Relation generation and fast similarity testing for unsupervised learning
Weibo Gong, Bo Jiang · 2017
Unsupervised learning of concepts is essential for analogical thinking, which is considered a hallmark of human intelligence. We argue that signals in the brain constantly generate relations among their constituents. Unsupervised learning is then accomplished by fast similarity testing of the stored relation sets using resonance transients and time sequenced memory units. We discuss specific algorithms to achieve these functions.