Gaussian Process Preference Elicitation

Edwin Bonilla, Shengbo Guo, Scott Sanner

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


    Bayesian approaches to preference elicitation (PE) are particularly attractive due to their ability to explicitly model uncertainty in users� latent utility functions. However, previous approaches to Bayesian PE have ignored the important problem of generalizing from previous users to an unseen user in order to reduce the elicitation burden on new users. In this paper, we address this deficiency by introducing a Gaussian Process (GP) prior over users� latent utility functions on the joint space of user and item features. We learn the hyper-parameters of this GP on a set of preferences of previous users and use it to aid in the elicitation process for a new user. This approach provides a flexible model of a multi-user utility function, facilitates an efficient value of information (VOI) heuristic query selection strategy, and provides a principled way to incorporate the elicitations of multiple users back into the model. We show the effectiveness of our method in comparison to previous work on a real dataset of user preferences over sushi types.
    Original languageEnglish
    Title of host publicationAdvances in Neural Information Processing Systems 24
    EditorsJohn Lafferty, Christopher Williams, John Shawe-Taylor, Richard S. Zemel and Aro
    Place of PublicationCambridge, MA
    PublisherMIT Press
    EditionPeer Reviewed
    ISBN (Print)9781617823800
    Publication statusPublished - 2010
    EventConference on Advances in Neural Information Processing Systems (NIPS 2010) - Vancouver Canada
    Duration: 1 Jan 2010 → …


    ConferenceConference on Advances in Neural Information Processing Systems (NIPS 2010)
    Period1/01/10 → …
    OtherDecember 6-9 2010
    Internet address


    Dive into the research topics of 'Gaussian Process Preference Elicitation'. Together they form a unique fingerprint.

    Cite this