Learning by dilution in a Neural Network
Disordered Systems and Neural Networks
2016-08-31 v1
Abstract
A perceptron with N random weights can store of the order of N patterns by removing a fraction of the weights without changing their strengths. The critical storage capacity as a function of the concentration of the remaining bonds for random outputs and for outputs given by a teacher perceptron is calculated. A simple Hebb-like dilution algorithm is presented which in the teacher case reaches the optimal generalization ability.
Cite
@article{arxiv.cond-mat/9611130,
title = {Learning by dilution in a Neural Network},
author = {B. Lopez and W. Kinzel},
journal= {arXiv preprint arXiv:cond-mat/9611130},
year = {2016}
}
Comments
LaTeX, 15 pages incl. figures