Model Concept
Attractor Neural Network
A recurrently connected neural network with iterative activation update. A noisy initial pattern (of neural activity) will converge to one of a set of template patterns that is stored in the network by its weighted connections. Because of this property it is also a model of Associative Memory. The Continuous Attractor Network uses continuous activation values (~ firing rates), while the Binary Attractor Network uses {0,1} or {-1,1} as possible neuronal activations.
Other categories referring to Attractor Neural Network
Computational model.Model Concept   (26)
Revisions: 3
Last Time: 12/13/2006 3:43:13 PM
Reviewer: Tom Morse - MoldelDB admin
Owner: Tom Morse - MoldelDB admin