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 is an emerging machine learning paradigm. Comparedto supervised learning which leverages high-quality labeled datasets,self-supervised learning relies on unlabeled datasets to pre-train powerfulencoders which can then be treated as feature extractors for various downstreamtasks. The huge amount of data and computational resources consumption makesthe encoders themselves become the valuable intellectual property of the modelowner. Recent research has shown that the machine learning model's copyright isthreatened by model stealing attacks, which aim to train a surrogate model tomimic the behavior of a given model. We empirically show that pre-trainedencoders are highly vulnerable to model stealing attacks. However, most of thecurrent efforts of copyright protection algorithms such as watermarkingconcentrate on classifiers. Meanwhile, the intrinsic challenges of pre-trainedencoder's copyright protection remain largely unstudied. We fill the gap byproposing SSLGuard, the first watermarking scheme for pre-trained encoders.Given a clean pre-trained encoder, SSLGuard injects a watermark into it andoutputs a watermarked version. The shadow training technique is also applied topreserve the watermark under potential model stealing attacks. Our extensiveevaluation shows that SSLGuard is effective in watermark injection andverification, and it is robust against model stealing and other watermarkremoval attacks such as input noising, output perturbing, overwriting, modelpruning, and fine-tuning.

SSLGuard: A Watermarking Scheme for Self-supervised Learning Pre-trained Encoders