INFORMATIVE ENERGY METRIC FOR SIMILARITY MEASURE IN REPRODUCING KERNEL HILBERT SPACES

Abstract

In this paper, information energy metric (IEM) is obtained by similarity computing for high-dimensional samples in a reproducing kernel Hilbert space (RKHS). Firstly, similar/dissimilar subsets and their corresponding informative energy functions are defined. Secondly, IEM is proposed for similarity measure of those subsets, which converts the non-metric distances into metric ones. Finally, applications of this metric is introduced, such as classification problems. Experimental results validate the effectiveness of the proposed method.

Keywords

• Kernel methods
• Similarity measure
• Reproducing kernel Hilbert space
• Non-metric distance