Mastering Machine Learning Algorithms - Second Edition

14

Hebbian Learning

In this chapter, we're going to introduce the concept of Hebbian learning, based on the methods defined by the psychologist Donald Hebb. These theories immediately showed how a very simple biological law is able to describe the behavior of multiple neurons in achieving complex goals and was a pioneering strategy that linked the research activities in the fields of artificial intelligence and computational neuroscience.

In particular, we are going to discuss the following topics:

• The Hebb rule for a single neuron, which is a simple but biologically plausible behavioral law.
• Some variants on the Hebb rule, like Oja's and Covariance rules, that have been introduced to overcome a few stability problems.
• The final result achieved ...