Timezone: »
Reliable evaluation of adversarial defenses is a challenging task, currently limited to an expert who manually crafts attacks that exploit the defense’s inner workings, or to approaches based on ensemble of fixed attacks, none of which may be effective for the specific defense at hand. Our key observation is that adaptive attacks are composed from a set of reusable building blocks that can be formalized in a search space and used to automatically discover attacks for unknown defenses. We evaluated our approach on 24 adversarial defenses and show that it outperforms AutoAttack, the current state-of-the-art tool for reliable evaluation of adversarial defenses: our tool discovered significantly stronger attacks by producing 3.0%-50.8% additional adversarial examples for 10 models, while obtaining attacks with slightly stronger or similar strength for the remaining models.
Author Information
Chengyuan Yao (Swiss Federal Institute of Technology)
Pavol Bielik (ETH Zurich)
Petar Tsankov (LatticeFlow)
Martin Vechev (ETH Zurich, Switzerland)
More from the Same Authors
-
2021 : Bayesian Framework for Gradient Leakage »
Mislav Balunovic · Dimitar Dimitrov · Martin Vechev -
2022 : Efficient Robustness Verification of Neural Ordinary Differential Equations »
Mustafa Zeqiri · Mark Müller · Marc Fischer · Martin Vechev -
2022 : Generating Intuitive Fairness Specifications for Natural Language Processing »
Florian E. Dorner · Momchil Peychev · Nikola Konstantinov · Naman Goel · Elliott Ash · Martin Vechev -
2022 : Just Avoid Robust Inaccuracy: Boosting Robustness Without Sacrificing Accuracy »
Yannick Merkli · Pavol Bielik · Petar Tsankov · Martin Vechev -
2022 : Certified Training: Small Boxes are All You Need »
Mark Müller · Franziska Eckert · Marc Fischer · Martin Vechev -
2022 : FARE: Provably Fair Representation Learning »
Nikola Jovanović · Mislav Balunovic · Dimitar Dimitrov · Martin Vechev -
2022 Poster: Learning to Configure Computer Networks with Neural Algorithmic Reasoning »
Luca Beurer-Kellner · Martin Vechev · Laurent Vanbever · Petar Veličković -
2022 Poster: (De-)Randomized Smoothing for Decision Stump Ensembles »
Miklós Horváth · Mark Müller · Marc Fischer · Martin Vechev -
2022 Poster: LAMP: Extracting Text from Gradients with Language Model Priors »
Mislav Balunovic · Dimitar Dimitrov · Nikola Jovanović · Martin Vechev -
2020 Poster: Learning Certified Individually Fair Representations »
Anian Ruoss · Mislav Balunovic · Marc Fischer · Martin Vechev -
2020 Poster: Certified Defense to Image Transformations via Randomized Smoothing »
Marc Fischer · Maximilian Baader · Martin Vechev -
2019 Poster: Beyond the Single Neuron Convex Barrier for Neural Network Certification »
Gagandeep Singh · Rupanshu Ganvir · Markus Püschel · Martin Vechev -
2019 Poster: Certifying Geometric Robustness of Neural Networks »
Mislav Balunovic · Maximilian Baader · Gagandeep Singh · Timon Gehr · Martin Vechev -
2018 Poster: Learning to Solve SMT Formulas »
Mislav Balunovic · Pavol Bielik · Martin Vechev -
2018 Oral: Learning to Solve SMT Formulas »
Mislav Balunovic · Pavol Bielik · Martin Vechev -
2018 Poster: Fast and Effective Robustness Certification »
Gagandeep Singh · Timon Gehr · Matthew Mirman · Markus Püschel · Martin Vechev