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

Federated learning (FL) is becoming a key component in many technology-basedapplications including language modeling -- where individual FL participantsoften have privacy-sensitive text data in their local datasets. However,realizing the extent of privacy leakage in federated language models is notstraightforward and the existing attacks only intend to extract data regardlessof how sensitive or naive it is. To fill this gap, in this paper, we introducetwo novel findings with regard to leaking privacy-sensitive user data fromfederated language models. Firstly, we make a key observation that modelsnapshots from the intermediate rounds in FL can cause greater privacy leakagethan the final trained model. Secondly, we identify that privacy leakage can beaggravated by tampering with a model's selective weights that are specificallyresponsible for memorizing the sensitive training data. We show how a maliciousclient can leak the privacy-sensitive data of some other user in FL evenwithout any cooperation from the server. Our best-performing method improvesthe membership inference recall by 29% and achieves up to 70% private datareconstruction, evidently outperforming existing attacks with strongerassumptions of adversary capabilities.

FLTrojan: Privacy Leakage Attacks against Federated Language Models Through Selective Weight Tampering