HackerPapers

arXiv:1802.08195cs.LG

Adversarial Examples that Fool both Computer Vision and Time-Limited Humans

Gamaleldin F. Elsayed, Shreya Shankar, Brian Cheung, Nicolas Papernot, Alex Kurakin, Ian Goodfellow, Jascha Sohl-Dickstein

Abstract

Machine learning models are vulnerable to adversarial examples: small changes to images can cause computer vision models to make mistakes such as identifying a school bus as an ostrich. However, it is still an open question whether humans are prone to similar mistakes. Here, we address this question by leveraging recent techniques that transfer adversarial examples from computer vision models with known parameters and architecture to other models with unknown parameters and architecture, and by matching the initial processing of the human visual system. We find that adversarial examples that strongly transfer across computer vision models influence the classifications made by time-limited human observers.

arXiv comments
Journal reference
Advances in Neural Information Processing Systems, 2018
DOI
Subjects
cs.LGMachine Learning cs.CVComputer Vision and Pattern Recognition q-bio.NCNeurons and Cognition stat.MLMachine Learning (Statistics)

On Hacker News

Linked by 4 identifiable commenters in 5 comments across 4 threads, from Mar 2018 to Nov 2022. Found in 4 HN submissions from Feb 2018 to Nov 2018. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments4 threads

  1. Study urges caution when comparing neural networks to the brain 1 comment
  2. ●●|●●●●●|●●|● 1 comment

    • I find these types of things to be a lot more interesting as well. With respect the behavioral modifiers I feel like they really help explain how societies actually function, once you start paying attention you see just how effective and pervasive they really are. This study[0]…

      katmannthreeView on HNParent context

  3. Artificial Intelligence Hits the Barrier of Meaning 2 comments

    • Recent update — we do know, and humans are vulnerable to perturbations created to fool a multitude of existing AI: https://arxiv.org/pdf/1802.08195.pdf

      ben_wView on HNParent context

    • https://arxiv.org/pdf/1802.08195.pdf Looks like we are starting to find examples. I think your intuition is wrong because humans are adapted to what exists naturally so of course there are no naturally occurring adversarial examples. It seems like the same is true for models…

      dontreactView on HNParent context

  4. Adversarial Examples That Fool Both Human and Computer Vision 1 comment

    • PDF: https://arxiv.org/pdf/1802.08195.pdf Figure 1 - an image of a cat, adversarially perturbed to look like a dog.

      ispView on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.