4 papers
MEEA: Mere Exposure Effect-Driven Confrontational Optimization for LLM Jailbreaking
Jianyi Zhang, Shizhao Liu, Ziyin Zhou +1
The rapid advancement of large language models (LLMs) has intensified concerns about the robustness of their safety alignment. While existing jailbreak studies explore both single-…
CAPTURE: A Benchmark and Evaluation for LVLMs in CAPTCHA Resolving
Jianyi Zhang, Ziyin Zhou, Xu Ji +2
Benefiting from strong and efficient multi-modal alignment strategies, Large Visual Language Models (LVLMs) are able to simulate human visual and reasoning capabilities, such as so…
Oedipus and the Sphinx: Benchmarking and Improving Visual Language Models for Complex Graphic Reasoning
Jianyi Zhang, Xu Ji, Ziyin Zhou +5
Evaluating the performance of visual language models (VLMs) in graphic reasoning tasks has become an important research topic. However, VLMs still show obvious deficiencies in simu…
FL-PLAS: Federated Learning with Partial Layer Aggregation for Backdoor Defense Against High-Ratio Malicious Clients
Jianyi Zhang, Ziyin Zhou, Yilong Li +1
Federated learning (FL) is gaining increasing attention as an emerging collaborative machine learning approach, particularly in the context of large-scale computing and data system…