Summary of Contextual Bandits with Arm Request Costs and Delays, by Lai Wei et al.
Contextual Bandits with Arm Request Costs and Delays
by Lai Wei, Ambuj Tewari, Michael A. Cianfrocco
First submitted to arxiv on: 17 Oct 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
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 introduces an extension of the contextual bandit problem where new sets of arms can be requested with stochastic time delays and associated costs. In this setting, the learner selects multiple arms from a decision set, taking one unit of time per selection. The problem is framed as a semi-Markov decision process (SMDP). Arm contexts, request times, and costs follow an unknown distribution. The goal is to minimize regret with respect to the optimal policy achieving maximum average reward. Algorithms leverage the Bellman optimality equation to select arms and determine when to request new ones. Regret upper bounds align with established results in contextual bandit literature under realizability assumption. Experiments on simulated data and movie recommendation dataset show consistent performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at a special type of decision-making problem called a contextual bandit. In this situation, you can ask for more options or “arms” to choose from, but it takes some time and may cost something. The goal is to figure out how to make good decisions while considering these costs and delays. The researchers come up with new algorithms that help solve this problem by using equations to decide which arms to choose and when to ask for new ones. They test their ideas on fake data and a real movie recommendation dataset, and the results match what they predicted. |
