This number is not credible. *Are we done with ImageNet?*[1] points out that the... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

Veedrac on Sept 25, 2021 | parent | context | favorite | on: Deep Learning's Diminishing Returns

This number is not credible. Are we done with ImageNet?[1] points out that the best ImageNet models that currently exist surpass the accuracy of the original labels they are trained on. Expecting to reduce error by a factor 10 on unseen data points is incoherent.

[1] https://arxiv.org/abs/2006.07159

abecedarius on Sept 25, 2021 [–]

I was unimpressed too; when I posted that snippet there were no other comments and I figured it'd help others judge whether it was worth the read.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact