap, and thread length gives maintainers an interpretable prioritization tool for high-friction discussions before they stall development.

cs.SE↗

arXiv2026

AgenticVBench: Can AI Agents Complete Real-World Post-Production Tasks?

Video production workflows offer a rich and demanding arena for evaluating multimodal AI agents: they require composite capabilities across text, image, audio, and video understanding, along with long-horizon planning, and tool use. To this end, we introduce AgenticVBench, a benchmark of 100 agentic tasks across 4 task families spanning the real world post-production workflow, constructed from real production workflows contributed by 20 industry experts averaging 6 years of professional experience. Tasks are paired with evaluation specifications that combine programmatic verifiers and expert rubrics. We evaluate frontier vision-language models (VLMs) with both vendor-native and open-source harnesses. The best evaluated agent stack barely crosses 30%, far below human expert performance on the same tasks. We further find that the choice of harness substantially affects model behavior, including scores, tool-use patterns, and failure modes. AgenticVBench provides a foundation for diagnosing and improving both models and harnesses for agentic video production. Benchmark website: https://agenticvbench.com.

cs.CR↗

arXiv2026

Modulation Consistency-based Contrastive Learning for Self-Supervised Automatic Modulation Classification

Deep learning-based AMC methods have achieved remarkable performance, but their practical deployment remains constrained by the high cost of labeled data. Although self-supervised learning (SSL) reduces the reliance on labels, existing SSL-based AMC methods often rely on task-agnostic pretext objectives misaligned with modulation classification, leading to representations entangled with nuisance factors such as symbol, channel, and noise. In this paper, we identify intra-instance modulation consistency as a task-aware structural prior, whereby different temporal segments of the same signal may differ in waveform while preserving the same modulation type, thus providing a principled cue for task-aligned self-supervision. Based on this prior, we propose Mod-CL, a Modulation consistency-based Contrastive Learning framework that constructs positive pairs from different temporal segments of the same signal instance, to encourage the model to learn shared modulation information while suppressing nuisance variations. We further develop a contrastive objective tailored to Mod-CL, which jointly exploits temporal segmentation and data augmentation to pull together views sharing the same modulation semantics while avoiding supervisory conflicts within each signal instance. Extensive experiments on RadioML datasets show that Mod-CL consistently outperforms strong baselines, especially in low-label regimes, achieving substantial improvements in linear probing accuracy.

eess.SP↗

arXiv2026

A double-exponential lower bound for $r_4(5,n)$

The Ramsey number $r_k(s,n)$ is the smallest integer $N$ such that every $N$-vertex $k$-graph contains either a copy of $K_s^{(k)}$ or an independent set of size $n$. We prove that $r_4(5,n)\ge 2^{2^{cn^{1/7}}}$, where $c>0$ is an absolute constant. As a consequence, we determine the tower growth rate of $r_k(k+1,n)$, which completely solves the problem of establishing the tower growth rate for all classical off-diagonal hypergraph Ramsey numbers, first posed by Erd\H{o}s and Hajnal in 1972.

math.CO↗

arXiv2026

A Note on Generalized Erd\H{o}s-Rogers Problems

For a $k$-uniform hypergraph $F$ and positive integers $s$ and $N$, the generalized Erd\H{o}s-Rogers function $f^{(k)}_{F,s}(N)$ denotes the largest integer $m$ such that every $K_s^{(k)}$-free $k$-graph on $N$ vertices contains an $F$-free induced subgraph on $m$ vertices. In particular, if $F = K^{(k)}_t$, then we write $f^{(k)}_{t,s}(N)$ for $f^{(k)}_{F,s}(N)$. Mubayi and Suk (\emph{J. London. Math. Soc. 2018}) conjectured that $f^{(4)}_{5,6}(N)=(\log \log N)^{\Theta(1)}$. Motivated by this conjecture, we prove that $f^{(4)}_{5^{-},6}(N)=(\log\log N)^{\Theta(1)}$, where $5^{-}$ denotes the $4$-graph obtained from $K_5^{(4)}$ by deleting one edge. Our proof combines a probabilistic construction of a $2$-coloring of pairs with a stepping-up construction and an analysis of multi-layer local extremum structures. Furthermore, we derive an upper bound for a more general Erd\H{o}s-Rogers function, which implies the lower bound $r_4(6,n)\ge 2^{2^{cn^{1/2}}}$. By applying a variant of the Erd\H{o}s-Hajnal stepping-up lemma due to Mubayi and Suk, we also slightly improve the lower bound for $r_k(k+2,n)$.

