Towards Trustworthy Machine Learning: Training-time and test-time integrity

Speaker: Dr. Minhao CHENG
         Assistant Professor
         Department of Computer Science and Engineering
         Hong Kong University of Science and Technology

Title:  "Towards Trustworthy Machine Learning: Training-time
         and test-time integrity"

Date:   Monday, 24 October 2022

Time:   4:00pm - 5:00pm

Venue:  Lecture Theater F
        (Leung Yat Sing Lecture Theater); near lift 25/26
        HKUST


Abstract:

Although machine learning methods, such as deep learning, have achieved
unprecedented success over a variety of tasks across different domains,
because of their black-box nature, deploying these methods often leads to
concerns as to their safety, especially in security-critical environments.
In this talk, I will show several emerging research areas related to
trustworthy machine learning. First, I will start by introducing the
backdoor attacks that insert backdoor functionality into models to make
them perform maliciously on trigger instances while maintaining similar
performance on normal data. I will also talk about existing studies to
detect and defend against backdoor attacks. Furthermore, for the test-time
integrity, I will introduce research topics concerning adversarial
robustness including attacks and defenses, and verification.


***************
Biography:

Dr. Minhao Cheng is an Assistant Professor in the Department of Computer
Science and Engineering (CSE), Hong Kong University of Science and
Technology (HKUST). He obtained his Ph.D. degree in the Department of
Computer Science from University of California, Los Angeles under the
supervision of Prof. Cho-Jui Hsieh. His research focus is broadly on
machine learning with a focus on machine learning robustness and AutoML.
He has published over 20 papers on top tier AI conferences including ICML,
NeurIPS, ICLR, ACL, AAAI etc. He is a recipient of ICLR 2021 Outstanding
Paper Award.