arXiv subjects

Xin Yin

Publications and source records attributed to Xin Yin.

At least 37 records · Page 2Linked to original sources

arXiv2025

Input Reduction Enhanced LLM-based Program Repair

Large Language Models (LLMs) have shown great potential in Automated Program Repair (APR). Test inputs, being crucial for reasoning the root cause of failures, are always included in the prompt for LLM-based APR. Unfortunately, LLMs struggle to retain key information in long prompts. When the test inputs are extensive in the prompt, this may trigger the "lost-in-the-middle" issue, compromising repair performance. ReduceFix prompts an LLM to generate a reducer that minimizes failure-inducing test inputs without human effort, and then feeds the reduced failure-inducing inputs to guide patch generation. For targeted evaluation, we constructed LFTBench, the first long-input APR benchmark with 200 real bugs from 20 programming tasks, each paired with a failure-inducing input whose median size is 1 MB. On this benchmark, ReduceFix shrinks inputs by 89.1% on average and improves overall pass@10 by up to 53.8% relative to a prompt that includes the original test, and by 17.6% compared with omitting the test entirely. Adding the same reduction step to ChatRepair and CREF increases their fix rate by 21.3% and 2.6%, respectively, without other changes. Our gains hold against a ddmin-only reducing template baseline and transfer to repository-level OSS-Fuzz cases. Ablation studies further highlight the impact of input length and compressed failure information on repair success. These results underscore that automatically reducing failing inputs is a practical and powerful complement to LLM-based APR, significantly improving its scalability and effectiveness.

cs.SE↗

arXiv2025

Learning to Align Human Code Preferences

Large Language Models (LLMs) have demonstrated remarkable potential in automating software development tasks. While recent advances leverage Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) to align models with human preferences, the optimal training strategy remains unclear across diverse code preference scenarios. This paper systematically investigates the roles of SFT and DPO in aligning LLMs with different code preferences. Through both theoretical analysis and empirical observation, we hypothesize that SFT excels in scenarios with objectively verifiable optimal solutions, while applying SFT followed by DPO (S&D) enables models to explore superior solutions in scenarios without objectively verifiable optimal solutions. Based on the analysis and experimental evidence, we propose Adaptive Preference Optimization (APO), a dynamic integration approach that adaptively amplifies preferred responses, suppresses dispreferred ones, and encourages exploration of potentially superior solutions during training. Extensive experiments across six representative code preference tasks validate our theoretical hypotheses and demonstrate that APO consistently matches or surpasses the performance of existing SFT and S&D strategies. Our work provides both theoretical foundations and practical guidance for selecting appropriate training strategies in different code preference alignment scenarios.

cs.SE↗

arXiv2025

Dynamic Log-Gaussian Process Control Barrier Function for Safe Robotic Navigation in Dynamic Environments

Control Barrier Functions (CBFs) have emerged as efficient tools to address the safe navigation problem for robot applications. However, synthesizing informative and obstacle motion-aware CBFs online using real-time sensor data remains challenging, particularly in unknown and dynamic scenarios. Motived by this challenge, this paper aims to propose a novel Gaussian Process-based formulation of CBF, termed the Dynamic Log Gaussian Process Control Barrier Function (DLGP-CBF), to enable real-time construction of CBF which are both spatially informative and responsive to obstacle motion. Firstly, the DLGP-CBF leverages a logarithmic transformation of GP regression to generate smooth and informative barrier values and gradients, even in sparse-data regions. Secondly, by explicitly modeling the DLGP-CBF as a function of obstacle positions, the derived safety constraint integrates predicted obstacle velocities, allowing the controller to proactively respond to dynamic obstacles' motion. Simulation results demonstrate significant improvements in obstacle avoidance performance, including increased safety margins, smoother trajectories, and enhanced responsiveness compared to baseline methods.

cs.RO↗

arXiv2025

A 3D-integrated BiCMOS-silicon photonics high-speed receiver realized using micro-transfer printing

Meeting the escalating demands of data transmission and computing, driven by artificial intelligence (AI), requires not only faster optical transceivers but also advanced integration technologies that can seamlessly combine photonic and electronic components. Traditional approaches struggle to overcome the parasitic limitations arising from fabricating those components using different processes. Here, we report a novel 3D heterogeneously integrated optical receiver based on micro-transfer printing (μTP), enabling the co-integration of a compact bipolar CMOS (BiCMOS) electronic chiplet (0.06 mm2) directly onto a silicon photonic integrated circuit (SiPIC). While previous μTP demonstrations have focused primarily on photonic integration, our work pioneers the direct integration of electronics and photonics, significantly enhancing performance and scalability. The resulting optical receiver achieves 224 Gb/s four-level pulse amplitude modulation (PAM-4) operation, delivering -5.2 dBm optical modulation amplitude(OMA) sensitivity at a bit-error rate (BER) of 2.4 x 10-4, a record-small footprint, and an excellent power efficiency of 0.51 pJ/b. This demonstration not only showcases the potential of μTP for high-density, cost-efficient integration but also represents a critical step toward next-generation optical interconnects in the AI era.

physics.optics↗

arXiv2025

SGLP: A Similarity Guided Fast Layer Partition Pruning for Compressing Large Deep Models

Layer pruning has emerged as a potent approach to remove redundant layers in the pre-trained network on the purpose of reducing network size and improve computational efficiency. However, existing layer pruning methods mostly overlook the intrinsic connections and inter-dependencies between different layers within complicated deep neural networks. This oversight can result in pruned models that do not preserve the essential characteristics of the pre-trained network as effectively as desired. To address these limitations, we propose a Similarity-Guided Layer Partition (SGLP) Pruning, a novel pruning framework that exploits representation similarity to guide efficient and informed layer removal for compressing large deep models. Our method begins by employing Centered Kernel Alignment (CKA) to quantify representational similarity between layers, uncovering structural patterns within the network. We then apply Fisher Optimal Segmentation on the similarity matrix to partition the network into semantically coherent layer segments. This segmentation allows pruning decisions to respect layer interdependencies and preserve essential knowledge. Within each segment, we introduce a fine-tuning-free importance evaluation using GradNorm, identifying and removing redundant layers in a targeted, segment-wise manner. Experimental results on both image classification tasks and large language models (LLMs) demonstrate that our proposed SGLP outperforms the state-of-the-art methods in accuracy and efficiency. Our approach achieves significant model compression with minimal performance degradation, making it well-suited for deployment in resource-limited environments.

cs.LG↗

arXiv2025

Diffusion Policies with Value-Conditional Optimization for Offline Reinforcement Learning

In offline reinforcement learning, value overestimation caused by out-of-distribution (OOD) actions significantly limits policy performance. Recently, diffusion models have been leveraged for their strong distribution-matching capabilities, enforcing conservatism through behavior policy constraints. However, existing methods often apply indiscriminate regularization to redundant actions in low-quality datasets, resulting in excessive conservatism and an imbalance between the expressiveness and efficiency of diffusion modeling. To address these issues, we propose DIffusion policies with Value-conditional Optimization (DIVO), a novel approach that leverages diffusion models to generate high-quality, broadly covered in-distribution state-action samples while facilitating efficient policy improvement. Specifically, DIVO introduces a binary-weighted mechanism that utilizes the advantage values of actions in the offline dataset to guide diffusion model training. This enables a more precise alignment with the dataset's distribution while selectively expanding the boundaries of high-advantage actions. During policy improvement, DIVO dynamically filters high-return-potential actions from the diffusion model, effectively guiding the learned policy toward better performance. This approach achieves a critical balance between conservatism and explorability in offline RL. We evaluate DIVO on the D4RL benchmark and compare it against state-of-the-art baselines. Empirical results demonstrate that DIVO achieves superior performance, delivering significant improvements in average returns across locomotion tasks and outperforming existing methods in the challenging AntMaze domain, where sparse rewards pose a major difficulty.

cs.LG↗

arXiv2025

Navigating the Labyrinth: Path-Sensitive Unit Test Generation with Large Language Models

Unit testing is essential for software quality assurance, yet writing and maintaining tests remains time-consuming and error-prone. To address this challenge, researchers have proposed various techniques for automating unit test generation, including traditional heuristic-based methods and more recent approaches that leverage large language models (LLMs). However, these existing approaches are inherently path-insensitive because they rely on fixed heuristics or limited contextual information and fail to reason about deep control-flow structures. As a result, they often struggle to achieve adequate coverage, particularly for deep or complex execution paths. In this work, we present a path-sensitive framework, JUnitGenie, to fill this gap by combining code knowledge with the semantic capabilities of LLMs in guiding context-aware unit test generation. After extracting code knowledge from Java projects, JUnitGenie distills this knowledge into structured prompts to guide the generation of high-coverage unit tests. We evaluate JUnitGenie on 2,258 complex focal methods from ten real-world Java projects. The results show that JUnitGenie generates valid tests and improves branch and line coverage by 29.60% and 31.00% on average over both heuristic and LLM-based baselines. We further demonstrate that the generated test cases can uncover real-world bugs, which were later confirmed and fixed by developers.

cs.SE↗

arXiv2025

A Preference-Driven Methodology for High-Quality Solidity Code Generation

While Large Language Models (LLMs) have demonstrated remarkable progress in generating functionally correct Solidity code, they continue to face critical challenges in producing gas-efficient and secure code, which are critical requirements for real-world smart contract deployment. Although recent advances leverage Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) for code preference alignment, existing approaches treat functional correctness, gas optimization, and security as independent objectives, resulting in contracts that may achieve operational soundness but suffer from prohibitive execution costs or dangerous vulnerabilities. To address these limitations, we propose PrefGen, a novel framework that extends standard DPO beyond human preferences to incorporate quantifiable blockchain-specific metrics, enabling holistic multi-objective optimization specifically tailored for smart contract generation. Our framework introduces a comprehensive evaluation methodology with four complementary metrics: Pass@k (functional correctness), Compile@k (syntactic correctness), Gas@k (gas efficiency), and Secure@k (security assessment), providing rigorous multi-dimensional contract evaluation. Through extensive experimentation, we demonstrate that PrefGen significantly outperforms existing approaches across all critical dimensions, achieving 66.7% Pass@5, 58.9% Gas@5, and 62.5% Secure@5, while generating production-ready smart contracts that are functionally correct, cost-efficient, and secure.

cs.SE↗

arXiv2025

Pre-training CLIP against Data Poisoning with Optimal Transport-based Matching and Alignment

Recent studies have shown that Contrastive Language-Image Pre-training (CLIP) models are threatened by targeted data poisoning and backdoor attacks due to massive training image-caption pairs crawled from the Internet. Previous defense methods correct poisoned image-caption pairs by matching a new caption for each image. However, the matching process relies solely on the global representations of images and captions, overlooking fine-grained features of visual and textual features. It may introduce incorrect image-caption pairs and harm the CLIP pre-training. To address their limitations, we propose an Optimal Transport-based framework to reconstruct image-caption pairs, named OTCCLIP. We propose a new optimal transport-based distance measure between fine-grained visual and textual feature sets and re-assign new captions based on the proposed optimal transport distance. Additionally, to further reduce the negative impact of mismatched pairs, we encourage the inter- and intra-modality fine-grained alignment by employing optimal transport-based objective functions. Our experiments demonstrate that OTCCLIP can successfully decrease the attack success rates of poisoning attacks. Also, compared to previous methods, OTCCLIP significantly improves CLIP's zero-shot and linear probing performance trained on poisoned datasets.

cs.CV↗

arXiv2025

SolEval: Benchmarking Large Language Models for Repository-level Solidity Code Generation

Large language models (LLMs) have transformed code generation. However, most existing approaches focus on mainstream languages such as Python and Java, neglecting the Solidity language, the predominant programming language for Ethereum smart contracts. Due to the lack of adequate benchmarks for Solidity, LLMs' ability to generate secure, cost-effective smart contracts remains unexplored. To fill this gap, we construct SolEval, the first repository-level benchmark designed for Solidity smart contract generation, to evaluate the performance of LLMs on Solidity. SolEval consists of 1,507 samples from 28 different repositories, covering 6 popular domains, providing LLMs with a comprehensive evaluation benchmark. Unlike the existing Solidity benchmark, SolEval not only includes complex function calls but also reflects the real-world complexity of the Ethereum ecosystem by incorporating Gas@k and Vul@k. We evaluate 16 LLMs on SolEval, and our results show that the best-performing LLM achieves only 26.29% Pass@10, highlighting substantial room for improvement in Solidity code generation by LLMs. Additionally, we conduct supervised fine-tuning (SFT) on Qwen-7B using SolEval, resulting in a significant performance improvement, with Pass@5 increasing from 16.67% to 58.33%, demonstrating the effectiveness of fine-tuning LLMs on our benchmark. We release our data and code at https://github.com/pzy2000/SolEval.

cs.SE↗

arXiv2025

RepoTransAgent: Multi-Agent LLM Framework for Repository-Aware Code Translation

Repository-aware code translation is critical for modernizing legacy systems, enhancing maintainability, and enabling interoperability across diverse programming languages. While recent advances in large language models (LLMs) have improved code translation quality, existing approaches face significant challenges in practical scenarios: insufficient contextual understanding, inflexible prompt designs, and inadequate error correction mechanisms. These limitations severely hinder accurate and efficient translation of complex, real-world code repositories. To address these challenges, we propose RepoTransAgent, a novel multi-agent LLM framework for repository-aware code translation. RepoTransAgent systematically decomposes the translation process into specialized subtasks-context retrieval, dynamic prompt construction, and iterative code refinement-each handled by dedicated agents. Our approach leverages retrieval-augmented generation (RAG) for contextual information gathering, employs adaptive prompts tailored to varying repository scenarios, and introduces a reflection-based mechanism for systematic error correction. We evaluate RepoTransAgent on hundreds of Java-C# translation pairs from six popular open-source projects. Experimental results demonstrate that RepoTransAgent significantly outperforms state-of-the-art baselines in both compile and pass rates. Specifically, RepoTransAgent achieves up to 55.34% compile rate and 45.84% pass rate. Comprehensive analysis confirms the robustness and generalizability of RepoTransAgent across different LLMs, establishing its effectiveness for real-world repository-aware code translation.

cs.SE↗

arXiv2025

Detecting LLM-generated Code with Subtle Modification by Adversarial Training

With the rapid development of Large Language Models (LLMs), their powerful code-generation capabilities have been widely applied in tasks like code completion and automated development, demonstrating the value of improving coding efficiency. However, the extensive use of LLM-generated code also raises several new challenges. On the one hand, issues such as the regulation of code provenance, copyright disputes, and code quality have become increasingly concerning. How to effectively detect LLM-generated code and ensure its compliant and responsible use has become a critical and urgent issue. On the other hand, in practical applications, LLM-generated code is often subject to manual modifications, such as variable renaming or structural adjustments. Although some recent studies have proposed training-based and zero-shot methods for detecting LLM-generated code, these approaches show insufficient robustness when facing modified LLM-generated code, and there is a lack of an effective solution. To address the real-world scenario where LLM-generated code may undergo minor modifications, we propose CodeGPTSensor+, an enhanced version of CodeGPTSensor, which employs adversarial training to improve robustness against input perturbations. CodeGPTSensor+ integrates an adversarial sample generation module, Multi-objective Identifier and Structure Transformation (MIST), which systematically generates both high-quality and representative adversarial samples. This module effectively enhances the model's resistance against diverse adversarial attacks. Experimental results on the HMCorp dataset demonstrate that CodeGPTSensor+ significantly improves detection accuracy on the adversarial test set while maintaining high accuracy on the original test set, showcasing superior robustness compared to CodeGPTSensor.

