Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

University of Toronto

$66,000.00

Robust preference assessment and online mechanism design

FY 2009-10

Summary
The assessment of user preferences is critical to the development of decision support and interactive optimization tools The ability to customize decisions to the needs or desires of specific users or organizations is of great interest in a wide range of domains, including medical decision making, financial planning, procurement and supply chain management, product recommendation, personalized Web search, advertizing, user interface configuration, and many others. Making good or optimal decisions requires that a system know enough about a user's preferences (or utility function) to address the task at hand, and have the ability to trade off the costs of obtaining such preference information with the value of that information (i.e., its ability to improve decision quality). The proposed research will develop new techniques for preference assessment and elicitation, both in single-agent and multiagent (or economic) contexts. The underlying methodology embodies assessment techniques that allow decisions to be made with only partial preference information, and elicitation techniques that obtain the most relevant information. Building on previous work using Bayesian and regret-based optimization and elicitation, we will explore two key themes that challenge current techniques. One is the assessment of preferences in sequential decision problems (such as Markov decision problems). Indirect (observational) and direct (querying) partial assessment methods will be developed, as will online methods that must act in the presence of an uncertain future (this is of special importance to online mechanism design). The second theme is the development of "open-ended" assessment and elicitation techniques, wherein the attributes or outcomes of interest are ill-specified or entirely unknown to the system. Feature identification/construction methods for utility functions will be developed as will techniques for handling vague features in multiattribute utility models.
Program
Discovery Grants Program - Individual
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
121843-2008
Name nom
Boutilier, Craig
Department département
Computer Science
Organization id
31
Institution établissement
University of Toronto
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2009
Competition year année de concours
2008
Award amount
66000
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
2
Nb partie
5
Application title
Robust preference assessment and online mechanism design
Keyword
NA
Application summary
The assessment of user preferences is critical to the development of decision support and interactive optimization tools The ability to customize decisions to the needs or desires of specific users or organizations is of great interest in a wide range of domains, including medical decision making, financial planning, procurement and supply chain management, product recommendation, personalized Web search, advertizing, user interface configuration, and many others. Making good or optimal decisions requires that a system know enough about a user's preferences (or utility function) to address the task at hand, and have the ability to trade off the costs of obtaining such preference information with the value of that information (i.e., its ability to improve decision quality). The proposed research will develop new techniques for preference assessment and elicitation, both in single-agent and multiagent (or economic) contexts. The underlying methodology embodies assessment techniques that allow decisions to be made with only partial preference information, and elicitation techniques that obtain the most relevant information. Building on previous work using Bayesian and regret-based optimization and elicitation, we will explore two key themes that challenge current techniques. One is the assessment of preferences in sequential decision problems (such as Markov decision problems). Indirect (observational) and direct (querying) partial assessment methods will be developed, as will online methods that must act in the presence of an uncertain future (this is of special importance to online mechanism design). The second theme is the development of "open-ended" assessment and elicitation techniques, wherein the attributes or outcomes of interest are ill-specified or entirely unknown to the system. Feature identification/construction methods for utility functions will be developed as will techniques for handling vague features in multiattribute utility models.