The Security Issues in Machine Learning Applicatio

报告题目:The Security Issues in Machine Learning Applications

报告人:张生智(Shengzhi Zhang

报告时间:2019-07-09  上午  1030

报告地点:信息楼 420

报告摘要:Recently Machine learning is widely used in various areas, e.g., image recognition, speech recognition, object detection, autonomous driving, strategic gaming, etc. However, its security issues are not fully understood. In this talk, I will summarize the inherent security problems in machine learning and our works investigating them. Then, I will focus on ASR (automatic speech recognition) systems, like Google Voice, Cortana, Amazon Echo, of which the performance has been significantly improved by the integration of deep learning. I will show that not only are more practical and surreptitious attacks against ASR systems feasible but they can even be automatically constructed. Specifically, the voice commands can be stealthily embedded into songs, which, when played, can effectively control the target system through ASR without being noticed. Our research shows that this can be done automatically against real world ASR systems, and even devices like Google Home, Amazon Echo, Apple Siri, etc.

报告人简介:Dr. Shengzhi Zhang is Assistant Professor in Computer Science department at Boston University Metropolitan College. Before joining BU, he worked as assistant professor at Florida Institute of Technology, and researcher in IBM research lab, Honeywell Aerospace, Cisco R&D. He got his Ph.D from the Department of Computer Science and Engineering at Penn State University in 2012 and bachelor from the Department of Electrical Engineering at Tongji University in 2006. Dr. Zhang's research interest includes, but not limited to machine learning security, IoT security, system security, vehicle security, and mobile security. He has published many papers and served as program committee members in top tier security conferences and journals.

1