cs.SE↗

arXiv2025

SepPrune: Structured Pruning for Efficient Deep Speech Separation

Although deep learning has substantially advanced speech separation in recent years, most existing studies continue to prioritize separation quality while overlooking computational efficiency, an essential factor for low-latency speech processing in real-time applications. In this paper, we propose SepPrune, the first structured pruning framework specifically designed to compress deep speech separation models and reduce their computational cost. SepPrune begins by analyzing the computational structure of a given model to identify layers with the highest computational burden. It then introduces a differentiable masking strategy to enable gradient-driven channel selection. Based on the learned masks, SepPrune prunes redundant channels and fine-tunes the remaining parameters to recover performance. Extensive experiments demonstrate that this learnable pruning paradigm yields substantial advantages for channel pruning in speech separation models, outperforming existing methods. Notably, a model pruned with SepPrune can recover 85% of the performance of a pre-trained model (trained over hundreds of epochs) with only one epoch of fine-tuning, and achieves convergence 36$\times$ faster than training from scratch. Code is available at https://github.com/itsnotacie/SepPrune.

cs.SD↗

arXiv2025

Chip-Based 16 GBaud Continuous-Variable Quantum Key Distribution

Quantum key distribution (QKD) stands as the most successful application of quantum information science, providing information-theoretic security for key exchange. While it has evolved from proof-of-concept experiments to commercial products, widespread adoption requires chip-based integration to reduce costs, enable mass production, facilitate miniaturization, and enhance system performance. Here, we demonstrate the first fully photonic-integrated continuous-variable QKD (CVQKD) system operating at a classical telecom symbol rate of 16 GBaud. Our system integrates a silicon photonic transmitter circuit (excluding the laser source) and a 20 GHz photonic-electronic receiver, which features a phase-diverse silicon photonic integrated circuit and custom-designed GaAs pHEMT transimpedance amplifiers. Advanced digital signal processing allows our system to achieve the highest reported secure key rate to date, reaching 0.289 Gb/s and 0.246 Gb/s over a 20 km fiber link in the asymptotic and finite-size regimes, respectively. These results establish a record key rate and represent a critical step toward scalable, cost-effective, and mass-deployable quantum-secure communication using photonic-integrated CVQKD systems.

quant-ph↗

arXiv2025

Real-time all-optical signal equalisation with silicon photonic recurrent neural networks

Communication through optical fibres experiences limitations due to chromatic dispersion and nonlinear Kerr effects that degrade the signal. Mitigating these impairments is typically done using complex digital signal processing algorithms. However, these equalisation methods require significant power consumption and introduce high latencies. Photonic reservoir computing (a subfield of neural networks) offers an alternative solution, processing signals in the analog optical Domain. In this work, we present to our knowledge the very first experimental demonstration of real-time online equalisation of fibre distortions using a silicon photonics chip that combines the recurrent reservoir and the programmable readout layer. We successfully equalize a 28 Gbps on-off keying signal across varying power levels and fibre lengths, even in the highly nonlinear regime. We obtain bit error rates orders of magnitude below previously reported optical equalisation methods, reaching as low as 4e-7 , far below the generic forward error correction limit of 5.8e-5 used in commercial Ethernet interfaces. Also, simulations show that simply by removing delay lines, the system becomes compatible with line rates of 896 Gpbs. Using wavelength multiplexing, this can result in a throughput in excess of 89.6 Tbps. Finally, incorporating non-volatile phase shifters, the power consumption can be less than 6 fJ/bit.

