REVIEW 6 cited by
Are adversarial examples inevitable?
Not yet reviewed by Pith; the record is open.
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
Signed reviews
read the original abstract
A wide range of defenses have been proposed to harden neural networks against adversarial attacks. However, a pattern has emerged in which the majority of adversarial defenses are quickly broken by new attacks. Given the lack of success at generating robust defenses, we are led to ask a fundamental question: Are adversarial attacks inevitable? This paper analyzes adversarial examples from a theoretical perspective, and identifies fundamental bounds on the susceptibility of a classifier to adversarial attacks. We show that, for certain classes of problems, adversarial examples are inescapable. Using experiments, we explore the implications of theoretical guarantees for real-world problems and discuss how factors such as dimensionality and image complexity limit a classifier's robustness against adversarial examples.
Forward citations
Cited by 6 Pith papers
-
Existence of Adversarial Examples for Random Convolutional Networks via Isoperimetric Inequalities on $\mathbb{so}(d)$
Using isoperimetric inequalities on SO(d), this paper proves that random convolutional networks with odd activations or ReLU have sign-flipping adversarial examples at distance O(||x0||/sqrt(d)).
-
Robust learning of halfspaces under log-concave marginals
An efficient agnostic learner for halfspaces over subgaussian isotropic log-concave distributions returns a classifier with error opt+O(ε) and boundary volume O(r+ε) in time d^{~O(1/ε^2)}.
-
Adversarial Examples Are Not Bugs, They Are Superposition
The paper argues that adversarial examples arise from superposition, and shows that changing superposition changes robustness and vice versa in toy models and ResNet18.
-
Adversarial Transferability in Deep Denoising Models: Theoretical Insights and Robustness Enhancement via Out-of-Distribution Typical Set Sampling
Adversarial transferability in denoisers is attributed to Gaussian noise concentrating on a typical set, and a low-density Gaussian sampling defense (TS) is shown to improve robustness while preserving denoising accuracy.
-
Unsafe at any AUC: Unlearned Lessons from Sociotechnical Disasters for Responsible AI
AI safety is a systems-governance problem: six recurring organizational failure patterns from past disasters remain unlearned in AI development, so component-level fixes like benchmarks and alignment cannot deliver safety.
-
Robustness Analysis against Adversarial Patch Attacks in Fully Unmanned Stores
Adversarial patches can disrupt YOLOv5 and Faster R-CNN detectors in a simulated unmanned store, with a new histogram loss yielding small success-rate gains.
Discussion (0). Continue with ORCID to comment.