Summary of Utility-based Reinforcement Learning: Unifying Single-objective and Multi-objective Reinforcement Learning, by Peter Vamplew et al.
Utility-Based Reinforcement Learning: Unifying Single-objective and Multi-objective Reinforcement Learning
by Peter Vamplew, Cameron Foale, Conor F. Hayes, Patrick Mannion, Enda Howley, Richard Dazeley, Scott Johnson, Johan Källström, Gabriel Ramos, Roxana Rădulescu, Willem Röpke, Diederik M. Roijers
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 extends the utility-based paradigm in multi-objective reinforcement learning (MORL) to single-objective reinforcement learning (RL), exploring benefits such as multi-policy learning across uncertain objectives, risk-aware RL, discounting, and safe RL. The proposed approach enables agents to learn from both environmental rewards and user-defined utilities, which could lead to more effective decision-making in complex scenarios. By examining the algorithmic implications of a utility-based approach, this research aims to improve the capabilities of single-objective RL systems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about using computers to make better decisions when there are multiple things we want it to achieve. Right now, computer programs can only do one thing at a time, but what if they could learn to do many things and choose which one to prioritize? This paper shows how this could be possible by combining two different ways of learning: one that rewards the program for doing certain actions, and another that gives it a score based on how well it does. This could help computers make better choices in situations where there are risks or uncertainties involved. |
Keywords
* Artificial intelligence * Reinforcement learning
