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

Modern face recognition systems leverage datasets containing images ofhundreds of thousands of specific individuals' faces to train deepconvolutional neural networks to learn an embedding space that maps anarbitrary individual's face to a vector representation of their identity. Theperformance of a face recognition system in face verification (1:1) and faceidentification (1:N) tasks is directly related to the ability of an embeddingspace to discriminate between identities. Recently, there has been significantpublic scrutiny into the source and privacy implications of large-scale facerecognition training datasets such as MS-Celeb-1M and MegaFace, as many peopleare uncomfortable with their face being used to train dual-use technologiesthat can enable mass surveillance. However, the impact of an individual'sinclusion in training data on a derived system's ability to recognize them hasnot previously been studied. In this work, we audit ArcFace, astate-of-the-art, open source face recognition system, in a large-scale faceidentification experiment with more than one million distractor images. We finda Rank-1 face identification accuracy of 79.71% for individuals present in themodel's training data and an accuracy of 75.73% for those not present. Thismodest difference in accuracy demonstrates that face recognition systems usingdeep learning work better for individuals they are trained on, which hasserious privacy implications when one considers all major open source facerecognition training datasets do not obtain informed consent from individualsduring their collection.

Investigating the Impact of Inclusion in Face Recognition Training Data on Individual Face Identification