No Pairs Left Behind: Improving Metric Learning with Regularized Triplet Objective

TL;DR

This paper introduces NPLB, a regularized triplet loss method that enhances metric learning performance without extra sampling costs, validated on benchmark and healthcare datasets for improved embeddings and health risk prediction.

Contribution

The paper presents a novel regularization approach for triplet loss that improves metric learning efficiency and effectiveness, especially in complex real-world healthcare data.

Findings

NPLB outperforms traditional triplet objectives on benchmark datasets.

Embeddings learned with NPLB significantly improve downstream health prediction tasks.

NPLB demonstrates potential for complex applications in healthcare and biological data analysis.

Abstract

We propose a novel formulation of the triplet objective function that improves metric learning without additional sample mining or overhead costs. Our approach aims to explicitly regularize the distance between the positive and negative samples in a triplet with respect to the anchor-negative distance. As an initial validation, we show that our method (called No Pairs Left Behind [NPLB]) improves upon the traditional and current state-of-the-art triplet objective formulations on standard benchmark datasets. To show the effectiveness and potentials of NPLB on real-world complex data, we evaluate our approach on a large-scale healthcare dataset (UK Biobank), demonstrating that the embeddings learned by our model significantly outperform all other current representations on tested downstream tasks. Additionally, we provide a new model-agnostic single-time health risk definition that, when used in tandem with the learned representations, achieves the most accurate prediction of subjects' future health complications. Our results indicate that NPLB is a simple, yet effective framework for improving existing deep metric learning models, showcasing the potential implications of metric learning in more complex applications, especially in the biological and healthcare domains.