Summary of Tabular and Deep Reinforcement Learning For Gittins Index, by Harshit Dhankhar and Kshitij Mishra and Tejas Bodas
Tabular and Deep Reinforcement Learning for Gittins Index
by Harshit Dhankhar, Kshitij Mishra, Tejas Bodas
First submitted to arxiv on: 2 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Performance (cs.PF); Machine Learning (stat.ML)
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 This paper proposes novel tabular (QGI) and deep reinforcement learning (DGN) algorithms for solving multi-arm bandit problems. The Gittins index policy is known to be optimal, but it relies on knowing the Markovian state transition probabilities, which are often unknown in realistic scenarios. To overcome this limitation, the authors develop RL algorithms that explore the state space to learn these indices while exploiting to maximize the reward collected. Compared to existing methods, the proposed algorithms demonstrate better empirical convergence to the Gittins index, with lower run times and reduced storage requirements. The paper also showcases a key application of the algorithm in minimizing mean flowtime in job scheduling problems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research develops new ways for computers to learn and make decisions when there are many possible choices. It’s like trying to decide which job to do first when you have many tasks to complete, but each task has a different time required to finish it. The computer needs to figure out the best way to choose the jobs in order to get all of them done as quickly as possible. The authors create two new methods for computers to learn how to make these decisions and show that they work better than other existing methods. |
Keywords
» Artificial intelligence » Reinforcement learning