math.CO↗

arXiv2026

A step towards the Erd\H{o}s-Rogers problem

For $2\le k\le t<s$, the Erd\H{o}s-Rogers function $f^{(k)}_{t,s}(N)$ denotes the largest $m$ such that every $K^{(k)}_s$-free $k$-graph on $N$ vertices contains a $K^{(k)}_t$-free induced subgraph on $m$ vertices. Mubayi and Suk (J. London Math. Soc. 2018) conjectured that $f^{(k)}_{k+1,k+2}(N)=(\log_{(k-2)}N)^{\Theta(1)}$ for $k\ge 4$, where $\log_{(i)}$ denotes the $i$-fold iterated logarithm. This is equivalent to the statement that $f^{(k)}_{k+1,s}(N)=(\log_{(k-2)}N)^{\Theta(1)}$ for every $s\ge k+2$. In this paper, we introduce multi-color patterns into a random construction of a $2$-graph to build a $4$-graph, and for the first time, combine them with multi-layer extremum structures to prove that $f^{(4)}_{5,s}(N)=(\log \log N)^{\Theta(1)}$ for every $s\ge 11$. More generally, using a variant of the Erd\H{o}s-Hajnal stepping-up lemma, we also establish that $f^{(k)}_{k+1,s}(N)=(\log_{(k-2)}N)^{\Theta(1)}$ for every $s\ge k+7$.

math.CO↗

arXiv2026

Spectral bounds for the independence number of graphs and even uniform hypergraphs

In this paper, we give spectral upper bounds for the independence number of even uniform hypergraphs and graphs, extend the Hoffman bound to even uniform hypergraphs, and give a simple spectral condition for determining the independence number, the Shannon capacity and the Lov\'{a}sz number of a graph. The Hoffman bound on the Lov\'{a}sz number is also extended from regular graphs to general graphs.

math.CO↗

arXiv2026

CE-RM: A Pointwise Generative Reward Model Optimized via Two-Stage Rollout and Unified Criteria

Automatic evaluation is crucial yet challenging for open-ended natural language generation, especially when rule-based metrics are infeasible. Compared with traditional methods, the recent LLM-as-a-Judge paradigms enable better and more flexible evaluation, and show promise as generative reward models for reinforcement learning. However, prior work has revealed a notable gap between their seemingly impressive benchmark performance and actual effectiveness in RL practice. We attribute this issue to some limitations in existing studies, including the dominance of pairwise evaluation and inadequate optimization of evaluation criteria. Therefore, we propose CE-RM-4B, a pointwise generative reward model trained with a dedicated two-stage rollout method, and adopting unified query-based criteria. Using only about 5.7K high-quality data curated from the open-source preference dataset, our CE-RM-4B achieves superior performance on diverse reward model benchmarks, especially in Best-of-N scenarios, and delivers more effective improvements in downstream RL practice.

cs.CL↗

arXiv2025

Deep Learning--Accelerated Multi-Start Large Neighborhood Search for Real-time Freight Bundling

Online Freight Exchange Systems (OFEX) play a crucial role in modern freight logistics by facilitating real-time matching between shippers and carrier. However, efficient combinatorial bundling of transporation jobs remains a bottleneck. We model the OFEX combinatorial bundling problem as a multi-commodity one-to-one pickup-and-delivery selective traveling salesperson problem (m1-PDSTSP), which optimizes revenue-driven freight bundling under capacity, precedence, and route-length constraints. The key challenge is to couple combinatorial bundle selection with pickup-and-delivery routing under sub-second latency. We propose a learning--accelerated hybrid search pipeline that pairs a Transformer Neural Network-based constructive policy with an innovative Multi-Start Large Neighborhood Search (MSLNS) metaheuristic within a rolling-horizon scheme in which the platform repeatedly freezes the current marketplace into a static snapshot and solves it under a short time budget. This pairing leverages the low-latency, high-quality inference of the learning-based constructor alongside the robustness of improvement search; the multi-start design and plausible seeds help LNS to explore the solution space more efficiently. Across benchmarks, our method outperforms state-of-the-art neural combinatorial optimization and metaheuristic baselines in solution quality with comparable time, achieving an optimality gap of less than 2\% in total revenue relative to the best available exact baseline method. To our knowledge, this is the first work to establish that a Deep Neural Network-based constructor can reliably provide high-quality seeds for (multi-start) improvement heuristics, with applicability beyond the \textit{m1-PDSTSP} to a broad class of selective traveling salesperson problems and pickup and delivery problems.

