David Stutz, Matthias Hein, Bernt Schiele.
Disentangling Adversarial Robustness and Generalization.
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019.
[ArXiv | BibTeX | Project Page]
David Stutz, Andreas Geiger.
Learning 3D Shape Completion from Laser Scan Data with Weak Supervision.
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018.
[PDF | BibTeX | Project Page]
David Stutz, Alexander Hermans, Bastian Leibe.
Superpixels: an evaluation of the state-of-the-art.
Computer Vision and Image Understanding, Volume 166, 2018.
[DOI | ArXiv | PDF | BibTeX | Project Page]
Our paper on confidence-calibrated adversarial training was accepted at ICML’20. In the revised paper, the proposed confidence-calibrated adversarial training tackles the problem of obtaining robustness that generalizes to attacks not seen during training. This is achieved by biasing the network towards low-confidence predictions on adversarial examples and rejecting these low-confidence examples at test time. This article gives a short abstract and includes paper and code.
Deep neural network (DNN) accelerators are specialized hardware for inference and have received considerable attention in the past years. Here, in order to reduce energy consumption, these accelerators are often operated at low voltage which causes the included accelerator memory to become unreliable. Additionally, recent work demonstrated attacks targeting individual bits in memory. The induced bit errors in both cases can cause significantly reduced accuracy of DNNs. In this paper, we tackle both random (due to low-voltage) and adversarial bit errors in DNNs. By explicitly taking such errors into account during training, wecan improve robustness significantly.
While robustness against imperceptible adversarial examples is well-studied, robustness against visible adversarial perturbations such as adversarial patches is poorly understood. In this pre-print, we present a practical approach to obtain adversarial patches while actively optimizing their location within the image. On Cifar10 and GTSRB, we show that adversarial training on these location-optimized adversarial patches improves robustness significantly while not reducing accuracy.
Adversarial training yields robust models against a specific threat model. However, robustness does not generalize to larger perturbations or threat models not seen during training. Confidence-calibrated adversarial training tackles this problem by biasing the network towards low-confidence predictions on adversarial examples. Through rejecting low-confidence (adversarial) examples, robustness generalizes to various threat models, including L2, L1 and L0 while training only on L∞ adversarial examples. This article gives a short abstract, discusses relevant updates to the previous version and includes paper and appendix.
Adversarial training is the de-facto standard to obtain models robust against adversarial examples. However, on complex datasets, a significant loss in accuracy is incurred and the robustness does not generalize to attacks not used during training. This paper introduces confidence-calibrated adversarial training. By forcing the confidence on adversarial examples to decay with their distance to the training data, the loss in accuracy is reduced and robustness generalizes to other attacks and larger perturbations.
Our paper on adversarial robustness and generalization was accepted at CVPR’19. In the revised paper, we show that adversarial examples usually leave the manifold, including a brief theoretical argumentation. Similarly, adversarial examples can be found on the manifold; then, robustness is nothing else than generalization. For (off-manifold) adversarial examples, in contrast, we show that generalization and robustness are not necessarily contradicting objectives. As example, on synthetic data, we adversarially train a robust and accurate model. This article gives a short abstract and provides the paper including appendix.