Unadversarial Examples: Designing Objects for Robust Vision

Hadi Salman, Andrew Ilyas, Logan Engstrom, Sai Vemprala, Aleksander Madry, Ashish Kapoor

Keywords: adversarial robustness, adversarial examples, computer vision

TL;DR: We introduce unadversarial examples; objects explicitly designed to be robustly classified.

Abstract: We study a class of computer vision settings wherein one can modify the design of the objects being recognized. We develop a framework that leverages this capability---and deep networks' unusual sensitivity to input perturbations---to design ``robust objects,'' i.e., objects that are explicitly optimized to be confidently classified. Our framework yields improved performance on standard benchmarks, a simulated robotics environment, and physical-world experiments.

Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.

Code: https://github.com/microsoft/unadversarial