stanislavfort.github.io/blog/OpenAI_CLIP_adversarial_examples

Preview meta tags from the stanislavfort.github.io website.

Linked Hostnames

11

Search Engine Appearance

Google

https://stanislavfort.github.io/blog/OpenAI_CLIP_adversarial_examples

Adversarial examples for the OpenAI CLIP in its zero-shot classification regime and their semantic generalization

It turns out that adversarial examples are very easy to find (<100 gradient steps typically) for the OpenAI CLIP model in the zero-shot classification regime. Those adversarial examples generalize to semantically related text descriptions of the adversarial class.



Bing

Adversarial examples for the OpenAI CLIP in its zero-shot classification regime and their semantic generalization

https://stanislavfort.github.io/blog/OpenAI_CLIP_adversarial_examples

It turns out that adversarial examples are very easy to find (<100 gradient steps typically) for the OpenAI CLIP model in the zero-shot classification regime. Those adversarial examples generalize to semantically related text descriptions of the adversarial class.



DuckDuckGo

https://stanislavfort.github.io/blog/OpenAI_CLIP_adversarial_examples

Adversarial examples for the OpenAI CLIP in its zero-shot classification regime and their semantic generalization

It turns out that adversarial examples are very easy to find (<100 gradient steps typically) for the OpenAI CLIP model in the zero-shot classification regime. Those adversarial examples generalize to semantically related text descriptions of the adversarial class.

  • General Meta Tags

    8
    • title
      Adversarial examples for the OpenAI CLIP in its zero-shot classification regime and their semantic generalization | Stanislav Fort
    • charset
      utf-8
    • X-UA-Compatible
      IE=edge
    • viewport
      width=device-width, initial-scale=1
    • generator
      Jekyll v3.10.0

  • Open Graph Meta Tags

    6
    • og:title
      Adversarial examples for the OpenAI CLIP in its zero-shot classification regime and their semantic generalization
    • US country flagog:locale
      en_US
    • og:description
      It turns out that adversarial examples are very easy to find (<100 gradient steps typically) for the OpenAI CLIP model in the zero-shot classification regime. Those adversarial examples generalize to semantically related text descriptions of the adversarial class.
    • og:url
      /blog/OpenAI_CLIP_adversarial_examples/
    • og:site_name
      Stanislav Fort

  • Twitter Meta Tags

    1
    • twitter:card
      summary

  • Link Tags

    2
    • canonical
      /blog/OpenAI_CLIP_adversarial_examples/
    • stylesheet
      /assets/main.css

Links

21