Timezone: »

 
Poster
Projective Preferential Bayesian Optimization
Petrus Mikkola · Milica Todorović · Jari Järvi · Patrick Rinke · Samuel Kaski

Thu Jul 16 01:00 PM -- 01:45 PM & Fri Jul 17 02:00 AM -- 02:45 AM (PDT) @
in Poster Session 46 »

Bayesian optimization is an effective method for finding extrema of a black-box function. We propose a new type of Bayesian optimization for learning user preferences in high-dimensional spaces. The central assumption is that the underlying objective function cannot be evaluated directly, but instead a minimizer along a projection can be queried, which we call a projective preferential query. The form of the query allows for feedback that is natural for a human to give, and which enables interaction. This is demonstrated in a user experiment in which the user feedback comes in the form of optimal position and orientation of a molecule adsorbing to a surface. We demonstrate that our framework is able to find a global minimum of a high-dimensional black-box function, which is an infeasible task for existing preferential Bayesian optimization frameworks that are based on pairwise comparisons.

Keywords: Bayesian Methods · Gaussian Processes · Ranking and Preference Learning

Author Information

Petrus Mikkola (Aalto University)
Milica Todorović (Aalto University)
Jari Järvi (Aalto University)
Patrick Rinke (Aalto University)
Samuel Kaski (Aalto University and University of Manchester)

More from the Same Authors