Two-Stage Metric Learning

In this paper, we present a novel two-stage metric learning algorithm. We first map each learning instance to a probability distribution by comput-ing its similarities to a set of fixed anchor points. Then, we define the distance in the input data space as the Fisher information distance on the associated statistical manifold. This induces in the input data space a new family of distance metric with unique properties. Unlike kernelized metric learning, we do not require the similarity measure to be positive semi-definite. Moreover, it can also be interpreted as a local metric learn-ing algorithm with well defined distance approx-imation. We evaluate its performance on a num-ber of datasets. It outperforms significantly other metric learning methods and SVM. 1