MiniFool: Physics-Constraint-Aware Minimizer-Based Adversarial Attacks in Deep Neural Networks

The European Physical Journal C, vol. 86

Abstract

Abstract In this paper, we present a new algorithm, , that implements physics-inspired adversarial attacks for testing neural network-based classification tasks in particle and astroparticle physics. While we initially developed the algorithm for the search for astrophysical tau neutrinos with the IceCube Neutrino Observatory, we apply it to further data from other science domains, thus demonstrating its general applicability. Here, we apply the algorithm to the well-known MNIST data set and furthermore, to public data from the CMS Experiment at the Large Hadron Collider. The algorithm is based on minimizing a cost function that combines a $$\chi ^2$$ χ 2 based test-statistic with the deviation from the desired target score. The test statistic quantifies the probability of the perturbations applied to the data based on the experimental uncertainties. For our studied use cases, we find that the likelihood of a flipped classification differs for both the initially correctly and incorrectly classified events. When testing changes of the classifications as a function of an attack parameter that scales the experimental uncertainties, the robustness of the network decision can be assessed. Furthermore, this allows testing the robustness of the classification of unlabeled experimental data.

Authors 10

1. Lucie Flek

   University of Bonn · Lamarr Institute for Machine Learning and Artificial Intelligence

   Affiliation as printed

   Lamarr Institute for Machine Learning and Artificial Intelligence

   University of Bonn

2. Oliver Janik Aachen

   Friedrich-Alexander-Universität Erlangen-Nürnberg · RWTH Aachen University

   Affiliation as printed

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   RWTH Aachen University

3. Akbar Karimi

   University of Bonn · Lamarr Institute for Machine Learning and Artificial Intelligence

   Affiliation as printed

   Lamarr Institute for Machine Learning and Artificial Intelligence

   University of Bonn

4. Timo Saala

   University of Bonn

   Affiliation as printed

   University of Bonn

5. Alexander G. Schmidt Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

6. M. Schott

   University of Bonn

   Affiliation as printed

   University of Bonn

7. Philipp Soldin Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

8. Matthias Thiesmeyer Aachen

   University of Wisconsin System · RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

   University of Wisconsin

9. Christopher H. Wiebusch corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

10. Ulrich Willemsen Aachen

    RWTH Aachen University

    Affiliation as printed

    RWTH Aachen University