arXiv subjects

Xinyu Xing

Publications and source records attributed to Xinyu Xing.

At least 19 recordsLinked to original sources

arXiv2026

Wrinkling of Randomly Heterogeneous Film-Substrate Systems

Wrinkling instabilities in stiff films on compliant substrates are strongly affected by spatial fluctuations in film stiffness. We develop a homogenized instability theory for one-dimensional film--substrate systems with random bending stiffness. The heterogeneous stability equation is reformulated as a Lippmann--Schwinger equation for the curvature field, and a strong-contrast expansion is derived using a local--nonlocal kernel decomposition and a cavity-field formulation. Truncation at third order yields an effective polarizability and a Dyson-type dispersion relation for predicting the critical load and wavenumber. The stiffness is modeled as an exponentially mapped Gaussian random field, allowing the required two- and three-point connected statistics to be obtained analytically. The theory is validated against generalized eigenvalue calculations and Fourier spectral simulations. Increasing stiffness contrast lowers the critical load and shifts the instability toward higher wavenumbers, producing shorter wrinkles. At weak contrast, the threshold follows the universal scaling $N_c^{(0)}-N_c\sim\varepsilon^2$, whereas at moderate and strong contrast the third-order approximation is more accurate than the second-order theory. Wavelength selection is controlled by the ratio of the dominant material wavelength $\lambda^\ast$ to the harmonic-mean reference wavelength $\lambda_H$. For $\lambda^\ast/\lambda_H<1$, the harmonic-mean model accurately predicts the wrinkle wavelength. The framework provides a mechanics-based tool for reliability assessment and design of statistically heterogeneous film--substrate systems.

cond-mat.mtrl-sci↗

arXiv2026

Patch Validation in Automated Vulnerability Repair

Automated Vulnerability Repair (AVR) systems, especially those leveraging large language models (LLMs), have demonstrated promising results in patching vulnerabilities -- that is, if we trust their patch validation methodology. Ground-truth patches from human developers often come with new tests that not only ensure mitigation of the vulnerability but also encode extra semantics such as root cause location, optimal fix strategy, or subtle coding styles or conventions. And yet, none of the recent AVR systems verify that the auto-generated patches additionally pass these new tests (termed as $\text{PoC}^+$ tests). This is a subtle yet critical omission. To fill this gap, we constructed a benchmark, $\textrm{PVBench}$, with 209 cases spanning 20 projects. Each case includes basic tests (functional tests before the patch and the PoC exploit) as well as the associated $\text{PoC}^+$ tests. Evaluated on three state-of-the-art AVR systems, we find that over 40\% of patches validated as correct by basic tests fail under $\text{PoC}^+$ testing, revealing substantial overestimation on patch success rates. Analyzing these patches that are falsely labeled as correct, we suggest that AVR tools should improve in three critical areas: root cause analysis, adherence to program specifications, and capturing developer intention.

cs.SE↗

arXiv2025

GPO: Learning from Critical Steps to Improve LLM Reasoning

Large language models (LLMs) are increasingly used in various domains, showing impressive potential on different tasks. Recently, reasoning LLMs have been proposed to improve the \textit{reasoning} or \textit{thinking} capabilities of LLMs to solve complex problems. Despite the promising results of reasoning LLMs, enhancing the multi-step reasoning capabilities of LLMs still remains a significant challenge. While existing optimization methods have advanced the LLM reasoning capabilities, they often treat reasoning trajectories as a whole, without considering the underlying critical steps within the trajectory. In this paper, we introduce \textbf{G}uided \textbf{P}ivotal \textbf{O}ptimization (GPO), a novel fine-tuning strategy that dives into the reasoning process to enable more effective improvements. GPO first identifies the `critical step' within a reasoning trajectory - a point that the model must carefully proceed to succeed at the problem. We locate the critical step by estimating the advantage function. GPO then resets the policy to the critical step, samples the new rollout and prioritizes the learning process on those rollouts. This focus allows the model to learn more effectively from pivotal moments within the reasoning process to improve the reasoning performance. We demonstrate that GPO is a general strategy that can be integrated with various optimization methods to improve reasoning performance. Besides theoretical analysis, our experiments across challenging reasoning benchmarks show that GPO can consistently and significantly enhance the performance of existing optimization methods, showcasing its effectiveness and generalizability in improving LLM reasoning by concentrating on pivotal moments within the generation process.

cs.AI↗

arXiv2025

Building Coding Agents via Entropy-Enhanced Multi-Turn Preference Optimization

Software engineering presents complex, multi-step challenges for Large Language Models (LLMs), requiring reasoning over large codebases and coordinated tool use. The difficulty of these tasks is exemplified by benchmarks like SWE-bench, where current LLMs still struggle to resolve real-world issues. A promising approach to enhance performance is test-time scaling (TTS), but its gains are heavily dependent on the diversity of model outputs. While standard alignment methods such as Direct Preference Optimization (DPO) and Kahneman-Tversky Optimization (KTO) are effective at aligning model outputs with human preferences, this process can come at the cost of reduced diversity, limiting the effectiveness of TTS. Additionally, existing preference optimization algorithms are typically designed for single-turn tasks and do not fully address the complexities of multi-turn reasoning and tool integration required for interactive coding agents. To bridge this gap, we introduce EntroPO, an entropy-enhanced framework that adapts existing preference optimization algorithms to the multi-turn, tool-assisted setting. EntroPO augments the preference objective to explicitly preserve policy entropy and generalizes learning to optimize over multi-turn interactions rather than single-turn responses. We validate EntroPO by fine-tuning a diverse suite of models from different families and sizes (up to 106B parameters).To maximize performance gains from TTS, we further propose a hybrid best-trajectory selection scheme combining a learned verifier model with model free approaches. On the SWEBENCH leaderboard, our approach establishes new state-of-the-art results among open-weight models. A 30B parameter model trained with EntroPO ranks 1st on SWEBENCH-LITE and 4th on SWEBENCH-VERIFIED on the open-weight leaderboard, surpassed only by models with over 10x more parameters(e.g., >$350B).

cs.AI↗

arXiv2025

BandFuzz: An ML-powered Collaborative Fuzzing Framework

Collaborative fuzzing combines multiple individual fuzzers and dynamically chooses appropriate combinations for different programs. Unlike individual fuzzers that rely on specific assumptions, collaborative fuzzing relaxes assumptions on target programs, providing robust performance across various programs. However, existing collaborative fuzzing frameworks face challenges including additional computational resource requirements and inefficient resource allocation among fuzzers. To tackle these challenges, we present BANDFUZZ, an ML-powered collaborative fuzzing framework that outperforms individual fuzzers without requiring additional computational resources. The key contribution of BANDFUZZ lies in its novel resource allocation algorithm driven by our proposed multi-armed bandits model. Different from greedy methods in existing frameworks, BANDFUZZ models the long-term impact of individual fuzzers, enabling discovery of globally optimal collaborative strategies. We propose a novel fuzzer evaluation method that assesses not only code coverage but also the fuzzer's capability of solving difficult branches. Finally, we integrate a real-time seed synchronization mechanism and implementation-wise optimizations to improve fuzzing efficiency and stability. Through extensive experiments on Fuzzbench and Fuzzer Test Suite, we show that BANDFUZZ outperforms state-of-the-art collaborative fuzzing framework autofz and widely used individual fuzzers. We verify BANDFUZZ's key designs through comprehensive ablation study. Notably, we demonstrate BANDFUZZ's effectiveness in real-world bug detection by analyzing results of a worldwide fuzzing competition, where BANDFUZZ won first place.

cs.CR↗

arXiv2025

UC-MOA: Utility-Conditioned Multi-Objective Alignment for Distributional Pareto-Optimality

Reinforcement Learning from Human Feedback (RLHF) has become a cornerstone for aligning large language models (LLMs) with human values. However, existing approaches struggle to capture the multi-dimensional, distributional nuances of human preferences. Methods such as RiC that directly inject raw reward values into prompts face significant numerical sensitivity issues--for instance, LLMs may fail to distinguish between 9.11 and 9.8--while alternatives like MORLHF, Rewarded Soups, and MODPO incur high computational costs by training multiple models. In this work, we introduce Utility-Conditioned Multi-Objective Alignment (UC-MOA), a novel framework that overcomes these limitations. Our approach leverages a diverse set of strictly increasing, non-linear utility functions to transform user-specified preferences into symbolic tokens, which are then used to condition a single LLM. This design not only mitigates numerical reasoning challenges but also substantially reduces training overhead, yielding models that achieve superior Pareto fronts and robust alignment across complex reward dimensions.

