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 (splitNN) has emerged as a popular strategy for addressing thehigh computational costs and low modeling efficiency in Vertical FederatedLearning (VFL). However, despite its popularity, vanilla splitNN lacksencryption protection, leaving it vulnerable to privacy leakage issues,especially Label Leakage from Gradients (LLG). Motivated by the LLG issueresulting from the use of labels during training, we propose the GenerativeAdversarial Federated Model (GAFM), a novel method designed specifically toenhance label privacy protection by integrating splitNN with GenerativeAdversarial Networks (GANs). GAFM leverages GANs to indirectly utilize labelinformation by learning the label distribution rather than relying on explicitlabels, thereby mitigating LLG. GAFM also employs an additional cross-entropyloss based on the noisy labels to further improve the prediction accuracy. Ourablation experiment demonstrates that the combination of GAN and thecross-entropy loss component is necessary to enable GAFM to mitigate LLGwithout significantly compromising the model utility. Empirical results onvarious datasets show that GAFM achieves a better and more robust trade-offbetween model utility and privacy compared to all baselines across multiplerandom runs. In addition, we provide experimental justification to substantiateGAFM's superiority over splitNN, demonstrating that it offers enhanced labelprotection through gradient perturbation relative to splitNN.

GAN-based Vertical Federated Learning for Label Protection in Binary Classification