Antonoglou, Ioannis;
(2023)
Learning to Search in Reinforcement Learning.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
PhD_Thesis_draft_corrected.pdf - Accepted Version Download (4MB) | Preview |
Abstract
In this thesis, we investigate the use of search based algorithms with deep neural networks to tackle a wide range of problems ranging from board games to video games and beyond. Drawing inspiration from AlphaGo, the first computer program to achieve superhuman performance in the game of Go, we developed a new algorithm AlphaZero. AlphaZero is a general reinforcement learning algorithm that combines deep neural networks with a Monte Carlo Tree search for planning and learning. Starting completely from scratch, without any prior human knowledge beyond the basic rules of the game, AlphaZero managed to achieve superhuman performance in Go, chess and shogi. Subsequently, building upon the success of AlphaZero, we investigated ways to extend our methods to problems in which the rules are not known or cannot be hand-coded. This line of work led to the development of MuZero, a model-based reinforcement learning agent that builds a deterministic internal model of the world and uses it to construct plans in its imagination. We applied our method to Go, chess, shogi and the classic Atari suite of video-games, achieving superhuman performance. MuZero is the first RL algorithm to master a variety of both canonical challenges for high performance planning and visually complex problems using the same principles. Finally, we describe Stochastic MuZero, a general agent that extends the applicability of MuZero to highly stochastic environments. We show that our method achieves superhuman performance in stochastic domains such as backgammon and the classic game of 2048 while matching the performance of MuZero in deterministic ones like Go.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Learning to Search in Reinforcement Learning |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2023. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10166147 |
Archive Staff Only
 |
View Item |
