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) has increasingly been deployed, in its vertical form,among organizations to facilitate secure collaborative training over siloeddata. In vertical FL (VFL), participants hold disjoint features of the same setof sample instances. Among them, only one has labels. This participant, knownas the active party, initiates the training and interacts with the otherparticipants, known as the passive parties. Despite the increasing adoption ofVFL, it remains largely unknown if and how the active party can extract featuredata from the passive party, especially when training deep neural network (DNN)models.  This paper makes the first attempt to study the feature security problem ofDNN training in VFL. We consider a DNN model partitioned between active andpassive parties, where the latter only holds a subset of the input layer andexhibits some categorical features of binary values. Using a reduction from theExact Cover problem, we prove that reconstructing those binary features isNP-hard. Through analysis, we demonstrate that, unless the feature dimension isexceedingly large, it remains feasible, both theoretically and practically, tolaunch a reconstruction attack with an efficient search-based algorithm thatprevails over current feature protection techniques. To address this problem,we develop a novel feature protection scheme against the reconstruction attackthat effectively misleads the search to some pre-specified random values. Withan extensive set of experiments, we show that our protection scheme sustainsthe feature reconstruction attack in various VFL applications at no expense ofaccuracy loss.

Feature Reconstruction Attacks and Countermeasures of DNN training in Vertical Federated Learning