Houdini: Fooling Deep Structured Prediction Models

TL;DR

Houdini is a novel method for generating adversarial examples that directly target the true performance measure of structured prediction tasks, improving attack success across various applications.

Contribution

Houdini introduces a flexible, task-specific adversarial generation approach for structured prediction models, surpassing traditional surrogate-based methods.

Findings

Houdini achieves higher attack success rates.

Adversarial examples are less perceptible.

Effective across speech recognition, pose estimation, and segmentation.

Abstract

Generating adversarial examples is a critical step for evaluating and improving the robustness of learning machines. So far, most existing methods only work for classification and are not designed to alter the true performance measure of the problem at hand. We introduce a novel flexible approach named Houdini for generating adversarial examples specifically tailored for the final performance measure of the task considered, be it combinatorial and non-decomposable. We successfully apply Houdini to a range of applications such as speech recognition, pose estimation and semantic segmentation. In all cases, the attacks based on Houdini achieve higher success rate than those based on the traditional surrogates used to train the models while using a less perceptible adversarial perturbation.