https://github.com/animesh-chourey/pommerman-agent-customized
Modified the Monte Carlo Tree Search algorithm by introducing progressive bias to refine the selection process on Pommerman game setup.
https://github.com/animesh-chourey/pommerman-agent-customized
heuristic-search-algorithms monte-carlo-tree-search pommerman
Last synced: 3 months ago
JSON representation
Modified the Monte Carlo Tree Search algorithm by introducing progressive bias to refine the selection process on Pommerman game setup.
- Host: GitHub
- URL: https://github.com/animesh-chourey/pommerman-agent-customized
- Owner: Animesh-Chourey
- Created: 2022-08-28T20:03:30.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2022-08-28T20:36:38.000Z (over 2 years ago)
- Last Synced: 2025-01-11T19:45:26.598Z (4 months ago)
- Topics: heuristic-search-algorithms, monte-carlo-tree-search, pommerman
- Language: Java
- Homepage:
- Size: 481 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
A modified MCTS agent has been implemented as part of the assignment.
The selection phase of the MCTS algorithm has been improvised through the process of adding a progressive strategy called progressive bias. The customised agent is designed to play the pommerman game.
The framework of the game can be found at : [java pommerman](https://github.com/GAIGResearch/java-pommerman)
Below are the instructions for executing this customised agent:
* The package required to run our enhanced MCTS agent (MSCT2) has been integrated into the folder "customised agent".
* The GameState.java file has been updated and needs to be replaced with the original one present.
* To run our agent(MCTS2), edit the configurations for Run.java and pass the value '7' in place of the indices 4-7.
For example, the configurations [0 1 1 -1 3 4 5 7] would run a tournament of Free for All between Simple Player, RHEA, MCTS and MCTS2(our customised agent) in full observability.