English

How much human-like visual experience do current self-supervised learning algorithms need in order to achieve human-level object recognition?

Computer Vision and Pattern Recognition 2022-05-25 v3 Machine Learning Neural and Evolutionary Computing

Abstract

This paper addresses a fundamental question: how good are our current self-supervised visual representation learning algorithms relative to humans? More concretely, how much "human-like" natural visual experience would these algorithms need in order to reach human-level performance in a complex, realistic visual object recognition task such as ImageNet? Using a scaling experiment, here we estimate that the answer is several orders of magnitude longer than a human lifetime: typically on the order of a million to a billion years of natural visual experience (depending on the algorithm used). We obtain even larger estimates for achieving human-level performance in ImageNet-derived robustness benchmarks. The exact values of these estimates are sensitive to some underlying assumptions, however even in the most optimistic scenarios they remain orders of magnitude larger than a human lifetime. We discuss the main caveats surrounding our estimates and the implications of these surprising results.

Keywords

Cite

@article{arxiv.2109.11523,
  title  = {How much human-like visual experience do current self-supervised learning algorithms need in order to achieve human-level object recognition?},
  author = {A. Emin Orhan},
  journal= {arXiv preprint arXiv:2109.11523},
  year   = {2022}
}

Comments

v3 adds DINO + robustness scaling experiments

R2 v1 2026-06-24T06:16:13.563Z