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

We propose an experimental method for measuring bias in face recognitionsystems. Existing methods to measure bias depend on benchmark datasets that arecollected in the wild and annotated for protected (e.g., race, gender) andnon-protected (e.g., pose, lighting) attributes. Such observational datasetsonly permit correlational conclusions, e.g., "Algorithm A's accuracy isdifferent on female and male faces in dataset X.". By contrast, experimentalmethods manipulate attributes individually and thus permit causal conclusions,e.g., "Algorithm A's accuracy is affected by gender and skin color."  Our method is based on generating synthetic faces using a neural facegenerator, where each attribute of interest is modified independently whileleaving all other attributes constant. Human observers crucially provide theground truth on perceptual identity similarity between synthetic image pairs.We validate our method quantitatively by evaluating race and gender biases ofthree research-grade face recognition models. Our synthetic pipeline revealsthat for these algorithms, accuracy is lower for Black and East Asianpopulation subgroups. Our method can also quantify how perceptual changes inattributes affect face identity distances reported by these models. Our largesynthetic dataset, consisting of 48,000 synthetic face image pairs (10,200unique synthetic faces) and 555,000 human annotations (individual attributesand pairwise identity comparisons) is available to researchers in thisimportant area.

Benchmarking Algorithmic Bias in Face Recognition: An Experimental Approach Using Synthetic Faces and Human Evaluation