- [R] Overinterpretation reveals image classification model pathologies - e.g. prominent models classify a black image w/ 4-5 gray pixels as "airplane" with >99% confidence https://arxiv.org/abs/2003.08907 47 comments machinelearning
Linking pages
- Farcical Self-Delusion – Piekniewski's blog https://blog.piekniewski.info/2021/12/18/farcical-self-delusion/ 10 comments
- If We Succeed | American Academy of Arts and Sciences https://www.amacad.org/publication/if-we-succeed 1 comment
- Researchers Propose Mitigation Strategies to Tackle Overinterpretation of Deep Learning Methods - MarkTechPost https://www.marktechpost.com/2022/01/02/researchers-propose-mitigation-strategies-to-tackle-overinterpretation-of-deep-learning-methods/ 0 comments
- Overinterpretation May Be a Bigger and More Intractable Threat Than Overfitting - Unite.AI https://www.unite.ai/overinterpretation-may-be-a-bigger-and-more-intractable-threat-than-overfitting/ 0 comments
Would you like to stay up to date with Computer science? Checkout Computer science
Weekly.
Related searches:
Search whole site: site:arxiv.org
Search title: [2003.08907] Overinterpretation reveals image classification model pathologies
See how to search.