1 citations · 2 across the 20 of their papers we have counts for
20 papers
Tracing Provenance and Detecting Tampering with Complementary LLM Watermarks
Xiaoyan Feng, Yanjun Zhang, He Zhang +2
Watermarking LLM-generated text is an important task for tracing its provenance. Existing LLM watermarks preserve provenance under editing, but this same robustness allows an adver…
Defending Jailbreak Attacks on Large Language Models via Manifold Trajectory Kinetics
Hangtao Zhang, Yucheng Zhao, Sishun Liu +8
Jailbreak prompts can bypass alignment guardrails in large language models (LLMs) and elicit unsafe outputs, making reliable deployment-time detection critical. Prior detection app…
SNARE: Adaptive Scenario Synthesis for Eliciting Overeager Behavior in Coding Agents
Yubin Qu, Yi Liu, Gelei Deng +4
A coding agent executes a benign task as a sequence of shell, file, and network actions, any of which can quietly exceed the authorized scope while the task still completes. We cal…
Rethinking Federated Unlearning via the Lens of Memorization
Jiaheng Wei, Yanjun Zhang, He Zhang +5
Federated learning (FL) increasingly needs machine unlearning to comply with privacy regulations. However, existing federated unlearning approaches may overlook the overlapping inf…
Overeager Coding Agents: Measuring Out-of-Scope Actions on Benign Tasks
Yubin Qu, Ying Zhang, Yanjun Zhang +4
Coding agents now run autonomously with shell, file, and network privileges. When a user issues a benign request, the agent sometimes does more than asked: it deletes unrelated fil…
How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Study
Zhihao Chen, Ying Zhang, Yi Liu +7
Large Language Model (LLM) agents increasingly rely on third-party skills that operate within privileged execution environments and routinely handle sensitive credentials, yet how…
Transferable Backdoor Attacks for Code Models via Sharpness-Aware Adversarial Perturbation
Shuyu Chang, Haiping Huang, Yanjun Zhang +3
Code models are increasingly adopted in software development but remain vulnerable to backdoor attacks via poisoned training data. Existing backdoor attacks on code models face a f…
"Do Not Mention This to the User": Detecting and Understanding Malicious Agent Skills in the Wild
Yi Liu, Zhihao Chen, Yanjun Zhang +4
LLM-based coding agents increasingly rely on third-party extensions called skills, which bundle natural language instructions and helper scripts that execute with full user privile…
UnlearnShield: Shielding Forgotten Privacy against Unlearning Inversion
Lulu Xue, Shengshan Hu, Wei Lu +6
Machine unlearning is an emerging technique that aims to remove the influence of specific data from trained models, thereby enhancing privacy protection. However, recent research h…
Less Is More -- Until It Breaks: Security Pitfalls of Vision Token Compression in Large Vision-Language Models
Xiaomei Zhang, Zhaoxi Zhang, Leo Yu Zhang +3
Visual token compression is widely adopted to improve the inference efficiency of Large Vision-Language Models (LVLMs), enabling their deployment in latency-sensitive and resource-…
Character-Level Perturbations Disrupt LLM Watermarks
Zhaoxi Zhang, Xiaomei Zhang, Yanjun Zhang +5
Large Language Model (LLM) watermarking embeds detectable signals into generated text for copyright protection, misuse prevention, and content detection. While prior studies evalua…
When Better Features Mean Greater Risks: The Performance-Privacy Trade-Off in Contrastive Learning
Ruining Sun, Hongsheng Hu, Wei Luo +4
With the rapid advancement of deep learning technology, pre-trained encoder models have demonstrated exceptional feature extraction capabilities, playing a pivotal role in the rese…
Towards Reliable Forgetting: A Survey on Machine Unlearning Verification
Lulu Xue, Shengshan Hu, Wei Lu +7
With growing demands for privacy protection, security, and legal compliance (e.g., GDPR), machine unlearning has emerged as a critical technique for ensuring the controllability an…
Improving Generalization of Universal Adversarial Perturbation via Dynamic Maximin Optimization
Yechao Zhang, Yingzhe Xu, Junyu Shi +4
Deep neural networks (DNNs) are susceptible to universal adversarial perturbations (UAPs). These perturbations are meticulously designed to fool the target model universally across…
Exploring Gradient-Guided Masked Language Model to Detect Textual Adversarial Attacks
Xiaomei Zhang, Zhaoxi Zhang, Yanjun Zhang +4
Textual adversarial examples pose serious threats to the reliability of natural language processing systems. Recent studies suggest that adversarial examples tend to deviate from t…
Test-Time Backdoor Detection for Object Detection Models
Hangtao Zhang, Yichen Wang, Shihui Yan +7
Object detection models are vulnerable to backdoor attacks, where attackers poison a small subset of training samples by embedding a predefined trigger to manipulate prediction. De…
Large Language Model Watermark Stealing With Mixed Integer Programming
Zhaoxi Zhang, Xiaomei Zhang, Yanjun Zhang +5
The Large Language Model (LLM) watermark is a newly emerging technique that shows promise in addressing concerns surrounding LLM copyright, monitoring AI-generated text, and preven…
Detector Collapse: Physical-World Backdooring Object Detection to Catastrophic Overload or Blindness in Autonomous Driving
Hangtao Zhang, Shengshan Hu, Yichen Wang +5
Object detection tasks, crucial in safety-critical systems like autonomous driving, focus on pinpointing object locations. These detectors are known to be susceptible to backdoor a…
Towards Model Extraction Attacks in GAN-Based Image Translation via Domain Shift Mitigation
Di Mi, Yanjun Zhang, Leo Yu Zhang +4
Model extraction attacks (MEAs) enable an attacker to replicate the functionality of a victim deep neural network (DNN) model by only querying its API service remotely, posing a se…
Masked Language Model Based Textual Adversarial Example Detection
Xiaomei Zhang, Zhaoxi Zhang, Qi Zhong +4
Adversarial attacks are a serious threat to the reliable deployment of machine learning models in safety-critical applications. They can misguide current models to predict incorrec…