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/

Zendy Plus

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

Zendy Tools

Zendy Open

Due to the pervasiveness of image capturing devices in every-day life, imagesof individuals are routinely captured. Although this has enabled many benefits,it also infringes on personal privacy. A promising direction in research onobfuscation of facial images has been the work in the k-same family of methodswhich employ the concept of k-anonymity from database privacy. However, thereare a number of deficiencies of k-anonymity that carry over to the k-samemethods, detracting from their usefulness in practice. In this paper, we firstoutline several of these deficiencies and discuss their implications in thecontext of facial obfuscation. We then develop a framework through which weobtain a formal differentially private guarantee for the obfuscation of facialimages in generative machine learning models. Our approach provides a provableprivacy guarantee that is not susceptible to the outlined deficiencies ofk-same obfuscation and produces photo-realistic obfuscated output. In addition,we demonstrate through experimental comparisons that our approach can achievecomparable utility to k-same obfuscation in terms of preservation of usefulfeatures in the images. Furthermore, we propose a method to achievedifferential privacy for any image (i.e., without restriction to facial images)through the direct modification of pixel intensities. Although the addition ofnoise to pixel intensities does not provide the high visual quality obtainedvia generative machine learning models, it offers greater versatility byeliminating the need for a trained model. We demonstrate that our proposed useof the exponential mechanism in this context is able to provide superior visualquality to pixel-space obfuscation using the Laplace mechanism.

Obfuscation of Images via Differential Privacy: From Facial Images to General Images