Abstract
This paper presents an important revision of Barlow's theory of redundancy reduction and greatly clarifies much of the confusion and misconceptions that have surrounded the theory. Barlow's original 1961 proposal was that neurons should attempt to encode sensory information so as to reduce the redundancy (or statistical dependencies) present in the input stream coming from the sensory receptors. Here Barlow points out...

This publication has 6 references indexed in Scilit: