Grant: Discovery Grants Program - Individual
University of Alberta
$28,000.00
Data-driven statistical agent modelling
FY 2007-08
Summary
Decision making in multiagent environments is a critical challenge for deploying artificially intelligent agents in real-world settings. Applications involving or requiring interaction between independent agents -- human, robotic, or software -- are rapidly increasing. The rise of inexpensive robot platforms and the pervasiveness of distributed agents make the assumption of isolated decision making less justifiable. Agent modelling is a key component for coping with the presence of other decision makers. Applications as diverse as assistive technologies, autonomous driving, electronic commerce, and interactive entertainment all require or could benefit significantly from accurate models of other agents (artificial or human).One approach to constructing such agent models is to consult domain experts and build generic, knowledge-intensive models of the typical human or artificial agent. The use of these entirely generic models, though, prevents tailoring the interaction to the particular participants. In addition, many tasks require models that are not easily elicited from a domain expert due to domain complexity and the difficulty in codifying expert intuitions.This research examines a data-driven, statistical approach to agent modelling. In particular, it focuses on the development of techniques for constructing statistical agent models from large datasets of previous interactions, both of the agent being modelled and past agents. In the same way that large corpuses of written text transformed the practice of natural language processing, it is expected that a similar statistical approach can overcome the significant limitations of knowledge-based agent models.
Program
Discovery Grants Program - Individual
Recipient
Location
Alberta, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
298239-2007
Name nom
Bowling, Michael
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
2007
Competition year année de concours
2007
Award amount
28000
Program id
RGPIN
Program name 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
1207
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Information, computer and communication technologies
Area of application fr
Technologie de l'information, des ordinateurs et des réseaux de communications
Research subject code
2805
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Learning and inference theories
Research subject fr
Théories de l'apprentissage et l'inférence
Partie
1
Nb partie
5
Application title
Data-driven statistical agent modelling
Keyword
NA
Application summary
Decision making in multiagent environments is a critical challenge for deploying artificially intelligent agents in real-world settings. Applications involving or requiring interaction between independent agents -- human, robotic, or software -- are rapidly increasing. The rise of inexpensive robot platforms and the pervasiveness of distributed agents make the assumption of isolated decision making less justifiable. Agent modelling is a key component for coping with the presence of other decision makers. Applications as diverse as assistive technologies, autonomous driving, electronic commerce, and interactive entertainment all require or could benefit significantly from accurate models of other agents (artificial or human).One approach to constructing such agent models is to consult domain experts and build generic, knowledge-intensive models of the typical human or artificial agent. The use of these entirely generic models, though, prevents tailoring the interaction to the particular participants. In addition, many tasks require models that are not easily elicited from a domain expert due to domain complexity and the difficulty in codifying expert intuitions.This research examines a data-driven, statistical approach to agent modelling. In particular, it focuses on the development of techniques for constructing statistical agent models from large datasets of previous interactions, both of the agent being modelled and past agents. In the same way that large corpuses of written text transformed the practice of natural language processing, it is expected that a similar statistical approach can overcome the significant limitations of knowledge-based agent models.