In this paper we study pairwise preference learning in a music setting with multitask Gaussian processes and examine the effect of sparsity in the input space as well as in the actual judgments. To introduce sparsity in the inputs, we extend a classic pairwise likelihood model to support sparse, multi-task Gaussian process priors based on the pseudo-input formulation. Sparsity in the actual pairwise judgments is potentially obtained by a sequential experimental design approach, and we discuss the combination of the sequential approach with the pseudo-input preference model. A preliminary simulation shows the performance on a real-world music preference dataset which motivates and demonstrates the potential of the sparse Gaussian process formulation for pairwise likelihoods.
|Published - 2011
|Annual Conference on Neural Information Processing Systems : CMPL workshop - Granada, Spain
Duration: 7 Mar 2012 → 8 Mar 2012
Conference number: 25
|Annual Conference on Neural Information Processing Systems : CMPL workshop
|07/03/2012 → 08/03/2012