Corpus ID: 223956855

On the surprising similarities between supervised and self-supervised models

@article{Geirhos2020OnTS,
  title={On the surprising similarities between supervised and self-supervised models},
  author={Robert Geirhos and Kantharaju Narayanappa and Benjamin Mitzkus and M. Bethge and Felix Wichmann and W. Brendel},
  journal={ArXiv},
  year={2020},
  volume={abs/2010.08377}
}
  • Robert Geirhos, Kantharaju Narayanappa, +3 authors W. Brendel
  • Published 2020
  • Computer Science, Biology
  • ArXiv
  • How do humans learn to acquire a powerful, flexible and robust representation of objects? While much of this process remains unknown, it is clear that humans do not require millions of object labels. Excitingly, recent algorithmic advancements in self-supervised learning now enable convolutional neural networks (CNNs) to learn useful visual object representations without supervised labels, too. In the light of this recent breakthrough, we here compare self-supervised networks to supervised… CONTINUE READING
    1 Citations

    Figures from this paper

    References

    SHOWING 1-10 OF 27 REFERENCES
    Unsupervised Feature Learning via Non-parametric Instance Discrimination
    • 338
    • PDF
    A Simple Framework for Contrastive Learning of Visual Representations
    • 684
    • PDF
    Generalisation in humans and deep neural networks
    • 163
    • PDF
    Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet
    • 188
    • PDF
    Deep convolutional networks do not classify based on global object shape
    • 93
    • PDF
    Unsupervised Neural Network Models of the Ventral Visual Stream
    • 6
    • PDF
    The Origins and Prevalence of Texture Bias in Convolutional Neural Networks
    • 8
    • PDF