Addressing the Loss-Metric Mismatch with Adaptive Loss Alignment

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.

Accepted to ICML 2019

Cited by in corpus (5)

• SoftAdapt: Techniques for Adaptive Loss Weighting of Neural Networks with Multi-Part Loss Functions
• Learning Adaptive Loss for Robust Learning with Noisy Labels
• Optimizing Black-box Metrics with Adaptive Surrogates
• A new approach to forecast service parts demand by integrating user preferences into multi-objective optimization
• Training Over-parameterized Models with Non-decomposable Objectives