cs.AI↗

arXiv2025

SCOPE: Intrinsic Semantic Space Control for Mitigating Copyright Infringement in LLMs

Large language models sometimes inadvertently reproduce passages that are copyrighted, exposing downstream applications to legal risk. Most existing studies for inference-time defences focus on surface-level token matching and rely on external blocklists or filters, which add deployment complexity and may overlook semantically paraphrased leakage. In this work, we reframe copyright infringement mitigation as intrinsic semantic-space control and introduce SCOPE, an inference-time method that requires no parameter updates or auxiliary filters. Specifically, the sparse autoencoder (SAE) projects hidden states into a high-dimensional, near-monosemantic space; benefiting from this representation, we identify a copyright-sensitive subspace and clamp its activations during decoding. Experiments on widely recognized benchmarks show that SCOPE mitigates copyright infringement without degrading general utility. Further interpretability analyses confirm that the isolated subspace captures high-level semantics.

cs.CL↗

arXiv2025

WebGraphEval: Multi-Turn Trajectory Evaluation for Web Agents using Graph Representation

Current evaluation of web agents largely reduces to binary success metrics or conformity to a single reference trajectory, ignoring the structural diversity present in benchmark datasets. We present WebGraphEval, a framework that abstracts trajectories from multiple agents into a unified, weighted action graph. This representation is directly compatible with benchmarks such as WebArena, leveraging leaderboard runs and newly collected trajectories without modifying environments. The framework canonically encodes actions, merges recurring behaviors, and applies structural analyses including reward propagation and success-weighted edge statistics. Evaluations across thousands of trajectories from six web agents show that the graph abstraction captures cross-model regularities, highlights redundancy and inefficiency, and identifies critical decision points overlooked by outcome-based metrics. By framing web interaction as graph-structured data, WebGraphEval establishes a general methodology for multi-path, cross-agent, and efficiency-aware evaluation of web agents.

cs.AI↗

arXiv2025

HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy

The increasing reliance on natural language generation (NLG) models, particularly large language models, has raised concerns about the reliability and accuracy of their outputs. A key challenge is hallucination, where models produce plausible but incorrect information. As a result, hallucination detection has become a critical task. In this work, we introduce a comprehensive hallucination taxonomy with 11 categories across various NLG tasks and propose the HAllucination Detection (HAD) models https://github.com/pku0xff/HAD, which integrate hallucination detection, span-level identification, and correction into a single inference process. Trained on an elaborate synthetic dataset of about 90K samples, our HAD models are versatile and can be applied to various NLG tasks. We also carefully annotate a test set for hallucination detection, called HADTest, which contains 2,248 samples. Evaluations on in-domain and out-of-domain test sets show that our HAD models generally outperform the existing baselines, achieving state-of-the-art results on HaluEval, FactCHD, and FaithBench, confirming their robustness and versatility.