cs.CL↗

arXiv2025

A Survey on Explainable Deep Reinforcement Learning

Deep Reinforcement Learning (DRL) has achieved remarkable success in sequential decision-making tasks across diverse domains, yet its reliance on black-box neural architectures hinders interpretability, trust, and deployment in high-stakes applications. Explainable Deep Reinforcement Learning (XRL) addresses these challenges by enhancing transparency through feature-level, state-level, dataset-level, and model-level explanation techniques. This survey provides a comprehensive review of XRL methods, evaluates their qualitative and quantitative assessment frameworks, and explores their role in policy refinement, adversarial robustness, and security. Additionally, we examine the integration of reinforcement learning with Large Language Models (LLMs), particularly through Reinforcement Learning from Human Feedback (RLHF), which optimizes AI alignment with human preferences. We conclude by highlighting open research challenges and future directions to advance the development of interpretable, reliable, and accountable DRL systems.

cs.LG↗

arXiv2024

Harnessing Large Language Models for Seed Generation in Greybox Fuzzing

Greybox fuzzing has emerged as a preferred technique for discovering software bugs, striking a balance between efficiency and depth of exploration. While research has focused on improving fuzzing techniques, the importance of high-quality initial seeds remains critical yet often overlooked. Existing methods for seed generation are limited, especially for programs with non-standard or custom input formats. Large Language Models (LLMs) has revolutionized numerous domains, showcasing unprecedented capabilities in understanding and generating complex patterns across various fields of knowledge. This paper introduces SeedMind, a novel system that leverages LLMs to boost greybox fuzzing through intelligent seed generation. Unlike previous approaches, SeedMind employs LLMs to create test case generators rather than directly producing test cases. Our approach implements an iterative, feedback-driven process that guides the LLM to progressively refine test case generation, aiming for increased code coverage depth and breadth. In developing SeedMind, we addressed key challenges including input format limitations, context window constraints, and ensuring consistent, progress-aware behavior. Intensive evaluations with real-world applications show that SeedMind effectively harnesses LLMs to generate high-quality test cases and facilitate fuzzing in bug finding, presenting utility comparable to human-created seeds and significantly outperforming the existing LLM-based solutions.

cs.CR↗

arXiv2024

Soft-Label Integration for Robust Toxicity Classification

Toxicity classification in textual content remains a significant problem. Data with labels from a single annotator fall short of capturing the diversity of human perspectives. Therefore, there is a growing need to incorporate crowdsourced annotations for training an effective toxicity classifier. Additionally, the standard approach to training a classifier using empirical risk minimization (ERM) may fail to address the potential shifts between the training set and testing set due to exploiting spurious correlations. This work introduces a novel bi-level optimization framework that integrates crowdsourced annotations with the soft-labeling technique and optimizes the soft-label weights by Group Distributionally Robust Optimization (GroupDRO) to enhance the robustness against out-of-distribution (OOD) risk. We theoretically prove the convergence of our bi-level optimization algorithm. Experimental results demonstrate that our approach outperforms existing baseline methods in terms of both average and worst-group accuracy, confirming its effectiveness in leveraging crowdsourced annotations to achieve more effective and robust toxicity classification.

cs.AI↗

arXiv2024

BlockScan: Detecting Anomalies in Blockchain Transactions

We propose BlockScan, a customized Transformer for anomaly detection in blockchain transactions. Unlike existing methods that rely on rule-based systems or directly apply off-the-shelf large language models (LLMs), BlockScan introduces a series of customized designs to effectively model the unique data structure of blockchain transactions. First, a blockchain transaction is multi-modal, containing blockchain-specific tokens, texts, and numbers. We design a novel modularized tokenizer to handle these multi-modal inputs, balancing the information across different modalities. Second, we design a customized masked language modeling mechanism for pretraining the Transformer architecture, incorporating RoPE embedding and FlashAttention for handling longer sequences. Finally, we design a novel anomaly detection method based on the model outputs. We further provide theoretical analysis for the detection method of our system. Extensive evaluations on Ethereum and Solana transactions demonstrate BlockScan's exceptional capability in anomaly detection while maintaining a low false positive rate. Remarkably, BlockScan is the only method that successfully detects anomalous transactions on Solana with high accuracy, whereas all other approaches achieved very low or zero detection recall scores. This work sets a new benchmark for applying Transformer-based approaches in blockchain data analysis.

cs.CR↗

arXiv2024

RICE: Breaking Through the Training Bottlenecks of Reinforcement Learning with Explanation

