Unsupervised learning of neural networks for separation of unknowndata

摘要

In neural networks, the learning scheme is very important and,basically, is divided into supervised learning and unsupervisedlearning. If one would like to classify a set of data, the statistics ofwhich are not known, then one cannot apply an ordinary supervisedlearning scheme. On the other hand, if one can embed a relation betweeninput data and teaching signals into an evaluation function, one canallow neural networks to learn the relation. In this paper, the authorspropose an unsupervised learning scheme and an evaluation function thatrealizes a classification of unknown data. Some simulation results arealso shown