General Video Game Evaluation Using Relative Algorithm Performance Profiles

Thorbjørn Nielsen; Gabriella Barros; Julian Togelius; Mark Nelson

doi:10.1007/978-3-319-16549-3_30

General Video Game Evaluation Using Relative Algorithm Performance Profiles

Thorbjørn Nielsen
, Gabriella Barros
, Julian Togelius
, Mark Nelson

Center for Digital Play

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

In order to generate complete games through evolution we need generic and reliably evaluation functions for games. It has been suggested that game quality could be characterised through playing a game with different controllers and comparing their performance. This paper explores that idea through investigating the relative performance of different general game-playing algorithms. Seven game-playing algorithms was used to play several hand-designed, mutated and randomly generated VGDL game descriptions. Results discussed appear to support the conjecture that well-designed games have, in average, a higher performance difference between better and worse game-playing algorithms.

Original language	English
Title of host publication	Applications of Evolutionary Computation : 18th European Conference, EvoApplications 2015, Copenhagen, Denmark, April 8-10, 2015, Proceedings
Number of pages	12
Publisher	Springer
Publication date	2015
Pages	369-380
ISBN (Print)	978-3-319-16548-6
ISBN (Electronic)	978-3-319-16549-3
DOIs	https://doi.org/10.1007/978-3-319-16549-3_30
Publication status	Published - 2015
Event	Evostar 2015: EvoGAMES - Bio-inspired Algorithms in Games - http://www.evostar.org/2015/cfp_evogames.php, Copenhagen, Denmark Duration: 8 Apr 2015 → 10 Apr 2015 http://www.evostar.org/2015/cfp_evogames.php

Conference

Conference	Evostar 2015
Location	http://www.evostar.org/2015/cfp_evogames.php
Country/Territory	Denmark
City	Copenhagen
Period	08/04/2015 → 10/04/2015
Internet address	http://www.evostar.org/2015/cfp_evogames.php

Series	Lecture Notes in Computer Science
Volume	9028
ISSN	0302-9743

Keywords

Game quality assessment
General game-playing algorithms
Evolutionary game generation
VGDL
Performance difference analysis

Access to Document

10.1007/978-3-319-16549-3_30

Nielsen2015GeneralAccepted author manuscript, 259 KB

Cite this

@inproceedings{24700217918946c8b7826dd02f4887e7,

title = "General Video Game Evaluation Using Relative Algorithm Performance Profiles",

abstract = "In order to generate complete games through evolution we need generic and reliably evaluation functions for games. It has been suggested that game quality could be characterised through playing a game with different controllers and comparing their performance. This paper explores that idea through investigating the relative performance of different general game-playing algorithms. Seven game-playing algorithms was used to play several hand-designed, mutated and randomly generated VGDL game descriptions. Results discussed appear to support the conjecture that well-designed games have, in average, a higher performance difference between better and worse game-playing algorithms.",

keywords = "Game quality assessment, General game-playing algorithms, Evolutionary game generation, VGDL, Performance difference analysis, Game quality assessment, General game-playing algorithms, Evolutionary game generation, VGDL, Performance difference analysis",

author = "Thorbj{\o}rn Nielsen and Gabriella Barros and Julian Togelius and Mark Nelson",

year = "2015",

doi = "10.1007/978-3-319-16549-3\_30",

language = "English",

isbn = "978-3-319-16548-6",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "369--380",

booktitle = "Applications of Evolutionary Computation",

address = "Germany",

note = "Evostar 2015 : EvoGAMES - Bio-inspired Algorithms in Games ; Conference date: 08-04-2015 Through 10-04-2015",

url = "http://www.evostar.org/2015/cfp\_evogames.php",

}

Nielsen, T, Barros, G, Togelius, J & Nelson, M 2015, General Video Game Evaluation Using Relative Algorithm Performance Profiles. in Applications of Evolutionary Computation: 18th European Conference, EvoApplications 2015, Copenhagen, Denmark, April 8-10, 2015, Proceedings. Springer, Lecture Notes in Computer Science, vol. 9028, pp. 369-380, Evostar 2015, Copenhagen, Denmark, 08/04/2015. https://doi.org/10.1007/978-3-319-16549-3_30

General Video Game Evaluation Using Relative Algorithm Performance Profiles. / Nielsen, Thorbjørn; Barros, Gabriella; Togelius, Julian et al.
Applications of Evolutionary Computation: 18th European Conference, EvoApplications 2015, Copenhagen, Denmark, April 8-10, 2015, Proceedings. Springer, 2015. p. 369-380 (Lecture Notes in Computer Science, Vol. 9028).

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - General Video Game Evaluation Using Relative Algorithm Performance Profiles

AU - Nielsen, Thorbjørn

AU - Barros, Gabriella

AU - Togelius, Julian

AU - Nelson, Mark

PY - 2015

Y1 - 2015

N2 - In order to generate complete games through evolution we need generic and reliably evaluation functions for games. It has been suggested that game quality could be characterised through playing a game with different controllers and comparing their performance. This paper explores that idea through investigating the relative performance of different general game-playing algorithms. Seven game-playing algorithms was used to play several hand-designed, mutated and randomly generated VGDL game descriptions. Results discussed appear to support the conjecture that well-designed games have, in average, a higher performance difference between better and worse game-playing algorithms.

AB - In order to generate complete games through evolution we need generic and reliably evaluation functions for games. It has been suggested that game quality could be characterised through playing a game with different controllers and comparing their performance. This paper explores that idea through investigating the relative performance of different general game-playing algorithms. Seven game-playing algorithms was used to play several hand-designed, mutated and randomly generated VGDL game descriptions. Results discussed appear to support the conjecture that well-designed games have, in average, a higher performance difference between better and worse game-playing algorithms.

KW - Game quality assessment

KW - General game-playing algorithms

KW - Evolutionary game generation

KW - VGDL

KW - Performance difference analysis

KW - Game quality assessment

KW - General game-playing algorithms

KW - Evolutionary game generation

KW - VGDL

KW - Performance difference analysis

U2 - 10.1007/978-3-319-16549-3_30

DO - 10.1007/978-3-319-16549-3_30

M3 - Article in proceedings

SN - 978-3-319-16548-6

T3 - Lecture Notes in Computer Science

SP - 369

EP - 380

BT - Applications of Evolutionary Computation

PB - Springer

T2 - Evostar 2015

Y2 - 8 April 2015 through 10 April 2015

ER -

Nielsen T, Barros G, Togelius J, Nelson M. General Video Game Evaluation Using Relative Algorithm Performance Profiles. In Applications of Evolutionary Computation: 18th European Conference, EvoApplications 2015, Copenhagen, Denmark, April 8-10, 2015, Proceedings. Springer. 2015. p. 369-380. (Lecture Notes in Computer Science, Vol. 9028). doi: 10.1007/978-3-319-16549-3_30

General Video Game Evaluation Using Relative Algorithm Performance Profiles

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this