Summary of Global Rewards in Restless Multi-armed Bandits, by Naveen Raman et al.
Global Rewards in Restless Multi-Armed Bandits
by Naveen Raman, Zheyuan Ryan Shi, Fei Fang
First submitted to arxiv on: 2 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
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 paper proposes a new algorithm for restless multi-armed bandits with global non-separable rewards. The existing models assume separability of rewards into a sum across arms, but this limitation is addressed by developing the Linear- and Shapley-Whittle indices, which extend Whittle indices from RMABs to RMAB-Gs. To overcome potential failures in highly non-linear reward functions, adaptive policies are proposed, including iterative index computation and combining indices with Monte-Carlo Tree Search (MCTS). The new algorithm outperforms baselines and index-based policies in both synthetic and real-world data. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about a new way to solve a problem called restless multi-armed bandits. This means that when you pull an arm, it changes how the other arms will behave later on. Right now, there are ways to solve this problem but they only work if the rewards from each arm add up in a simple way. The new solution works even if the rewards don’t add up in a simple way and can make good decisions based on different types of data. |