Bayesian Active Distance Metric Learning

TL;DR

This paper introduces a Bayesian framework for distance metric learning that estimates a posterior distribution, improving reliability with small datasets and enabling active learning to select the most informative pairs.

Contribution

It proposes a Bayesian approach with a variational algorithm for distance metric learning and applies active learning to enhance efficiency and accuracy.

Findings

Higher classification accuracy compared to non-Bayesian methods

More informative training examples identified through active learning

Effective estimation of posterior distribution for the distance metric

Abstract

Distance metric learning is an important component for many tasks, such as statistical classification and content-based image retrieval. Existing approaches for learning distance metrics from pairwise constraints typically suffer from two major problems. First, most algorithms only offer point estimation of the distance metric and can therefore be unreliable when the number of training examples is small. Second, since these algorithms generally select their training examples at random, they can be inefficient if labeling effort is limited. This paper presents a Bayesian framework for distance metric learning that estimates a posterior distribution for the distance metric from labeled pairwise constraints. We describe an efficient algorithm based on the variational method for the proposed Bayesian approach. Furthermore, we apply the proposed Bayesian framework to active distance metric learning by selecting those unlabeled example pairs with the greatest uncertainty in relative distance. Experiments in classification demonstrate that the proposed framework achieves higher classification accuracy and identifies more informative training examples than the non-Bayesian approach and state-of-the-art distance metric learning algorithms.