Active Preference Learning with Discrete Choice Data

We propose an active learning algorithm that learns a continuous valuation model from discrete preferences. The algorithm automatically decides what items are best presented to an individual in order to find the item that they value highly in as few trials as possible, and exploits quirks of human psychology to minimize time and cognitive burden. To do this, our algorithm maximizes the expected improvement at each query without accurately modelling the entire valuation surface, which would be needlessly expensive. The problem is particularly difficult because the space of choices is infinite. We demonstrate the effectiveness of the new algorithm compared to related active learning methods. We also embed the algorithm within a decision making tool for assisting digital artists in rendering materials. The tool finds the best parameters while minimizing the number of queries.

[1]  Frédo Durand,et al.  Image-driven navigation of analytical BRDF models , 2006, EGSR '06.

[2]  Michael James Sasena,et al.  Flexibility and efficiency enhancements for constrained global design optimization with kriging approximations. , 2002 .

[3]  F. Mosteller.  Remarks on the method of paired comparisons: I. The least squares solution assuming equal standard deviations and equal correlations , 1951 .

[4]  Shane T. Jensen,et al.  Adaptive Paired Comparison Design , 2005 .

[5]  Aaron Hertzmann,et al.  Active learning for real-time motion controllers , 2007, SIGGRAPH 2007.

[6]  Daphne Koller,et al.  Support Vector Machine Active Learning with Application sto Text Classification , 2000, ICML.

[7]  Harold J. Kushner,et al.  A New Method of Locating the Maximum Point of an Arbitrary Multipeak Curve in the Presence of Noise , 1964 .

[8]  L. Thurstone.  A law of comparative judgment. , 1994 .

[9]  Daphne Koller,et al.  Support Vector Machine Active Learning with Applications to Text Classification , 2000, J. Mach. Learn. Res..

[10]  David A. Cohn,et al.  Active Learning with Statistical Models , 1996, NIPS.

[11]  A. Elo.  The rating of chessplayers, past and present , 1978 .

[12]  Edward H. Adelson,et al.  How do Humans Determine Reflectance Properties under Unknown Illumination , 2001, CVPR 2001.

[13]  C. D. Perttunen,et al.  Lipschitzian optimization without the Lipschitz constant , 1993 .

[14]  Hal S. Stern,et al.  A continuum of paired comparisons models , 1990 .

[15]  Andreas Krause,et al.  Near-optimal sensor placements in Gaussian processes , 2005, ICML.

[16]  C. F. Kossack,et al.  Rank Correlation Methods , 1949 .

[17]  Frédo Durand,et al.  Experimental analysis of BRDF models , 2005, EGSR '05.

[18]  M. Kendall,et al.  Rank Correlation Methods , 1949 .

[19]  Bernice W. Polemis.  Nonparametric Statistics for the Behavioral Sciences , 1959 .

[20]  Wei Chu,et al.  Extensions of Gaussian Processes for Ranking : Semi-supervised and Active Learning , 2005 .

[21]  Donald R. Jones,et al.  Efficient Global Optimization of Expensive Black-Box Functions , 1998, J. Glob. Optim..

[22]  F. Mosteller.  Remarks on the method of paired comparisons: I. The least squares solution assuming equal standard deviations and equal correlations , 1951 .

[23]  Klaus Obermayer,et al.  Gaussian Process Regression: Active Data Selection and Test Point Rejection , 2000, DAGM-Symposium.

[24]  Wei Chu,et al.  Preference learning with Gaussian processes , 2005, ICML.

[25]  Paul A. Beardsley,et al.  Design galleries: a general approach to setting parameters for computer graphics and animation , 1997, SIGGRAPH.

[26]  Peter Shirley,et al.  An Anisotropic Phong BRDF Model , 2000, J. Graphics, GPU, & Game Tools.