ML Anthology

Learnable and Non-Learnable Visual Concepts

Haim Shvaytser
ICCV 1988 pp. 264-268
doi:10.1109/CCV.1988.589998 /iccv/1988/shvaytser1988iccv-learnable/

Abstract

Valiant’s theory of the learnable is applied to visual concepts in digital pictures. Several visual concepts that are easily perceived by humans are shown to be learnable from examples by a simple algorithm. These include inaccurate copies of a picture, multiple objects, and lines in a fixed slope. However, there are simple characterizations of visual concepts which are apparently non-learnable from examples. It is shown that templates for template matching cannot be learned from a feasible number of positive examples.

ICCV Semantic Scholar

Cite

Text

Shvaytser. "Learnable and Non-Learnable Visual Concepts." IEEE/CVF International Conference on Computer Vision, 1988. doi:10.1109/CCV.1988.589998

Markdown

[Shvaytser. "Learnable and Non-Learnable Visual Concepts." IEEE/CVF International Conference on Computer Vision, 1988.](https://mlanthology.org/iccv/1988/shvaytser1988iccv-learnable/) doi:10.1109/CCV.1988.589998

BibTeX

@inproceedings{shvaytser1988iccv-learnable,
  title     = {{Learnable and Non-Learnable Visual Concepts}},
  author    = {Shvaytser, Haim},
  booktitle = {IEEE/CVF International Conference on Computer Vision},
  year      = {1988},
  pages     = {264-268},
  doi       = {10.1109/CCV.1988.589998},
  url       = {https://mlanthology.org/iccv/1988/shvaytser1988iccv-learnable/}
}