English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Test-time adaptation (TTA) seeks to tackle potential distribution shiftsbetween training and testing data by adapting a given model w.r.t. any testingsample. This task is particularly important for deep models when the testenvironment changes frequently. Although some recent attempts have been made tohandle this task, we still face two practical challenges: 1) existing methodshave to perform backward computation for each test sample, resulting inunbearable prediction cost to many applications; 2) while existing TTAsolutions can significantly improve the test performance on out-of-distributiondata, they often suffer from severe performance degradation on in-distributiondata after TTA (known as catastrophic forgetting). In this paper, we point outthat not all the test samples contribute equally to model adaptation, andhigh-entropy ones may lead to noisy gradients that could disrupt the model.Motivated by this, we propose an active sample selection criterion to identifyreliable and non-redundant samples, on which the model is updated to minimizethe entropy loss for test-time adaptation. Furthermore, to alleviate theforgetting issue, we introduce a Fisher regularizer to constrain importantmodel parameters from drastic changes, where the Fisher importance is estimatedfrom test samples with generated pseudo labels. Extensive experiments onCIFAR-10-C, ImageNet-C, and ImageNet-R verify the effectiveness of our proposedmethod.

Efficient Test-Time Model Adaptation without Forgetting