Self-Taught Metric Learning without Labels

TL;DR

This paper introduces a self-taught, unsupervised metric learning framework that predicts class relations to learn embeddings, outperforming existing methods and enhancing semi-supervised learning.

Contribution

The paper proposes a novel end-to-end self-taught algorithm for unsupervised metric learning that does not require external pseudo-labeling modules.

Findings

Outperforms existing unsupervised metric learning methods on standard benchmarks.

Sometimes surpasses supervised models with the same backbone.

Achieves state-of-the-art results in semi-supervised metric learning.

Abstract

We present a novel self-taught framework for unsupervised metric learning, which alternates between predicting class-equivalence relations between data through a moving average of an embedding model and learning the model with the predicted relations as pseudo labels. At the heart of our framework lies an algorithm that investigates contexts of data on the embedding space to predict their class-equivalence relations as pseudo labels. The algorithm enables efficient end-to-end training since it demands no off-the-shelf module for pseudo labeling. Also, the class-equivalence relations provide rich supervisory signals for learning an embedding space. On standard benchmarks for metric learning, it clearly outperforms existing unsupervised learning methods and sometimes even beats supervised learning models using the same backbone network. It is also applied to semi-supervised metric learning as a way of exploiting additional unlabeled data, and achieves the state of the art by boosting performance of supervised learning substantially.