MT3: Meta Test-Time Training for Self-Supervised Test-Time Adaption
This addresses the challenge of domain adaptation in deep learning for improved robustness in image classification, though it appears incremental as it builds on existing test-time training and meta-learning concepts.
The paper tackles the problem of neural networks adapting to domain shifts during test-time by proposing Meta Test-Time Training (MT3), which combines meta-learning and self-supervision to enable adaptation with a single unlabeled image, achieving significant improvements on the CIFAR-10-Corrupted benchmark.
An unresolved problem in Deep Learning is the ability of neural networks to cope with domain shifts during test-time, imposed by commonly fixing network parameters after training. Our proposed method Meta Test-Time Training (MT3), however, breaks this paradigm and enables adaption at test-time. We combine meta-learning, self-supervision and test-time training to learn to adapt to unseen test distributions. By minimizing the self-supervised loss, we learn task-specific model parameters for different tasks. A meta-model is optimized such that its adaption to the different task-specific models leads to higher performance on those tasks. During test-time a single unlabeled image is sufficient to adapt the meta-model parameters. This is achieved by minimizing only the self-supervised loss component resulting in a better prediction for that image. Our approach significantly improves the state-of-the-art results on the CIFAR-10-Corrupted image classification benchmark. Our implementation is available on GitHub.