Summary of Indexed Minimum Empirical Divergence-based Algorithms For Linear Bandits, by Jie Bian and Vincent Y. F. Tan
Indexed Minimum Empirical Divergence-Based Algorithms for Linear Bandits
by Jie Bian, Vincent Y. F. Tan
First submitted to arxiv on: 24 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Information Theory (cs.IT)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The Indexed Minimum Empirical Divergence (IMED) algorithm offers a stronger theoretical guarantee of asymptotic optimality for multi-armed bandits compared to the Kullback-Leibler Upper Confidence Bound (KL-UCB) algorithm. IMED has been shown to empirically outperform UCB-based algorithms and Thompson Sampling. However, its application to contextual bandits with linear payoffs remained unexplored. This paper presents novel linear versions of IMED, known as the LinIMED family, which provide a (d) upper regret bound for contextual bandits. Empirical studies demonstrate that LinIMED and its variants outperform traditional algorithms like LinUCB and Linear Thompson Sampling in certain regimes. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The Indexed Minimum Empirical Divergence (IMED) algorithm is a new way to solve a problem called the multi-armed bandit problem. This problem is important because it helps us make good decisions when we don’t know which option will work best. IMED is better than some other algorithms at solving this problem. But so far, it’s only been used for one type of problem. The authors of this paper want to use IMED for a new type of problem called contextual bandits with linear payoffs. They create a new family of algorithms called LinIMED that can be used for this problem. Their algorithm is better than some other algorithms at solving this problem. |
