Addressing the Loss-Metric Mismatch with Adaptive Loss Alignment

TL;DR

This paper introduces a reinforcement learning-based method to adaptively align loss functions with evaluation metrics during training, improving performance across various tasks and metrics.

Contribution

It presents a novel, sample-efficient reinforcement learning approach for dynamic loss adaptation that enhances metric optimization in machine learning models.

Findings

Improves performance by directly optimizing evaluation metrics.

Enhances smoothing of the loss landscape during training.

Demonstrates transferability of learned policies across tasks.

Abstract

In most machine learning training paradigms a fixed, often handcrafted, loss function is assumed to be a good proxy for an underlying evaluation metric. In this work we assess this assumption by meta-learning an adaptive loss function to directly optimize the evaluation metric. We propose a sample efficient reinforcement learning approach for adapting the loss dynamically during training. We empirically show how this formulation improves performance by simultaneously optimizing the evaluation metric and smoothing the loss landscape. We verify our method in metric learning and classification scenarios, showing considerable improvements over the state-of-the-art on a diverse set of tasks. Importantly, our method is applicable to a wide range of loss functions and evaluation metrics. Furthermore, the learned policies are transferable across tasks and data, demonstrating the versatility of the method.