Record Details
Evaluation of Parallel Monte Carlo Tree Search Algorithms in Python
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | Evaluation of Parallel Monte Carlo Tree Search Algorithms in Python |
| Names |
Jothi, Shankar
(creator) Fern, Alan (advisor) |
| Date Issued | 2015-06-08 (iso8601) |
| Note | Graduation date: 2015 |
| Abstract | Monte-Carlo Tree Search (MCTS) is an online-planning algorithm for decision-theoretic planning in domains with stochastic and combinatorial structure. The general applicability of MCTS makes it an ideal first choice to investigate when developing planners for complex applications requiring automated control and planning. The first contribution of this thesis is to develop a new open-source Python MCTS planning library (PyPlan), which allows for fast prototyping of MCTS solutions in new domains. While the library can offer a final solution for applications that do not impose strict real-time constraints, for other applications the interpreted nature of Python may not meet the time constraints. This issue leads to the second contribution of this thesis, which is to study the effectiveness of various parallel versions of MCTS for speeding up the performance of PyPlan. We evaluate these algorithms on several complex planning problems using cloud-based resources. Our findings point to a preferred method for parallelizing PyPlan and also indicates that some prior methods are quite ineffective due to particular aspects of the Python language. |
| Genre | Thesis/Dissertation |
| Access Condition | http://creativecommons.org/licenses/by/3.0/us/ |
| Topic | Artificial Intelligence |
| Identifier | http://hdl.handle.net/1957/56230 |
