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

Self-supervised learning (SSL), a paradigm harnessing unlabeled datasets totrain robust encoders, has recently witnessed substantial success. Theseencoders serve as pivotal feature extractors for downstream tasks, demandingsignificant computational resources. Nevertheless, recent studies have shedlight on vulnerabilities in pre-trained encoders, including backdoor andadversarial threats. Safeguarding the intellectual property of encoder trainersand ensuring the trustworthiness of deployed encoders pose notable challengesin SSL. To bridge these gaps, we introduce SSL-Auth, the first authenticationframework designed explicitly for pre-trained encoders. SSL-Auth leveragesselected key samples and employs a well-trained generative network toreconstruct watermark information, thus affirming the integrity of the encoderwithout compromising its performance. By comparing the reconstruction outcomesof the key samples, we can identify any malicious alterations. Comprehensiveevaluations conducted on a range of encoders and diverse downstream tasksdemonstrate the effectiveness of our proposed SSL-Auth.

SSL-Auth: An Authentication Framework by Fragile Watermarking for Pre-trained Encoders in Self-supervised Learning