Summary of Distributed Multi-task Learning For Stochastic Bandits with Context Distribution and Stage-wise Constraints, by Jiabin Lin and Shana Moothedath
Distributed Multi-Task Learning for Stochastic Bandits with Context Distribution and Stage-wise Constraints
by Jiabin Lin, Shana Moothedath
First submitted to arxiv on: 21 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Multiagent Systems (cs.MA)
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 presents a problem in distributed multi-task learning for stochastic linear contextual bandits with heterogeneous agents. It extends conservative linear bandits to a distributed setting where multiple agents tackle different tasks while adhering to performance constraints. The exact context is unknown, and only a context distribution is available. The authors propose DiSC-UCB, a distributed UCB algorithm that constructs a pruned action set and shares estimates among agents via a central server. They prove regret and communication bounds for the algorithm and extend it to a setting where agents are unaware of baseline rewards. Empirical validation shows strong performance on synthetic and real-world data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, scientists explore how multiple computers can work together to make predictions about different things, like stock prices or weather forecasts. The problem is that each computer only has some information about what’s happening, but they all need to agree on the best prediction. The researchers created a new way for these computers to share information and make decisions while staying within certain limits. They tested this method on fake data and real data from a movie rating website, and it worked well. |
Keywords
* Artificial intelligence * Multi task