cs.CL↗

arXiv2025

SoK: Potentials and Challenges of Large Language Models for Reverse Engineering

Reverse Engineering (RE) is central to software security, enabling tasks such as vulnerability discovery and malware analysis, but it remains labor-intensive and requires substantial expertise. Earlier advances in deep learning start to automate parts of RE, particularly for malware detection and vulnerability classification. More recently, a rapidly growing body of work has applied Large Language Models (LLMs) to similar purposes. Their role compared to prior machine learning remains unclear, since some efforts simply adapt existing pipelines with minimal change while others seek to exploit broader reasoning and generative abilities. These differences, combined with varied problem definitions, methods, and evaluation practices, limit comparability, reproducibility, and cumulative progress. This paper systematizes the field by reviewing 44 research papers, including peer-reviewed publications and preprints, and 18 additional open-source projects that apply LLMs in RE. We propose a taxonomy that organizes existing work by objective, target, method, evaluation strategy, and data scale. Our analysis identifies strengths and limitations, highlights reproducibility and evaluation gaps, and examines emerging risks. We conclude with open challenges and future research directions that aim to guide more coherent and security-relevant applications of LLMs in RE.

cs.CR↗

arXiv2025

Ransomware Negotiation: Dynamics and Privacy-Preserving Mechanism Design

Ransomware attacks have become a pervasive and costly form of cybercrime, causing tens of millions of dollars in losses as organizations increasingly pay ransoms to mitigate operational disruptions and financial risks. While prior research has largely focused on proactive defenses, the post-infection negotiation dynamics between attackers and victims remains underexplored. This paper presents a formal analysis of attacker-victim interactions in modern ransomware incidents using a finite-horizon alternating-offers bargaining game model. Our analysis demonstrates how bargaining alters the optimal strategies of both parties. In practice, incomplete information-attackers lacking knowledge of victims' data valuations and victims lacking knowledge of attackers' reservation ransoms-can prolong negotiations and increase victims' business interruption costs. To address this, we design a Bayesian incentive-compatible mechanism that facilitates rapid agreement on a fair ransom without requiring either party to disclose private valuations. We further implement this mechanism using secure two-party computation based on garbled circuits, thereby eliminating the need for trusted intermediaries and preserving the privacy of both parties throughout the negotiation. To the best of our knowledge, this is the first automated, privacy-preserving negotiation mechanism grounded in a formal analysis of ransomware negotiation dynamics.

cs.GT↗

arXiv2025

Exploring Causal Effect of Social Bias on Faithfulness Hallucinations in Large Language Models

Large language models (LLMs) have achieved remarkable success in various tasks, yet they remain vulnerable to faithfulness hallucinations, where the output does not align with the input. In this study, we investigate whether social bias contributes to these hallucinations, a causal relationship that has not been explored. A key challenge is controlling confounders within the context, which complicates the isolation of causality between bias states and hallucinations. To address this, we utilize the Structural Causal Model (SCM) to establish and validate the causality and design bias interventions to control confounders. In addition, we develop the Bias Intervention Dataset (BID), which includes various social biases, enabling precise measurement of causal effects. Experiments on mainstream LLMs reveal that biases are significant causes of faithfulness hallucinations, and the effect of each bias state differs in direction. We further analyze the scope of these causal effects across various models, specifically focusing on unfairness hallucinations, which are primarily targeted by social bias, revealing the subtle yet significant causal effect of bias on hallucination generation.

cs.CL↗

arXiv2025

Part I: Tricks or Traps? A Deep Dive into RL for LLM Reasoning

Reinforcement learning for LLM reasoning has rapidly emerged as a prominent research area, marked by a significant surge in related studies on both algorithmic innovations and practical applications. Despite this progress, several critical challenges remain, including the absence of standardized guidelines for employing RL techniques and a fragmented understanding of their underlying mechanisms. Additionally, inconsistent experimental settings, variations in training data, and differences in model initialization have led to conflicting conclusions, obscuring the key characteristics of these techniques and creating confusion among practitioners when selecting appropriate techniques. This paper systematically reviews widely adopted RL techniques through rigorous reproductions and isolated evaluations within a unified open-source framework. We analyze the internal mechanisms, applicable scenarios, and core principles of each technique through fine-grained experiments, including datasets of varying difficulty, model sizes, and architectures. Based on these insights, we present clear guidelines for selecting RL techniques tailored to specific setups, and provide a reliable roadmap for practitioners navigating the RL for the LLM domain. Finally, we reveal that a minimalist combination of two techniques can unlock the learning capability of critic-free policies using vanilla PPO loss. The results demonstrate that our simple combination consistently improves performance, surpassing strategies like GRPO and DAPO.

cs.LG↗