Time: December 22, 2021 (Wednesday) 18:30-20:30
Venue: Tencent Meeting ID (online lecture): 176 546 184
Speaker: Chai Yidong, Researcher, School of Management, Hefei University of Technology
Host: Associate Professor Zhang Mingyue
Topic：An Explainable Multi-Modal Hierarchical Attention Model for Developing Phishing Threat Intelligence
Phishing website attack, as one of the most persistent forms of cyber threats, evolves and remains a major cyber threat. Previous deep representation-based methods fail to analyze two important modalities of website content: textual information and visual design. Moreover, the interpretability of these deep learning based methods is limited. As such, we propose a multi-modal hierarchical attention model (MMHAM) which jointly learns the deep fraud cues from the three major modalities of website content for phishing website detection. Specifically, MMHAM features an innovative shared dictionary learning approach for aligning representations from different modalities in the attention mechanism. In our evaluation experiments, the proposed MMHAM provided a hierarchical interpretability system from which we could develop phishing threat intelligence to inform phishing websites detection at different levels.
Yidong Chai is a Huangshan Young Scholar of Hefei University of Technology, a researcher and a Ph. D. from the Department of Management Science and Engineering, School of Economics and Management, Tsinghua University. His research areas include machine learning, big data management and technology, business intelligence, cyberspace management, and smart healthcare. He has hosted three National Natural Science Youth Funds and the University's Young Scholar Support Cultivation Fund. He has published research results as first author or corresponding author in top international journals such as MISQ(UTD 24, ABS 4*), IEEE TDSC(CCF A), JMIS(FT 50) and international conferences such as ICIS, WITS, PACIS, CSWIM, INFORMS Workshop on Data Science. He is a reviewer for top international journals such as POM and IEEE TDSC.