Tomar, DevavratVray, Guillaume Marc GeorgesBozorgtabar, BehzadThiran, Jean-Philippe2023-08-022023-08-022023-08-022023-01-0110.1109/CVPR52729.2023.01948https://infoscience.epfl.ch/handle/20.500.14299/199603WOS:001062531304064Most recent test-time adaptation methods focus on only classification tasks, use specialized network architectures, destroy model calibration or rely on lightweight information from the source domain. To tackle these issues, this paper proposes a novel Test-time Self-Learning method with auto- matic Adversarial augmentation dubbed TeSLA for adapt- ing a pre-trained source model to the unlabeled streaming test data. In contrast to conventional self-learning meth- ods based on cross-entropy, we introduce a new test-time loss function through an implicitly tight connection with the mutual information and online knowledge distillation. Furthermore, we propose a learnable efficient adversarial augmentation module that further enhances online knowl- edge distillation by simulating high entropy augmented im- ages. Our method achieves state-of-the-art classification and segmentation results on several benchmarks and types of domain shifts, particularly on challenging measurement shifts of medical images. TeSLA also benefits from several desirable properties compared to competing methods in terms of calibration, uncertainty metrics, insensitivity to model architectures, and source training strategies, all supported by extensive ablations.text::conference output::conference proceedings::conference paper