Connecting the Dots: Detecting Adversarial Perturbations Using Context Inconsistency

TL;DR

This paper introduces a context-based detection system for adversarial perturbations in deep neural networks, leveraging auto-encoders trained to identify violations of scene consistency, significantly improving detection accuracy over previous methods.

Contribution

The novel approach uses class-specific auto-encoders to detect adversarial attacks by identifying context violations, outperforming existing context-agnostic detection methods.

Findings

Achieves ROC-AUC over 0.95 in detecting adversarial attacks

Over 20% improvement in detection accuracy compared to prior methods

Effective on PASCAL VOC and MS COCO datasets

Abstract

There has been a recent surge in research on adversarial perturbations that defeat Deep Neural Networks (DNNs) in machine vision; most of these perturbation-based attacks target object classifiers. Inspired by the observation that humans are able to recognize objects that appear out of place in a scene or along with other unlikely objects, we augment the DNN with a system that learns context consistency rules during training and checks for the violations of the same during testing. Our approach builds a set of auto-encoders, one for each object class, appropriately trained so as to output a discrepancy between the input and output if an added adversarial perturbation violates context consistency rules. Experiments on PASCAL VOC and MS COCO show that our method effectively detects various adversarial attacks and achieves high ROC-AUC (over 0.95 in most cases); this corresponds to over 20% improvement over a state-of-the-art context-agnostic method.