Academy of Mathematics and Systems Science, CAS
Colloquia & Seminars

Speaker:	王琦，国防科技大学
Inviter:
Title:	A Simple Yet Effective Strategy to Robustify the Meta Learning Paradigm
Language:	Chinese
Time & Venue:	2023.06.21 11:00 N213
Abstract:	Meta learning is a promising paradigm to enable skill transfer across tasks. Most previous methods employ the empirical risk minimization principle in optimization. However, the resulting worst fast adaptation to a subset of tasks can be catastrophic in risk-sensitive scenarios. To robustify fast adaptation, this paper optimizes meta learning pipelines from a distributionally robust perspective and meta trains models with the measure of expected tail risk. We take the two-stage strategy as heuristics to solve the robust meta learning problem, controlling the worst fast adaptation cases at a certain probabilistic level. Experimental results show that our simple method can improve the robustness of meta learning to task distribution, alleviate the heavy tail effect in risk, and reduce the conditional expectation of the worst fast adaptation risk.