CS Seminar - Sijia Liu - MIT-IBM Watson AI Lab, IBM Research

By: mh1271 | Posted on: Sept. 26, 2018, 3:41 p.m.

The Department of Computer Science is pleased to announce that we will offer a seminar presentation.  All are welcome to attend.  Graduate students are invited to attend these computer science seminars as part of their degree fulfillment requirements. 

Out of respect for the guest speaker, no admittance to the seminar room will be granted after the scheduled start time.

Who:    Sijia Liu, MIT-IBM Watson AI Lab, IBM Research
When:  Oct 12, 2018, from 1-2 p.m.
Where: Comal Bldg CS Conference Room 212

 

 Titleļ¼š Zeroth-order optimization with application to adversarial machine learning 

 Abstract: 

 Zeroth-order (gradient-free) optimization is increasingly embraced for solving big data and machine learning problems when explicit expressions of the gradients are difficult or even infeasible to obtain.

This talk will cover some recent advances in zeroth-order (ZO) optimization methods in both theory and applications. On the theory side, I will elaborate on fundamental basics of ZO algorithms including convergence rate and query complexity analysis, and make comparisons to their first-order counterparts. On the application side, I will highlight one appealing application of ZO optimization to studying the robustness of deep neural networks -- practical and efficient adversarial attacks that generate adversarial examples from a black-box deep learning model in information-limited scenarios. I will also summarize potential research directions regarding ZO optimization, big data challenges and some open-ended machine learning problems.

Bio:

 Sijia Liu is a Research Staff Member at MIT-IBM Watson AI Lab, IBM research. He received the Ph.D. degree (with All University Doctoral Prize) in electrical and computer engineering from Syracuse University, NY, USA, in 2016. He was a Postdoctoral Research Fellow at the University of Michigan, Ann Arbor, before joining in IBM Research. His recent research interests include adversarial machine learning, gradient-free optimization, machine learning algorithms, and graph signal processing. He has published papers in top machine learning and signal processing conferences, including NIPS, AISTATS, ACMMM, ICASSP, and IEEE T-SP. He received the Best Student Paper Award at the 42nd IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). He is also the co-chair of workshop “Signal Processing for Adversarial Machine Learning” at GlobalSIP 2018.

 

Deadline: Oct. 12, 2018, midnight