arXiv subjects

Zhiyuan Zhong

Publications and source records attributed to Zhiyuan Zhong.

6 recordsLinked to original sources

arXiv2025

Multi-agent Assisted Automatic Test Generation for Java JSON Libraries

JSON is a widely used format for data exchange between applications. In the Java ecosystem, JSON libraries serve as fundamental toolkits for processing JSON data, powering real-world applications such as web services, Android apps, or data management systems. However, without effective quality assurance methods such as automatic test generation (ATG), developers risk introducing subtle data inconsistency bugs, compatibility issues, and even security vulnerabilities. These flaws can affect billions of end users and potentially cause severe financial losses. Recently, large language models (LLMs) have shown strong potential in enhancing ATG. However, existing LLM-based methods like TitanFuzz and YanHui lack specialization in the JSON domain. For Java JSON libraries (JJLs), effective test cases should capture the constraints between formatted data and application programs, leaving critical aspects of quality assurance unaddressed. To fill this gap, we propose JsonATG, a multi-agent ATG system that generates diverse bug-triggering tests for JJLs. With historical bug information as seeds, JsonATG introduces a code summarization agent and a test validation agent into the generation pipeline to produce new and valid test cases. It applies agent-generated program mutation rules tailored specifically for the structural and semantic characteristics of JJLs, such as data streaming operations, serialization formats, and data-binding patterns. The generated tests are further refined through post-processing to ensure syntactic and semantic correctness. Our experiments show that JsonATG achieves higher coverage than two state-of-the-art LLM-based test generation methods on the widely used JSON library. With a \$25 budget, we reported 59 bugs (including non-crashing functional bugs) in fastjson, of which 47 were confirmed and 28 have already been fixed.

cs.SE↗

arXiv2025

Can VLMs Detect and Localize Fine-Grained AI-Edited Images?

Fine-grained detection and localization of localized image edits is crucial for assessing content authenticity, especially as modern diffusion models and image editors can produce highly realistic manipulations. However, this problem faces three key challenges: (1) most AIGC detectors produce only a global real-or-fake label without indicating where edits occur; (2) traditional computer vision methods for edit localization typically rely on costly pixel-level annotations; and (3) there is no large-scale, modern benchmark specifically targeting edited-image detection. To address these gaps, we develop an automated data-generation pipeline and construct FragFake, a large-scale benchmark of AI-edited images spanning multiple source datasets, diverse editing models, and several common edit types. Building on FragFake, we are the first to systematically study vision language models (VLMs) for edited-image classification and edited-region localization. Our experiments show that pretrained VLMs, including GPT4o, perform poorly on this task, whereas fine-tuned models such as Qwen2.5-VL achieve high accuracy and substantially higher object precision across all settings. We further explore GRPO-based RLVR training, which yields modest metric gains while improving the interpretability of model outputs. Ablation and transfer analyses reveal how data balancing, training size, LoRA rank, and training domain affect performance, and highlight both the potential and the limitations of cross-editor and cross-dataset generalization. We anticipate that this work will establish a solid foundation to facilitate and inspire subsequent research endeavors in the domain of multimodal content authenticity.

cs.CV↗

arXiv2025

Backdoor Attack on Vision Language Models with Stealthy Semantic Manipulation

Vision Language Models (VLMs) have shown remarkable performance, but are also vulnerable to backdoor attacks whereby the adversary can manipulate the model's outputs through hidden triggers. Prior attacks primarily rely on single-modality triggers, leaving the crucial cross-modal fusion nature of VLMs largely unexplored. Unlike prior work, we identify a novel attack surface that leverages cross-modal semantic mismatches as implicit triggers. Based on this insight, we propose BadSem (Backdoor Attack with Semantic Manipulation), a data poisoning attack that injects stealthy backdoors by deliberately misaligning image-text pairs during training. To perform the attack, we construct SIMBad, a dataset tailored for semantic manipulation involving color and object attributes. Extensive experiments across four widely used VLMs show that BadSem achieves over 98% average ASR, generalizes well to out-of-distribution datasets, and can transfer across poisoning modalities. Our detailed analysis using attention visualization shows that backdoored models focus on semantically sensitive regions under mismatched conditions while maintaining normal behavior on clean inputs. To mitigate the attack, we try two defense strategies based on system prompt and supervised fine-tuning but find that both of them fail to mitigate the semantic backdoor. Our findings highlight the urgent need to address semantic vulnerabilities in VLMs for their safer deployment.

cs.CV↗

arXiv2025

On the Generalization and Adaptation Ability of Machine-Generated Text Detectors in Academic Writing

The rising popularity of large language models (LLMs) has raised concerns about machine-generated text (MGT), particularly in academic settings, where issues like plagiarism and misinformation are prevalent. As a result, developing a highly generalizable and adaptable MGT detection system has become an urgent priority. Given that LLMs are most commonly misused in academic writing, this work investigates the generalization and adaptation capabilities of MGT detectors in three key aspects specific to academic writing: First, we construct MGT-Acedemic, a large-scale dataset comprising over 336M tokens and 749K samples. MGT-Acedemic focuses on academic writing, featuring human-written texts (HWTs) and MGTs across STEM, Humanities, and Social Sciences, paired with an extensible code framework for efficient benchmarking. Second, we benchmark the performance of various detectors for binary classification and attribution tasks in both in-domain and cross-domain settings. This benchmark reveals the often-overlooked challenges of attribution tasks. Third, we introduce a novel attribution task where models have to adapt to new classes over time without (or with very limited) access to prior training data in both few-shot and many-shot scenarios. We implement eight different adapting techniques to improve the performance and highlight the inherent complexity of the task. Our findings provide insights into the generalization and adaptation ability of MGT detectors across diverse scenarios and lay the foundation for building robust, adaptive detection systems. The code framework is available at https://github.com/Y-L-LIU/MGTBench-2.0.

cs.AI↗

arXiv2024

Large Language Models Based JSON Parser Fuzzing for Bug Discovery and Behavioral Analysis

Fuzzing has been incredibly successful in uncovering bugs and vulnerabilities across diverse software systems. JSON parsers play a vital role in modern software development, and ensuring their reliability is of great importance. This research project focuses on leveraging Large Language Models (LLMs) to enhance JSON parser testing. The primary objectives are to generate test cases and mutants using LLMs for the discovery of potential bugs in open-source JSON parsers and the identification of behavioral diversities among them. We aim to uncover underlying bugs, plus discovering (and overcoming) behavioral diversities.

cs.SE↗

arXiv2016

Scalable Image Retrieval by Sparse Product Quantization

Fast Approximate Nearest Neighbor (ANN) search technique for high-dimensional feature indexing and retrieval is the crux of large-scale image retrieval. A recent promising technique is Product Quantization, which attempts to index high-dimensional image features by decomposing the feature space into a Cartesian product of low dimensional subspaces and quantizing each of them separately. Despite the promising results reported, their quantization approach follows the typical hard assignment of traditional quantization methods, which may result in large quantization errors and thus inferior search performance. Unlike the existing approaches, in this paper, we propose a novel approach called Sparse Product Quantization (SPQ) to encoding the high-dimensional feature vectors into sparse representation. We optimize the sparse representations of the feature vectors by minimizing their quantization errors, making the resulting representation is essentially close to the original data in practice. Experiments show that the proposed SPQ technique is not only able to compress data, but also an effective encoding technique. We obtain state-of-the-art results for ANN search on four public image datasets and the promising results of content-based image retrieval further validate the efficacy of our proposed method.

cs.CV↗