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

Cross-domain few-shot learning (CDFSL) remains a largely unsolved problem inthe area of computer vision, while self-supervised learning presents apromising solution. Both learning methods attempt to alleviate the dependencyof deep networks on the requirement of large-scale labeled data. Althoughself-supervised methods have recently advanced dramatically, their utility onCDFSL is relatively unexplored. In this paper, we investigate the role ofself-supervised representation learning in the context of CDFSL via a thoroughevaluation of existing methods. It comes as a surprise that even with shallowarchitectures or small training datasets, self-supervised methods can performfavorably compared to the existing SOTA methods. Nevertheless, no singleself-supervised approach dominates all datasets indicating that existingself-supervised methods are not universally applicable. In addition, we findthat representations extracted from self-supervised methods exhibit strongerrobustness than the supervised method. Intriguingly, whether self-supervisedrepresentations perform well on the source domain has little correlation withtheir applicability on the target domain. As part of our study, we conduct anobjective measurement of the performance for six kinds of representativeclassifiers. The results suggest Prototypical Classifier as the standardevaluation recipe for CDFSL.

How Well Do Self-Supervised Methods Perform in Cross-Domain Few-Shot Learning?