New Generalization Bounds for Learning Kernels - Computer Science > Artificial IntelligenceReport as inadecuate




New Generalization Bounds for Learning Kernels - Computer Science > Artificial Intelligence - Download this document for free, or read online. Document in PDF available to download.

Abstract: This paper presents several novel generalization bounds for the problem oflearning kernels based on the analysis of the Rademacher complexity of thecorresponding hypothesis sets. Our bound for learning kernels with a convexcombination of p base kernels has only a logp dependency on the number ofkernels, p, which is considerably more favorable than the previous best boundgiven for the same problem. We also give a novel bound for learning with alinear combination of p base kernels with an L 2 regularization whosedependency on p is only in p^{1-4}.



Author: Corinna Cortes, Mehryar Mohri, Afshin Rostamizadeh

Source: https://arxiv.org/



DOWNLOAD PDF




Related documents