Relational Surrogate Loss Learning

TL;DR

This paper introduces a novel rank correlation-based method for learning surrogate losses that better approximate evaluation metrics, leading to improved performance across multiple machine learning tasks.

Contribution

It proposes a new surrogate loss learning approach focusing on model relation preservation, simplifying optimization and enhancing efficiency and accuracy.

Findings

Outperforms state-of-the-art methods on human pose estimation

Achieves significant improvements in image classification and machine translation

Demonstrates effectiveness across diverse tasks like reading comprehension

Abstract

Evaluation metrics in machine learning are often hardly taken as loss functions, as they could be non-differentiable and non-decomposable, e.g., average precision and F1 score. This paper aims to address this problem by revisiting the surrogate loss learning, where a deep neural network is employed to approximate the evaluation metrics. Instead of pursuing an exact recovery of the evaluation metric through a deep neural network, we are reminded of the purpose of the existence of these evaluation metrics, which is to distinguish whether one model is better or worse than another. In this paper, we show that directly maintaining the relation of models between surrogate losses and metrics suffices, and propose a rank correlation-based optimization method to maximize this relation and learn surrogate losses. Compared to previous works, our method is much easier to optimize and enjoys significant efficiency and performance gains. Extensive experiments show that our method achieves improvements on various tasks including image classification and neural machine translation, and even outperforms state-of-the-art methods on human pose estimation and machine reading comprehension tasks. Code is available at: https://github.com/hunto/ReLoss.