Human Imperceptible Attacks and Applications to Improve Fairness

Abstract

Modern neural networks are able to perform at least as well as humans in numerous tasks involving object classification and image generation. However, small perturbations which are imperceptible to humans may significantly degrade the performance of well-trained deep neural networks. We provide a Distributionally Robust Optimization (DRO) framework which integrates human-based image quality assessment methods to design optimal attacks that are imperceptible to humans but significantly damaging to deep neural networks. Through extensive experiments, we show that our attack algorithm generates better-quality (less perceptible to humans) attacks than other state-of-the-art human imperceptible attack methods. Moreover, we demonstrate that DRO training using our optimally designed human imperceptible attacks can improve group fairness in image classification. Towards the end, we provide an algorithmic implementation to speed up DRO training significantly, which could be of independent interest.

How Biased is Your Feature?: Computing Fairness Influence Functions with Global Sensitivity Analysis

“…Particularly, we define and compute Fairness Influence Function (FIF) that quantifies the contribution of individual and subset of features to the resulting bias. FIFs do not only allow practitioners to identify the features to act up on but also to quantify the effect of various affirmative [8,19,23,[45][46][47][48] or punitive actions [21,32,42] on the resulting bias.…”