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

Split Learning (SL) has emerged as a practical and efficient alternative totraditional federated learning. While previous attempts to attack SL have oftenrelied on overly strong assumptions or targeted easily exploitable models, weseek to develop more practical attacks. We introduce SDAR, a novel attackframework against SL with an honest-but-curious server. SDAR leveragesauxiliary data and adversarial regularization to learn a decodable simulator ofthe client's private model, which can effectively infer the client's privatefeatures under the vanilla SL, and both features and labels under the U-shapedSL. We perform extensive experiments in both configurations to validate theeffectiveness of our proposed attacks. Notably, in challenging but practicalscenarios where existing passive attacks struggle to reconstruct the client'sprivate data effectively, SDAR consistently achieves attack performancecomparable to active attacks. On CIFAR-10, at the deep split level of 7, SDARachieves private feature reconstruction with less than 0.025 mean squared errorin both the vanilla and the U-shaped SL, and attains a label inference accuracyof over 98% in the U-shaped setting, while existing attacks fail to producenon-trivial results.

Passive Inference Attacks on Split Learning via Adversarial Regularization