LGAICVROOct 17, 2022

Learning Less Generalizable Patterns with an Asymmetrically Trained Double Classifier for Better Test-Time Adaptation

arXiv:2210.09834v14 citationsh-index: 3
Originality Incremental advance
AI Analysis

This addresses the generalization issue in deep learning for domain adaptation, offering a novel training-time modification that enhances test-time adaptation, though it is incremental in combining existing ideas with new loss mechanisms.

The paper tackles the problem of deep neural networks failing to generalize outside their training distribution due to shortcut learning, by proposing a novel approach using a pair of classifiers with a shortcut patterns avoidance loss. The result is improved state-of-the-art performance on PACS and Office-Home benchmarks, with concrete gains in test-time adaptation.

Deep neural networks often fail to generalize outside of their training distribution, in particular when only a single data domain is available during training. While test-time adaptation has yielded encouraging results in this setting, we argue that, to reach further improvements, these approaches should be combined with training procedure modifications aiming to learn a more diverse set of patterns. Indeed, test-time adaptation methods usually have to rely on a limited representation because of the shortcut learning phenomenon: only a subset of the available predictive patterns is learned with standard training. In this paper, we first show that the combined use of existing training-time strategies, and test-time batch normalization, a simple adaptation method, does not always improve upon the test-time adaptation alone on the PACS benchmark. Furthermore, experiments on Office-Home show that very few training-time methods improve upon standard training, with or without test-time batch normalization. We therefore propose a novel approach using a pair of classifiers and a shortcut patterns avoidance loss that mitigates the shortcut learning behavior by reducing the generalization ability of the secondary classifier, using the additional shortcut patterns avoidance loss that encourages the learning of samples specific patterns. The primary classifier is trained normally, resulting in the learning of both the natural and the more complex, less generalizable, features. Our experiments show that our method improves upon the state-of-the-art results on both benchmarks and benefits the most to test-time batch normalization.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes