Ensemble decision making in real-time games

Philip Rodgers; John Levine; Damien Anderson

doi:10.1109/CIG.2018.8490401

Ensemble decision making in real-time games

Philip Rodgers, John Levine, Damien Anderson

Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

78 Downloads (Pure)

Abstract

This paper describes an Ensemble Agent for the classic arcade game Ms. Pac-Man. Our approach decomposes the problem into sub-goals. An expert agent is created for each sub-goal, with all experts reporting to a central arbiter. Our Ensemble Agent has achieved the AI world record for the arcade version of Ms. Pac-Man with a score of 162,280. For comparison, a MCTS-based monolithic agent was also created, based on the same accurate forward model that the Ensemble Agent uses, reaching a score of 115,180.

Original language	English
Title of host publication	2018 IEEE Conference on Computational Intelligence and Games (CIG)
Publisher	IEEE
Number of pages	8
ISBN (Electronic)	978-1-5386-4359-4
ISBN (Print)	978-1-5386-4360-0
DOIs	https://doi.org/10.1109/CIG.2018.8490401
Publication status	Published - 15 Oct 2018

Publication series

Name
ISSN (Print)	2325-4270
ISSN (Electronic)	2325-4289

Keywords

ensemble
mcts
Pac-Man
real-time
decision

Documents and Links

10.1109/CIG.2018.8490401

Rodgers, P. et al (2018) Ensemble decision making in real-time games
Copyright © 2018 IEEE. This is a post-peer review, pre-copy edited version of the article and may differ slightly from the final published version of the work. This work is made available in line with the publisher policy. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
Accepted author manuscript, 264 KB

Fingerprint Dive into the research topics of 'Ensemble decision making in real-time games'. Together they form a unique fingerprint.

Cite this

@inproceedings{7621442588fa4012af3b39042f136567,

title = "Ensemble decision making in real-time games",

abstract = "This paper describes an Ensemble Agent for the classic arcade game Ms. Pac-Man. Our approach decomposes the problem into sub-goals. An expert agent is created for each sub-goal, with all experts reporting to a central arbiter. Our Ensemble Agent has achieved the AI world record for the arcade version of Ms. Pac-Man with a score of 162,280. For comparison, a MCTS-based monolithic agent was also created, based on the same accurate forward model that the Ensemble Agent uses, reaching a score of 115,180.",

keywords = "ensemble, mcts, Pac-Man, real-time, decision",

author = "Philip Rodgers and John Levine and Damien Anderson",

note = "Changed template to proceedings ET 10/2/20 Acceptance in SAN AAM: no embargo In scope of REF policy - has ISSN. ",

year = "2018",

month = oct,

day = "15",

doi = "10.1109/CIG.2018.8490401",

language = "English",

isbn = "978-1-5386-4360-0",

publisher = "IEEE",

booktitle = "2018 IEEE Conference on Computational Intelligence and Games (CIG)",

}

TY - GEN

T1 - Ensemble decision making in real-time games

AU - Rodgers, Philip

AU - Levine, John

AU - Anderson, Damien

N1 - Changed template to proceedings ET 10/2/20 Acceptance in SAN AAM: no embargo In scope of REF policy - has ISSN.

PY - 2018/10/15

Y1 - 2018/10/15

N2 - This paper describes an Ensemble Agent for the classic arcade game Ms. Pac-Man. Our approach decomposes the problem into sub-goals. An expert agent is created for each sub-goal, with all experts reporting to a central arbiter. Our Ensemble Agent has achieved the AI world record for the arcade version of Ms. Pac-Man with a score of 162,280. For comparison, a MCTS-based monolithic agent was also created, based on the same accurate forward model that the Ensemble Agent uses, reaching a score of 115,180.

AB - This paper describes an Ensemble Agent for the classic arcade game Ms. Pac-Man. Our approach decomposes the problem into sub-goals. An expert agent is created for each sub-goal, with all experts reporting to a central arbiter. Our Ensemble Agent has achieved the AI world record for the arcade version of Ms. Pac-Man with a score of 162,280. For comparison, a MCTS-based monolithic agent was also created, based on the same accurate forward model that the Ensemble Agent uses, reaching a score of 115,180.

KW - ensemble

KW - mcts

KW - Pac-Man

KW - real-time

KW - decision

U2 - 10.1109/CIG.2018.8490401

DO - 10.1109/CIG.2018.8490401

M3 - Conference contribution

SN - 978-1-5386-4360-0

BT - 2018 IEEE Conference on Computational Intelligence and Games (CIG)

PB - IEEE

ER -