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

Protecting personal data against exploitation of machine learning models iscrucial. Recently, availability attacks have shown great promise to provide anextra layer of protection against the unauthorized use of data to train neuralnetworks. These methods aim to add imperceptible noise to clean data so thatthe neural networks cannot extract meaningful patterns from the protected data,claiming that they can make personal data "unexploitable." This paper providesa strong countermeasure against such approaches, showing that unexploitabledata might only be an illusion. In particular, we leverage the power ofdiffusion models and show that a carefully designed denoising process cancounteract the effectiveness of the data-protecting perturbations. Werigorously analyze our algorithm, and theoretically prove that the amount ofrequired denoising is directly related to the magnitude of the data-protectingperturbations. Our approach, called AVATAR, delivers state-of-the-artperformance against a suite of recent availability attacks in variousscenarios, outperforming adversarial training even under distribution mismatchbetween the diffusion model and the protected data. Our findings call for moreresearch into making personal data unexploitable, showing that this goal is farfrom over. Our implementation is available at this repository:https://github.com/hmdolatabadi/AVATAR.

The Devil's Advocate: Shattering the Illusion of Unexploitable Data using Diffusion Models