Publications (19)
MGC: A Compiler Framework Exploiting Compositional Blindness in Aligned LLMs for Malware Generation
Lu Yan, Zhuo Zhang, Xiangzhe Xu +5
Large language models (LLMs) have democratized software development, reducing the expertise barrier for programming complex applications. This accessibility extends to malicious so…
BEAGLE: Forensics of Deep Learning Backdoor Attack for Better Defense
Siyuan Cheng, Guanhong Tao, Yingqi Liu +8
Deep Learning backdoor attacks have a threat model similar to traditional cyber attacks. Attack forensics, a critical counter-measure for traditional cyber attacks, is hence of imp…
Backdoor Scanning for Deep Neural Networks through K-Arm Optimization
Guangyu Shen, Yingqi Liu, Guanhong Tao +5
Back-door attack poses a severe threat to deep learning systems. It injects hidden malicious behaviors to a model such that any input stamped with a special pattern can trigger suc…
LOTUS: Evasive and Resilient Backdoor Attacks through Sub-Partitioning
Siyuan Cheng, Guanhong Tao, Yingqi Liu +7
Backdoor attack poses a significant security threat to Deep Learning applications. Existing attacks are often not evasive to established backdoor detection techniques. This suscept…
Assistant or Actor? Student Trust, Control, and Delegation Regret When Using a General-Purpose AI Agent
Shiva Pochampally, Shengwei An, Yan Chen
When AI agents shift from answering questions to taking actions, users face a new problem: deciding what to delegate, to a system whose action space they cannot fully anticipate. W…
Constrained Optimization with Dynamic Bound-scaling for Effective NLPBackdoor Defense
Guangyu Shen, Yingqi Liu, Guanhong Tao +5
We develop a novel optimization method for NLPbackdoor inversion. We leverage a dynamically reducing temperature coefficient in the softmax function to provide changing loss landsc…
CENSOR: Defense Against Gradient Inversion via Orthogonal Subspace Bayesian Sampling
Kaiyuan Zhang, Siyuan Cheng, Guangyu Shen +5
Federated learning collaboratively trains a neural network on a global server, where each local client receives the current global model weights and sends back parameter updates (g…
DECK: Model Hardening for Defending Pervasive Backdoors
Guanhong Tao, Yingqi Liu, Siyuan Cheng +5
Pervasive backdoors are triggered by dynamic and pervasive input perturbations. They can be intentionally injected by attackers or naturally exist in normally trained models. They…
ManimAgent: Self-Evolving Multimodal Agents for Visual Education
Wenjia Jiang, Zongyuan Cai, Yuanhang Shao +7
Multi-round reflection lets agents built on large language models recover from failures within a single task, but each task remains an isolated episode: lessons learned across many…
FLIP: A Provable Defense Framework for Backdoor Mitigation in Federated Learning
Kaiyuan Zhang, Guanhong Tao, Qiuling Xu +8
Federated Learning (FL) is a distributed learning paradigm that enables different parties to train a model together for high quality and strong privacy protection. In this scenario…
Backdoor Attack through Frequency Domain
Tong Wang, Yuan Yao, Feng Xu +3
Backdoor attacks have been shown to be a serious threat against deep learning systems such as biometric authentication and autonomous driving. An effective backdoor attack could en…
Backdooring Masked Diffusion Language Models
Daniel Yiming Cao, Chengzhong Wang, Sheng-Yen Chou +3
Masked diffusion language models (MDLMs) are emerging as a compelling new paradigm for text generation, but their training-time security remains largely unexplored. Existing backdo…
Temporal Logic-Based Multi-Vehicle Backdoor Attacks against Offline RL Agents in End-to-end Autonomous Driving
Xuan Chen, Shiwei Feng, Zikang Xiong +6
Assessing the safety of autonomous driving (AD) systems against security threats, particularly backdoor attacks, is a stepping stone for real-world deployment. However, existing wo…
Elijah: Eliminating Backdoors Injected in Diffusion Models via Distribution Shift
Shengwei An, Sheng-Yen Chou, Kaiyuan Zhang +8
Diffusion models (DM) have become state-of-the-art generative models because of their capability to generate high-quality images from noises without adversarial training. However,…
Confidence Matters: Inspecting Backdoors in Deep Neural Networks via Distribution Transfer
Tong Wang, Yuan Yao, Feng Xu +3
Backdoor attacks have been shown to be a serious security threat against deep learning models, and detecting whether a given model has been backdoored becomes a crucial task. Exist…
UNIT: Backdoor Mitigation via Automated Neural Distribution Tightening
Siyuan Cheng, Guangyu Shen, Kaiyuan Zhang +5
Deep neural networks (DNNs) have demonstrated effectiveness in various fields. However, DNNs are vulnerable to backdoor attacks, which inject a unique pattern, called trigger, into…
Rapid Optimization for Jailbreaking LLMs via Subconscious Exploitation and Echopraxia
Guangyu Shen, Siyuan Cheng, Kaiyuan Zhang +6
Large Language Models (LLMs) have become prevalent across diverse sectors, transforming human life with their extraordinary reasoning and comprehension abilities. As they find incr…
Backdoor Vulnerabilities in Normally Trained Deep Learning Models
Guanhong Tao, Zhenting Wang, Siyuan Cheng +7
We conduct a systematic study of backdoor vulnerabilities in normally trained Deep Learning models. They are as dangerous as backdoors injected by data poisoning because both can b…
SOFT: Selective Data Obfuscation for Protecting LLM Fine-tuning against Membership Inference Attacks
Kaiyuan Zhang, Siyuan Cheng, Hanxi Guo +8
Large language models (LLMs) have achieved remarkable success and are widely adopted for diverse applications. However, fine-tuning these models often involves private or sensitive…