Grant: Discovery Grants Program - Individual
University of Alberta
$61,000.00
General game-playing programs
EF 2008-09
Résumé
For over 25 years, I have been working on applying heuristic search and knowledge techniques towards building high-performance artificial intelligence (AI) systems. I often use game-playing and puzzle-solving programs as my experimental test domains. This has led to success in numerous domains (chess, checkers, poker, sliding tile puzzles, Rubik's Cube - all yielding valuable insights). However, each program is "one of a kind", with limited carryover from domain to domain. Although some of the algorithms apply to classes of applications, each high-performance AI system represents a complete re-implementation.
Humans are very good at transferring their experience in one domain (a game/puzzle in this case) to another. The state of the art in computing science is that the program designer transfers their experience from building a program for one domain to that of building a program for another domain. Ideally, the human should be factored out of this process.
There has been movement in the AI community to build general systems that can handle a broad class of applications. In this particular case, can we build a program capable of taking any game and, based solely on the rules and past experience, build a new game-playing program? Such a program would be a step towards the realization of "human level" AI.
This project will investigate the research issues around building a high-performance (high level of demonstrated skill) general game-playing program. The intent is to have a program that is a) capable of learning to play a broad spectrum of games and b) is scalable in its ability to learn (more learning time equates to better performance).
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Alberta, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
8173-2008
Name nom
Schaeffer, Jonathan
Department département
Computing Science
Organization id
9
Institution établissement
University of Alberta
Province en
Alberta
Province fr
Alberta
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2008
Competition year année de concours
2008
Award amount
61000
Program id
RGPIN
Program nane en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
331
Committee name en
Computing and Information Sciences - B
Committee name fr
Informatique - B
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2800
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Artificial intelligence (computer vision, use2603)
Research subject fr
Intelligence artificielle (vision artificielle, utiliser 2603)
Partie
1
Nb partie
5
Application title
General game-playing programs
Keyword
NA
Application summary
For over 25 years, I have been working on applying heuristic search and knowledge techniques towards building high-performance artificial intelligence (AI) systems. I often use game-playing and puzzle-solving programs as my experimental test domains. This has led to success in numerous domains (chess, checkers, poker, sliding tile puzzles, Rubik's Cube - all yielding valuable insights). However, each program is "one of a kind", with limited carryover from domain to domain. Although some of the algorithms apply to classes of applications, each high-performance AI system represents a complete re-implementation.
Humans are very good at transferring their experience in one domain (a game/puzzle in this case) to another. The state of the art in computing science is that the program designer transfers their experience from building a program for one domain to that of building a program for another domain. Ideally, the human should be factored out of this process.
There has been movement in the AI community to build general systems that can handle a broad class of applications. In this particular case, can we build a program capable of taking any game and, based solely on the rules and past experience, build a new game-playing program? Such a program would be a step towards the realization of "human level" AI.
This project will investigate the research issues around building a high-performance (high level of demonstrated skill) general game-playing program. The intent is to have a program that is a) capable of learning to play a broad spectrum of games and b) is scalable in its ability to learn (more learning time equates to better performance).