physics.optics↗

arXiv2025

A high-speed heterogeneous lithium tantalate silicon photonics platform

The rapid expansion of cloud computing and artificial intelligence has driven the demand for faster optical components in data centres to unprecedented levels. A key advancement in this field is the integration of multiple photonic components onto a single chip, enhancing the performance of optical transceivers. Here, silicon photonics, benefiting from mature fabrication processes, has gained prominence. The platform combines modulators, switches, photodetectors and low-loss waveguides on a single chip. However, emerging standards like 1600ZR+ potentially exceed the capabilities of silicon-based modulators. To address these limitations, thin-film lithium niobate has been proposed as an alternative to silicon photonics, offering a low voltage-length product and exceptional high-speed modulation properties. More recently, the first demonstrations of thin-film lithium tantalate circuits have emerged, addressing some of the disadvantages of lithium niobate enabling a reduced bias drift and enhanced resistance to optical damage. As such, making it a promising candidate for next-generation photonic platforms. However, a persistent drawback of such platforms is the lithium contamination, which complicates integration with CMOS fabrication processes. Here, we present for the first time the integration of lithium tantalate onto a silicon photonics chip. This integration is achieved without modifying the standard silicon photonics process design kit. Our device achieves low half-wave voltage (3.5 V), low insertion loss (2.9 dB) and high-speed operation (> 70 GHz), paving the way for next-gen applications. By minimising lithium tantalate material use, our approach reduces costs while leveraging existing silicon photonics technology advancements, in particular supporting ultra-fast monolithic germanium photodetectors and established process design kits.

physics.optics↗

arXiv2025

MulChain: Enabling Advanced Cross-Modal Queries in Hybrid-Storage Blockchains

With its decentralization and immutability, blockchain has emerged as a trusted foundation for data management and querying. Because blockchain storage space is limited, large multimodal data files, such as videos, are often stored offline, leaving only lightweight metadata on the chain. While this hybrid storage approach enhances storage efficiency, it introduces significant challenges for executing advanced queries on multimodal data. The metadata stored on-chain is often minimal and may not include all the attributes necessary for queries like time range or fuzzy queries. In addition, existing blockchains do not provide native support for multimodal data querying. Achieving this capability would necessitate extensive modifications to the underlying blockchain framework, even reconstructing its core architecture. Consequently, enabling blockchains with multimodal query capabilities remains a significant problem, which necessitates overcoming the following three key challenges: (1) Designing efficient indexing methods to adapt to varying workloads that involve frequent insertions and query operations; (2) Achieving seamless integration with existing blockchains without altering the underlying infrastructure; (3) Ensuring high query performance while minimizing gas consumption. To address these challenges, we propose MulChain, a novel middleware architecture to enable smooth integration with existing blockchains. At the core of MulChain is the BHashTree, a flexible data structure that dynamically switches between tree and hash nodes based on workload characteristics, ensuring efficient insertion and query operations. Furthermore, the middleware provides standardized interfaces for blockchain systems, unifying query methods across different platforms.

cs.DB↗

arXiv2025

Improving the Ability of Pre-trained Language Model by Imparting Large Language Model's Experience

Large Language Models (LLMs) and pre-trained Language Models (LMs) have achieved impressive success on many software engineering tasks (e.g., code completion and code generation). By leveraging huge existing code corpora (e.g., GitHub), these models can understand the patterns in source code and use these patterns to predict code properties. However, LLMs under few-shot learning perform poorly on non-generative tasks (e.g., fault localization and vulnerability localization), and fine-tuning LLMs is time-consuming and costly for end users and small organizations. Furthermore, the performance of fine-tuning LMs for non-generative tasks is impressive, yet it heavily depends on the amount and quality of data. As a result, the current lack of data and the high cost of collecting it in real-world scenarios further limit the applicability of LMs. In this paper, we leverage the powerful generation capabilities of LLMs to enhance pre-trained LMs. Specifically, we use LLMs to generate domain-specific data, thereby improving the performance of pre-trained LMs on the target tasks. We conduct experiments by combining different LLMs in our generation phase and introducing various LMs to learn from the LLM-generated data. Then, we compare the performance of these LMs before and after learning the data. We find that LLM-generated data significantly enhances the performance of LMs. The improvement can reach up to 58.36% for fault localization and up to 6.09% for clone detection.

cs.SE↗