Abstract: Systems and methods for diagnosing and testing fairness of machine learning models based on detecting individual violations of group definitions of fairness, via adversarial attacks that aim to perturb model inputs to generate individual violations. The systems and methods employ auxiliary machine learning models using a local surrogate for identifying group membership and assess fairness by measuring the transferability of attacks from this model. The systems and methods generate fairness indicator values indicative of discrimination risk due to the target predictions generated by the machine learning model, by comparing gradients of the machine learning model to gradients of an auxiliary machine learning model.

Abstract: This disclosure relates to methods of transforming an image. Disclosed herein is a method for manipulating an image using at least one image control handle. The image comprises pixels, and at least one set of constrained pixels defines a constrained region having a transformation constraint. The method comprises transforming pixels of the image based on input received from the manipulation of the at least one image control handle. The transformation constraint applies to the pixels inside the constrained region, and the degree to which the transformation constraint applies to any respective pixel outside the constrained region is conditional on the distance between the respective pixel and the constrained region.