When Neural Networks Fail to Generalize? A Model Sensitivity Perspective
DOI:
https://doi.org/10.1609/aaai.v37i9.26328Keywords:
ML: Transfer, Domain Adaptation, Multi-Task Learning, CV: Adversarial Attacks & Robustness, CV: Representation Learning for Vision, ML: Adversarial Learning & Robustness, ML: Classification and Regression, ML: Representation LearningAbstract
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 by up to 2.55%. The source code is available at https://github.com/DIAL-RPI/Spectral-Adversarial-Data-Augmentation.Downloads
Published
2023-06-26
How to Cite
Zhang, J., Chao, H., Dhurandhar, A., Chen, P.-Y., Tajer, A., Xu, Y., & Yan, P. (2023). When Neural Networks Fail to Generalize? A Model Sensitivity Perspective. Proceedings of the AAAI Conference on Artificial Intelligence, 37(9), 11219-11227. https://doi.org/10.1609/aaai.v37i9.26328
Issue
Section
AAAI Technical Track on Machine Learning IV