Contrastive Test-Time Adaptation
This addresses the challenge of domain adaptation in machine learning for scenarios where source data is unavailable, representing a novel method rather than an incremental improvement.
The paper tackles the problem of test-time adaptation, where a model must adapt to a new domain without access to source data, by proposing AdaContrast, which uses self-supervised contrastive learning and an online pseudo labeling scheme with refinement. It achieves state-of-the-art performance on major benchmarks, offering memory efficiency, insensitivity to hyper-parameters, and better model calibration.
Test-time adaptation is a special setting of unsupervised domain adaptation where a trained model on the source domain has to adapt to the target domain without accessing source data. We propose a novel way to leverage self-supervised contrastive learning to facilitate target feature learning, along with an online pseudo labeling scheme with refinement that significantly denoises pseudo labels. The contrastive learning task is applied jointly with pseudo labeling, contrasting positive and negative pairs constructed similarly as MoCo but with source-initialized encoder, and excluding same-class negative pairs indicated by pseudo labels. Meanwhile, we produce pseudo labels online and refine them via soft voting among their nearest neighbors in the target feature space, enabled by maintaining a memory queue. Our method, AdaContrast, achieves state-of-the-art performance on major benchmarks while having several desirable properties compared to existing works, including memory efficiency, insensitivity to hyper-parameters, and better model calibration. Project page: sites.google.com/view/adacontrast.