EvoMCTS: Enhancing MCTS-based players through genetic programming

Amit Benbassat; Moshe Sipper

doi:10.1109/CIG.2013.6633631

EvoMCTS: Enhancing MCTS-based players through genetic programming

Amit Benbassat
, Moshe Sipper

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

15 Scopus citations

Abstract

We present EvoMCTS, a genetic programming method for enhancing level of play in games. Our work focuses on the zero-sum, deterministic, perfect-information board game of Reversi. Expanding on our previous work on evolving board-state evaluation functions for alpha-beta search algorithm variants, we now evolve evaluation functions that augment the MTCS algorithm. We use strongly typed genetic programming, explicitly defined introns, and a selective directional crossover method. Our system regularly evolves players that outperform MCTS players that use the same amount of search. Our results prove scalable and EvoMCTS players whose search is increased offline still outperform MCTS counterparts. To demonstrate the generality of our method we apply EvoMCTS successfully to the game of Dodgem.

Original language	English
Title of host publication	2013 IEEE Conference on Computational Intelligence in Games, CIG 2013
DOIs	https://doi.org/10.1109/CIG.2013.6633631
State	Published - 1 Dec 2013
Event	2013 IEEE Conference on Computational Intelligence in Games, CIG 2013 - Niagara Falls, ON, Canada Duration: 11 Aug 2013 → 13 Aug 2013

Publication series

Name	IEEE Conference on Computatonal Intelligence and Games, CIG
ISSN (Print)	2325-4270
ISSN (Electronic)	2325-4289

Conference

Conference	2013 IEEE Conference on Computational Intelligence in Games, CIG 2013
Country/Territory	Canada
City	Niagara Falls, ON
Period	11/08/13 → 13/08/13

ASJC Scopus subject areas

Artificial Intelligence
Computer Graphics and Computer-Aided Design
Computer Vision and Pattern Recognition
Human-Computer Interaction
Software

Access to Document

10.1109/CIG.2013.6633631

Cite this

@inproceedings{719b3462f38a4f24a1567776e5ac734b,

title = "EvoMCTS: Enhancing MCTS-based players through genetic programming",

abstract = "We present EvoMCTS, a genetic programming method for enhancing level of play in games. Our work focuses on the zero-sum, deterministic, perfect-information board game of Reversi. Expanding on our previous work on evolving board-state evaluation functions for alpha-beta search algorithm variants, we now evolve evaluation functions that augment the MTCS algorithm. We use strongly typed genetic programming, explicitly defined introns, and a selective directional crossover method. Our system regularly evolves players that outperform MCTS players that use the same amount of search. Our results prove scalable and EvoMCTS players whose search is increased offline still outperform MCTS counterparts. To demonstrate the generality of our method we apply EvoMCTS successfully to the game of Dodgem.",

author = "Amit Benbassat and Moshe Sipper",

year = "2013",

month = dec,

day = "1",

doi = "10.1109/CIG.2013.6633631",

language = "English",

isbn = "9781467353113",

series = "IEEE Conference on Computatonal Intelligence and Games, CIG",

booktitle = "2013 IEEE Conference on Computational Intelligence in Games, CIG 2013",

note = "2013 IEEE Conference on Computational Intelligence in Games, CIG 2013 ; Conference date: 11-08-2013 Through 13-08-2013",

}

Benbassat, A & Sipper, M 2013, EvoMCTS: Enhancing MCTS-based players through genetic programming. in 2013 IEEE Conference on Computational Intelligence in Games, CIG 2013., 6633631, IEEE Conference on Computatonal Intelligence and Games, CIG, 2013 IEEE Conference on Computational Intelligence in Games, CIG 2013, Niagara Falls, ON, Canada, 11/08/13. https://doi.org/10.1109/CIG.2013.6633631

TY - GEN

T1 - EvoMCTS

T2 - 2013 IEEE Conference on Computational Intelligence in Games, CIG 2013

AU - Benbassat, Amit

AU - Sipper, Moshe

PY - 2013/12/1

Y1 - 2013/12/1

N2 - We present EvoMCTS, a genetic programming method for enhancing level of play in games. Our work focuses on the zero-sum, deterministic, perfect-information board game of Reversi. Expanding on our previous work on evolving board-state evaluation functions for alpha-beta search algorithm variants, we now evolve evaluation functions that augment the MTCS algorithm. We use strongly typed genetic programming, explicitly defined introns, and a selective directional crossover method. Our system regularly evolves players that outperform MCTS players that use the same amount of search. Our results prove scalable and EvoMCTS players whose search is increased offline still outperform MCTS counterparts. To demonstrate the generality of our method we apply EvoMCTS successfully to the game of Dodgem.

AB - We present EvoMCTS, a genetic programming method for enhancing level of play in games. Our work focuses on the zero-sum, deterministic, perfect-information board game of Reversi. Expanding on our previous work on evolving board-state evaluation functions for alpha-beta search algorithm variants, we now evolve evaluation functions that augment the MTCS algorithm. We use strongly typed genetic programming, explicitly defined introns, and a selective directional crossover method. Our system regularly evolves players that outperform MCTS players that use the same amount of search. Our results prove scalable and EvoMCTS players whose search is increased offline still outperform MCTS counterparts. To demonstrate the generality of our method we apply EvoMCTS successfully to the game of Dodgem.

UR - https://www.scopus.com/pages/publications/84892427149

U2 - 10.1109/CIG.2013.6633631

DO - 10.1109/CIG.2013.6633631

M3 - Conference contribution

AN - SCOPUS:84892427149

SN - 9781467353113

T3 - IEEE Conference on Computatonal Intelligence and Games, CIG

BT - 2013 IEEE Conference on Computational Intelligence in Games, CIG 2013

Y2 - 11 August 2013 through 13 August 2013

ER -

EvoMCTS: Enhancing MCTS-based players through genetic programming

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this