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

Currently, deep neural networks (DNNs) are widely adopted in differentapplications. Despite its commercial values, training a well-performed DNN isresource-consuming. Accordingly, the well-trained model is valuableintellectual property for its owner. However, recent studies revealed thethreats of model stealing, where the adversaries can obtain a function-similarcopy of the victim model, even when they can only query the model. In thispaper, we propose an effective and harmless model ownership verification (MOVE)to defend against different types of model stealing simultaneously, withoutintroducing new security risks. In general, we conduct the ownershipverification by verifying whether a suspicious model contains the knowledge ofdefender-specified external features. Specifically, we embed the externalfeatures by tempering a few training samples with style transfer. We then traina meta-classifier to determine whether a model is stolen from the victim. Thisapproach is inspired by the understanding that the stolen models should containthe knowledge of features learned by the victim model. In particular, wedevelop our MOVE method under both white-box and black-box settings to providecomprehensive model protection. Extensive experiments on benchmark datasetsverify the effectiveness of our method and its resistance to potential adaptiveattacks. The codes for reproducing the main experiments of our method areavailable at \url{https://github.com/THUYimingLi/MOVE}.

MOVE: Effective and Harmless Ownership Verification via Embedded External Features