Probability Theory

Heavy-tail behaviour of SGD - Part 1
Event Weekly OSI Lab Seminar Short summary In this seminar, I will talk about a recent line of works that propose to analyze SGD under heavy-tail noise assumptions. Abstract One of the popular ways of analyzing the behavior of SGD and SGDm(SGD with momentum) is by considering it as a discretization of Langevin-type SDE.
Aug 14, 2020
A Unified Confidence Sequence for Generalized Linear Models, with Applications to Bandits
We present a unified likelihood ratio-based confidence sequence (CS) for any (self-concordant) generalized linear model (GLM) that is …
Junghyun Lee, Se-Young Yun, Kwang-Sung Jun
PDF Cite Project Poster
A Unified Confidence Sequence for Generalized Linear Models, with Applications to Bandits
Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion
Logistic bandit is a ubiquitous framework of modeling users’ choices, e.g., click vs. no click for advertisement recommender …
Junghyun Lee, Se-Young Yun, Kwang-Sung Jun
PDF Cite Project Poster Slides
Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion
Deep Learning Theory - Optimization
specific topics tbd
Junghyun Lee
Sample Complexity of Learning in Structured Markov Chains and MDPs
(tbd)
Junghyun Lee
Theoretical Analyses of Reinforcement Learning with Human Feedback (RLHF) and Related Problems
(tbd)
Junghyun Lee