Unsupervised Learning by Backward Inhibition
Abstract
Backward inhibition in a two-layer connectionist network can be used as an alternative to, or an enhancement of, the competitive model for unsupervised learning. Two feature discovery algorithms based on backward inhibition are presented. It is shown that they are superior to the competitive feature discovery algorithm in feature independence and controllable grain. Moreover, the representation in the feature layer is distributed, and a certain "classification hierarchy " is defined by the
Cite
Text
Hrycej. "Unsupervised Learning by Backward Inhibition." International Joint Conference on Artificial Intelligence, 1989.Markdown
[Hrycej. "Unsupervised Learning by Backward Inhibition." International Joint Conference on Artificial Intelligence, 1989.](https://mlanthology.org/ijcai/1989/hrycej1989ijcai-unsupervised/)BibTeX
@inproceedings{hrycej1989ijcai-unsupervised,
title = {{Unsupervised Learning by Backward Inhibition}},
author = {Hrycej, Tomas},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {1989},
pages = {170-175},
url = {https://mlanthology.org/ijcai/1989/hrycej1989ijcai-unsupervised/}
}