ICML 2020poster22 citations

Projective Preferential Bayesian Optimization

Petrus Mikkola, Milica Todorović, Jari Järvi, Patrick Rinke, Samuel Kaski

Abstract

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.

BibTeX
@InProceedings{pmlr-v119-mikkola20a,
  title = 	 {Projective Preferential {B}ayesian Optimization},
  author =       {Mikkola, Petrus and Todorovi{\'c}, Milica and J{\"a}rvi, Jari and Rinke, Patrick and Kaski, Samuel},
  booktitle = 	 {Proceedings of the 37th International Conference on Machine Learning},
  pages = 	 {6884--6892},
  year = 	 {2020},
  editor = 	 {III, Hal Daumé and Singh, Aarti},
  volume = 	 {119},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {13--18 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v119/mikkola20a/mikkola20a.pdf},
  url = 	 {https://proceedings.mlr.press/v119/mikkola20a.html},
  abstract = 	 {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.}
}
Projective Preferential Bayesian Optimization · ICML 2020