5 papers
SPARK: Jailbreaking T2V Models by Synergistically Prompting Auditory and Recontextualized Knowledge
Zonghao Ying, Moyang Chen, Nizhang Li +6
Jailbreak attacks can circumvent model safety guardrails and reveal critical blind spots. Prior attacks on text-to-video (T2V) models typically add adversarial perturbations to obv…
Learnable Game-theoretic Policy Optimization for Data-centric Self-explanation Rationalization
Yunxiao Zhao, Zhiqiang Wang, Xingtong Yu +3
Rationalization, a data-centric framework, aims to build self-explanatory models to explain the prediction outcome by generating a subset of human-intelligible pieces of the input…
VisMoDAl: Visual Analytics for Evaluating and Improving Corruption Robustness of Vision-Language Models
Huanchen Wang, Wencheng Zhang, Zhiqiang Wang +2
Vision-language (VL) models have shown transformative potential across various critical domains due to their capability to comprehend multi-modal information. However, their perfor…
GeoSketch: A Neural-Symbolic Approach to Geometric Multimodal Reasoning with Auxiliary Line Construction and Affine Transformation
Shichao Weng, Zhiqiang Wang, Yuhua Zhou +5
Geometric Problem Solving (GPS) poses a unique challenge for Multimodal Large Language Models (MLLMs), requiring not only the joint interpretation of text and diagrams but also ite…
Explaining Black-box Language Models with Knowledge Probing Systems: A Post-hoc Explanation Perspective
Yunxiao Zhao, Hao Xu, Zhiqiang Wang +3
Pre-trained Language Models (PLMs) are trained on large amounts of unlabeled data, yet they exhibit remarkable reasoning skills. However, the trustworthiness challenges posed by th…