RL

Event Weekly OptiML Lab Group Meeting Short summary In this seminar, I will talk about my own paper “Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion” (Lee et al.

Event Weekly OptiML Lab Group Meeting Short summary In this talk, I will first (somewhat rigorously) introduce the framework of reinforcement learning with human feedback (RLHF). Then I will go over three recent breakthroughs in the analysis and improvement of RLHF.

Event Weekly OSI Lab Seminar Short summary In this talk, I will first (somewhat rigorously) introduce the framework of reinforcement learning with human feedback (RLHF). Then I will go over three recent breakthroughs in the analysis and improvement of RLHF.

Event Weekly OSI Lab Seminar Short summary In this seminar, I start by introducing the problem of bandits, overall proof techniques for the fundamental lower bounds, and a brief overview of variants of bandits.

Event Weekly OSI Lab Seminar Short summary In this seminar, I will talk about the paper “Improved Sample Complexity for Reward-free Reinforcement Learning under Low-rank MDPs” (Cheng et al., ICLR 2023).

Event Weekly OptiML Lab Group Meeting Short summary In this seminar, I will first introduce bandits, then move on to talk about a special class of bandits, called combinatorial (semi-)bandits, and even more specific example of it, namely, matroid bandits.

Event Weekly OptiML Lab Group Meeting Short summary In this seminar, I will talk about my own paper “Nearly Optimal Latent State Decoding in Block MDPs” (Jedra et al., arXiv 2022).

Event Weekly OSI Lab Seminar Short summary In this seminar, I will talk about the generalized linear bandits and logistic bandits. Especially for the logistic bandits, I will talk about its connection to the generalized linear bandits and some recent progresses on it from Criteo AI Lab.

Event Weekly OSI Lab Seminar Short summary In this seminar, I will talk about the paper “Clustering in Block Markov Chains” (Sanders et al., Ann. Stat. 2020). Abstract (taken directly from the paper)

A novel problem setting where heterogeneous multi-agent bandits collaborate over a network to minimize their group regret. To deal with the high communication complexity of the classic flooding protocol combined with UCB, a new network protocol called Flooding with Absorption (FwA) is proposed. Theoretical and empirical analyses are provded for flooding and FwA, showing the efficacy of our proposed FwA.