Summary of Investigating Relational State Abstraction in Collaborative Marl, by Sharlin Utke et al.
Investigating Relational State Abstraction in Collaborative MARL
by Sharlin Utke, Jeremie Houssineau, Giovanni Montana
First submitted to arxiv on: 19 Dec 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Machine Learning (cs.LG); 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 proposed paper investigates how relational state abstraction affects sample efficiency and performance in collaborative Multi-Agent Reinforcement Learning (MARL). The authors introduce MARC, a simple yet effective critic architecture that incorporates spatial relational inductive biases by transforming the state into a spatial graph. This is achieved through a relational graph neural network processing. The performance of MARC is evaluated across six collaborative tasks, including a novel environment with heterogeneous agents. The results demonstrate improvements in both sample efficiency and asymptotic performance, as well as generalization capabilities. Comparisons are made against state-of-the-art MARL baselines, highlighting the potential benefits of relational state abstraction without requiring complex designs or task-specific engineering. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how to make team AI (Multi-Agent Reinforcement Learning) work better by understanding how agents relate to each other and their environment. The authors created a new way called MARC that uses spatial relationships to help agents learn faster and better. They tested MARC on six different tasks and found it outperformed current methods in terms of learning speed and ability to generalize to new situations. |
Keywords
» Artificial intelligence » Generalization » Graph neural network » Reinforcement learning
