When Neural Networks Fail to Generalize? A Model Sensitivity Perspective

TL;DR

This paper investigates the failure of neural networks to generalize in single domain scenarios by analyzing model sensitivity and proposes a spectral adversarial data augmentation method to improve generalization performance.

Contribution

It introduces the concept of model sensitivity related to generalization and proposes SADA, a novel data augmentation technique targeting sensitive frequencies.

Findings

SADA improves model generalization in single domain settings.

Models trained with SADA outperform state-of-the-art methods.

Sensitivity analysis correlates strongly with generalization failure.

Abstract

Domain generalization (DG) aims to train a model to perform well in unseen domains under different distributions. This paper considers a more realistic yet more challenging scenario,namely Single Domain Generalization (Single-DG), where only a single source domain is available for training. To tackle this challenge, we first try to understand when neural networks fail to generalize? We empirically ascertain a property of a model that correlates strongly with its generalization that we coin as "model sensitivity". Based on our analysis, we propose a novel strategy of Spectral Adversarial Data Augmentation (SADA) to generate augmented images targeted at the highly sensitive frequencies. Models trained with these hard-to-learn samples can effectively suppress the sensitivity in the frequency space, which leads to improved generalization performance. Extensive experiments on multiple public datasets demonstrate the superiority of our approach, which surpasses the state-of-the-art single-DG methods.