Test-Time Adaptation to Distribution Shift by Confidence Maximization and Input Transformation
arXiv:2106.14999
Abstract
Deep neural networks often exhibit poor performance on data that is unlikely under the train-time data distribution, for instance data affected by corruptions. Previous works demonstrate that test-time adaptation to data shift, for instance using entropy minimization, effectively improves performance on such shifted distributions. This paper focuses on the fully test-time adaptation setting, where only unlabeled data from the target distribution is required. This allows adapting arbitrary pretrained networks. Specifically, we propose a novel loss that improves test-time adaptation by addressing both premature convergence and instability of entropy minimization. This is achieved by replacing the entropy by a non-saturating surrogate and adding a diversity regularizer based on batch-wise entropy maximization that prevents convergence to trivial collapsed solutions. Moreover, we propose to prepend an input transformation module to the network that can partially undo test-time distribution shifts. Surprisingly, this preprocessing can be learned solely using the fully test-time adaptation loss in an end-to-end fashion without any target domain labels or source domain data. We show that our approach outperforms previous work in improving the robustness of publicly available pretrained image classifiers to common corruptions on such challenging benchmarks as ImageNet-C.
arXiv comment16 pages, 5 figures, 7 tables
References in corpus (14)
- ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation
- ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
- VisDA: The Visual Domain Adaptation Challenge
- AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty
- Model Adaptation: Unsupervised Domain Adaptation without Source Data
- Revisiting Batch Normalization For Practical Domain Adaptation
- Tent: Fully Test-time Adaptation by Entropy Minimization
- Unsupervised Domain Adaptation through Self-Supervision
- Adversarial Examples Are a Natural Consequence of Test Error in Noise
- Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Coming
- Evaluating Prediction-Time Batch Normalization for Robustness under Covariate Shift
- Improving Robustness Without Sacrificing Accuracy with Patch Gaussian Augmentation
- Entropy Minimization vs. Diversity Maximization for Domain Adaptation
- On Interaction Between Augmentations and Corruptions in Natural Corruption Robustness