Deep reinforcement learning (DRL) is playing an increasingly important role in real-world applications. However, obtaining an optimally performing DRL agent for complex tasks, especially with sparse rewards, remains a significant challenge. The training of a DRL agent can be often trapped in a bottleneck without further progress. In this paper, we propose RICE, an innovative refining scheme for reinforcement learning that incorporates explanation methods to break through the training bottlenecks. The high-level idea of RICE is to construct a new initial state distribution that combines both the default initial states and critical states identified through explanation methods, thereby encouraging the agent to explore from the mixed initial states. Through careful design, we can theoretically guarantee that our refining scheme has a tighter sub-optimality bound. We evaluate RICE in various popular RL environments and real-world applications. The results demonstrate that RICE significantly outperforms existing refining schemes in enhancing agent performance.

cs.LG↗

arXiv2024

ShadowBound: Efficient Heap Memory Protection Through Advanced Metadata Management and Customized Compiler Optimization

In software development, the prevalence of unsafe languages such as C and C++ introduces potential vulnerabilities, especially within the heap, a pivotal component for dynamic memory allocation. Despite its significance, heap management complexities have made heap corruption pervasive, posing severe threats to system security. While prior solutions aiming for temporal and spatial memory safety exhibit overheads deemed impractical, we present ShadowBound, a unique heap memory protection design. At its core, ShadowBound is an efficient out-of-bounds defense that can work with various use-after-free defenses (e.g. MarkUs, FFMalloc, PUMM) without compatibility constraints. We harness a shadow memory-based metadata management mechanism to store heap chunk boundaries and apply customized compiler optimizations tailored for boundary checking. We implemented ShadowBound atop the LLVM framework and integrated three state-of-the-art use-after-free defenses. Our evaluations show that ShadowBound provides robust heap protection with minimal time and memory overhead, suggesting its effectiveness and efficiency in safeguarding real-world programs against prevalent heap vulnerabilities.

cs.CR↗

arXiv2024

CAMP: Compiler and Allocator-based Heap Memory Protection

The heap is a critical and widely used component of many applications. Due to its dynamic nature, combined with the complexity of heap management algorithms, it is also a frequent target for security exploits. To enhance the heap's security, various heap protection techniques have been introduced, but they either introduce significant runtime overhead or have limited protection. We present CAMP, a new sanitizer for detecting and capturing heap memory corruption. CAMP leverages a compiler and a customized memory allocator. The compiler adds boundary-checking and escape-tracking instructions to the target program, while the memory allocator tracks memory ranges, coordinates with the instrumentation, and neutralizes dangling pointers. With the novel error detection scheme, CAMP enables various compiler optimization strategies and thus eliminates redundant and unnecessary check instrumentation. This design minimizes runtime overhead without sacrificing security guarantees. Our evaluation and comparison of CAMP with existing tools, using both real-world applications and SPEC CPU benchmarks, show that it provides even better heap corruption detection capability with lower runtime overhead.

cs.CR↗

arXiv2024

Decoupled Alignment for Robust Plug-and-Play Adaptation

We introduce a training-free safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning or reinforcement learning from human feedback. Our main idea is to provide a robust plug-and-play approach to prevent shadow alignment when models are adapted to downstream tasks. Specifically, we leverage knowledge distillation to extract alignment signals from well-aligned LLMs and inject them into shadow-aligned models via model fusion, enabling plug-and-play alignment correction. In our methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense success rate by approximately 14.42%, reaching as high as 51.39% across 17 influenced LLMs, without compromising performance. Our code is available at https://github.com/NWULIST/DAPA.

cs.CL↗

arXiv2024

Take a Step Further: Understanding Page Spray in Linux Kernel Exploitation

Recently, a novel method known as Page Spray emerges, focusing on page-level exploitation for kernel vulnerabilities. Despite the advantages it offers in terms of exploitability, stability, and compatibility, comprehensive research on Page Spray remains scarce. Questions regarding its root causes, exploitation model, comparative benefits over other exploitation techniques, and possible mitigation strategies have largely remained unanswered. In this paper, we conduct a systematic investigation into Page Spray, providing an in-depth understanding of this exploitation technique. We introduce a comprehensive exploit model termed the \sys model, elucidating its fundamental principles. Additionally, we conduct a thorough analysis of the root causes underlying Page Spray occurrences within the Linux Kernel. We design an analyzer based on the Page Spray analysis model to identify Page Spray callsites. Subsequently, we evaluate the stability, exploitability, and compatibility of Page Spray through meticulously designed experiments. Finally, we propose mitigation principles for addressing Page Spray and introduce our own lightweight mitigation approach. This research aims to assist security researchers and developers in gaining insights into Page Spray, ultimately enhancing our collective understanding of this emerging exploitation technique and making improvements to the community.

cs.CR↗

arXiv2024

Mind the Inconspicuous: Revealing the Hidden Weakness in Aligned LLMs' Refusal Boundaries

Recent advances in Large Language Models (LLMs) have led to impressive alignment where models learn to distinguish harmful from harmless queries through supervised finetuning (SFT) and reinforcement learning from human feedback (RLHF). In this paper, we reveal a subtle yet impactful weakness in these aligned models. We find that simply appending multiple end of sequence (eos) tokens can cause a phenomenon we call context segmentation, which effectively shifts both harmful and benign inputs closer to the refusal boundary in the hidden space. Building on this observation, we propose a straightforward method to BOOST jailbreak attacks by appending eos tokens. Our systematic evaluation shows that this strategy significantly increases the attack success rate across 8 representative jailbreak techniques and 16 open-source LLMs, ranging from 2B to 72B parameters. Moreover, we develop a novel probing mechanism for commercial APIs and discover that major providers such as OpenAI, Anthropic, and Qwen do not filter eos tokens, making them similarly vulnerable. These findings highlight a hidden yet critical blind spot in existing alignment and content filtering approaches. We call for heightened attention to eos tokens' unintended influence on model behaviors, particularly in production systems. Our work not only calls for an input-filtering based defense, but also points to new defenses that make refusal boundaries more robust and generalizable, as well as fundamental alignment techniques that can defend against context segmentation attacks.

cs.AI↗

arXiv2024

Assessing Prompt Injection Risks in 200+ Custom GPTs

In the rapidly evolving landscape of artificial intelligence, ChatGPT has been widely used in various applications. The new feature - customization of ChatGPT models by users to cater to specific needs has opened new frontiers in AI utility. However, this study reveals a significant security vulnerability inherent in these user-customized GPTs: prompt injection attacks. Through comprehensive testing of over 200 user-designed GPT models via adversarial prompts, we demonstrate that these systems are susceptible to prompt injections. Through prompt injection, an adversary can not only extract the customized system prompts but also access the uploaded files. This paper provides a first-hand analysis of the prompt injection, alongside the evaluation of the possible mitigation of such attacks. Our findings underscore the urgent need for robust security frameworks in the design and deployment of customizable GPT models. The intent of this paper is to raise awareness and prompt action in the AI community, ensuring that the benefits of GPT customization do not come at the cost of compromised security and privacy.

cs.CR↗

arXiv2023

GPTFUZZER: Red Teaming Large Language Models with Auto-Generated Jailbreak Prompts

Large language models (LLMs) have recently experienced tremendous popularity and are widely used from casual conversations to AI-driven programming. However, despite their considerable success, LLMs are not entirely reliable and can give detailed guidance on how to conduct harmful or illegal activities. While safety measures can reduce the risk of such outputs, adversarial jailbreak attacks can still exploit LLMs to produce harmful content. These jailbreak templates are typically manually crafted, making large-scale testing challenging. In this paper, we introduce GPTFuzz, a novel black-box jailbreak fuzzing framework inspired by the AFL fuzzing framework. Instead of manual engineering, GPTFuzz automates the generation of jailbreak templates for red-teaming LLMs. At its core, GPTFuzz starts with human-written templates as initial seeds, then mutates them to produce new templates. We detail three key components of GPTFuzz: a seed selection strategy for balancing efficiency and variability, mutate operators for creating semantically equivalent or similar sentences, and a judgment model to assess the success of a jailbreak attack. We evaluate GPTFuzz against various commercial and open-source LLMs, including ChatGPT, LLaMa-2, and Vicuna, under diverse attack scenarios. Our results indicate that GPTFuzz consistently produces jailbreak templates with a high success rate, surpassing human-crafted templates. Remarkably, GPTFuzz achieves over 90% attack success rates against ChatGPT and Llama-2 models, even with suboptimal initial seed templates. We anticipate that GPTFuzz will be instrumental for researchers and practitioners in examining LLM robustness and will encourage further exploration into enhancing LLM safety.

cs.AI↗