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.

References in corpus (26)

• Semantics derived automatically from language corpora contain human-like biases
• Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks
• Towards Deep Learning Models Resistant to Adversarial Attacks
• Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
• Image Quality Assessment: Unifying Structure and Texture Similarity
• The Unreasonable Effectiveness of Deep Features as a Perceptual Metric
• Adversarial Attacks and Defences: A Survey
• Generating Natural Adversarial Examples
• Countering Adversarial Images using Input Transformations
• Generalizing to Unseen Domains via Adversarial Data Augmentation
• Robust Wasserstein Profile Inference and Applications to Machine Learning
• A study of the effect of JPG compression on adversarial images
• Exploring the Landscape of Spatial Robustness
• No Classification without Representation: Assessing Geodiversity Issues in Open Data Sets for the Developing World
• Distributionally Robust Logistic Regression
• Predictive Inequity in Object Detection
• Simple Black-box Adversarial Attacks
• Noise or Signal: The Role of Image Backgrounds in Object Recognition
• Wasserstein Adversarial Examples via Projected Sinkhorn Iterations
• Adversarial Distributional Training for Robust Deep Learning
• Sample Out-Of-Sample Inference Based on Wasserstein Distance
• Image Quality Assessment for Perceptual Image Restoration: A New Dataset, Benchmark and Metric
• Optimal Transport Based Distributionally Robust Optimization: Structural Properties and Iterative Schemes
• Towards Feature Space Adversarial Attack
• An Unsupervised Information-Theoretic Perceptual Quality Metric
• Sparse and Imperceptible Adversarial Attack via a Homotopy Algorithm