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

Recent works have demonstrated that it is possible to reconstruct trainingimages and their labels from gradients of an image-classification model whenits architecture is known. Unfortunately, there is still an incompletetheoretical understanding of the efficacy and failure of these gradient-leakageattacks. In this paper, we propose a novel framework to analyse training-dataleakage from gradients that draws insights from both analytic andoptimisation-based gradient-leakage attacks. We formulate the reconstructionproblem as solving a linear system from each layer iteratively, accompanied bycorrections using gradient matching. Under this framework, we claim that thesolubility of the reconstruction problem is primarily determined by that of thelinear system at each layer. As a result, we are able to partially attributethe leakage of the training data in a deep network to its architecture. We alsopropose a metric to measure the level of security of a deep learning modelagainst gradient-based attacks on the training data.

Analysing Training-Data Leakage from Gradients through Linear Systems and Gradient Matching