English

Sensitivity of sparse codes to image distortions

Computer Vision and Pattern Recognition 2022-04-18 v1

Abstract

Sparse coding has been proposed as a theory of visual cortex and as an unsupervised algorithm for learning representations. We show empirically with the MNIST dataset that sparse codes can be very sensitive to image distortions, a behavior that may hinder invariant object recognition. A locally linear analysis suggests that the sensitivity is due to the existence of linear combinations of active dictionary elements with high cancellation. A nearest neighbor classifier is shown to perform worse on sparse codes than original images. For a linear classifier with a sufficiently large number of labeled examples, sparse codes are shown to yield higher accuracy than original images, but no higher than a representation computed by a random feedforward net. Sensitivity to distortions seems to be a basic property of sparse codes, and one should be aware of this property when applying sparse codes to invariant object recognition.

Keywords

Cite

@article{arxiv.2204.07466,
  title  = {Sensitivity of sparse codes to image distortions},
  author = {Kyle Luther and H. Sebastian Seung},
  journal= {arXiv preprint arXiv:2204.07466},
  year   = {2022}
}
R2 v1 2026-06-24T10:49:11.600Z