Pith. sign in

REVIEW 4 cited by

Exploring the Landscape of Spatial Robustness

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1712.02779 v4 pith:RYN6V3NP submitted 2017-12-07 cs.LG cs.CVcs.NEstat.ML

Logan Engstrom , Brandon Tran , Dimitris Tsipras , Ludwig Schmidt , Aleksander Madry This is my paper · ORCID
classification cs.LGcs.CVcs.NEstat.ML
keywords robustnessperturbationsspatialadversarialfindgithubhttpsmadrylab
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

The study of adversarial robustness has so far largely focused on perturbations bound in p-norms. However, state-of-the-art models turn out to be also vulnerable to other, more natural classes of perturbations such as translations and rotations. In this work, we thoroughly investigate the vulnerability of neural network--based classifiers to rotations and translations. While data augmentation offers relatively small robustness, we use ideas from robust optimization and test-time input aggregation to significantly improve robustness. Finally we find that, in contrast to the p-norm case, first-order methods cannot reliably find worst-case perturbations. This highlights spatial robustness as a fundamentally different setting requiring additional study. Code available at https://github.com/MadryLab/adversarial_spatial and https://github.com/MadryLab/spatial-pytorch.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Testing Robustness Against Unforeseen Adversaries

    cs.LG 2019-08 conditional novelty 8.0 of 10

    ImageNet-UA, a six-attack benchmark with four new attacks, shows L-infinity robustness does not transfer to unforeseen distortions and that L2 training and AugMix generalize better.

  2. Fool the Stoplight: Realistic Adversarial Patch Attacks on Traffic Light Detectors

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A universal adversarial patch placed below a traffic light can flip red detections to green on YOLO models, with partial real-world success on a full autonomous driving perception pipeline.

  3. Are Adversarial Robustness and Common Perturbation Robustness Independent Attributes ?

    cs.LG 2019-09 reject novelty 6.0 of 10

    A new benchmark for common perturbation robustness is proposed, and experiments conclude that adversarial robustness and common perturbation robustness are independent, a claim that the provided data only weakly support.

  4. Robustifying deep networks for image segmentation

    eess.IV 2019-08 conditional novelty 4.0 of 10

    Adversarial perturbations based on gradient methods reduce brain tumor segmentation Dice scores by up to 65 percent, and defensive distillation outperforms adversarial training while all defenses still underperform on...

Pith tools