Multimodal-Aware Weakly Supervised Metric Learning with Self-weighting Triplet Loss

TL;DR

This paper introduces MDaML, a novel weakly supervised metric learning method that partitions data into clusters, assigns local centers and weights, and optimizes on the SPD manifold to improve local separability and handle multimodal distributions.

Contribution

The paper proposes MDaML, a new metric learning algorithm that incorporates local clustering, weighting, and Riemannian optimization to address multimodal data distributions in weak supervision.

Findings

MDaML outperforms existing methods on 13 datasets.

It effectively handles multimodal class distributions.

The approach improves local separability and metric validity.

Abstract

In recent years, we have witnessed a surge of interests in learning a suitable distance metric from weakly supervised data. Most existing methods aim to pull all the similar samples closer while push the dissimilar ones as far as possible. However, when some classes of the dataset exhibit multimodal distribution, these goals conflict and thus can hardly be concurrently satisfied. Additionally, to ensure a valid metric, many methods require a repeated eigenvalue decomposition process, which is expensive and numerically unstable. Therefore, how to learn an appropriate distance metric from weakly supervised data remains an open but challenging problem. To address this issue, in this paper, we propose a novel weakly supervised metric learning algorithm, named MultimoDal Aware weakly supervised Metric Learning (MDaML). MDaML partitions the data space into several clusters and allocates the local cluster centers and weight for each sample. Then, combining it with the weighted triplet loss can further enhance the local separability, which encourages the local dissimilar samples to keep a large distance from the local similar samples. Meanwhile, MDaML casts the metric learning problem into an unconstrained optimization on the SPD manifold, which can be efficiently solved by Riemannian Conjugate Gradient Descent (RCGD). Extensive experiments conducted on 13 datasets validate the superiority of the proposed MDaML.