Tent: Fully Test-time Adaptation by Entropy Minimization
This addresses the challenge of model adaptation during testing for applications like image classification and semantic segmentation, representing a novel method rather than an incremental improvement.
The paper tackles the problem of fully test-time adaptation, where a model must adapt to new data using only its parameters and test data, by proposing test entropy minimization (tent) to optimize model confidence. The result is a new state-of-the-art error on ImageNet-C and improved generalization on various benchmarks like CIFAR-10/100 and domain adaptation tasks, achieved in one epoch without altering training.
A model must adapt itself to generalize to new and different data during testing. In this setting of fully test-time adaptation the model has only the test data and its own parameters. We propose to adapt by test entropy minimization (tent): we optimize the model for confidence as measured by the entropy of its predictions. Our method estimates normalization statistics and optimizes channel-wise affine transformations to update online on each batch. Tent reduces generalization error for image classification on corrupted ImageNet and CIFAR-10/100 and reaches a new state-of-the-art error on ImageNet-C. Tent handles source-free domain adaptation on digit recognition from SVHN to MNIST/MNIST-M/USPS, on semantic segmentation from GTA to Cityscapes, and on the VisDA-C benchmark. These results are achieved in one epoch of test-time optimization without altering training.