Overlapping cell assemblies from correlators.

Huyck, Christian R. (2004) Overlapping cell assemblies from correlators. Neural Computing Letters, 56 . pp. 435-439. ISSN 0925-2312

[img]
Preview
PDF (paper) - Published version (with publisher's formatting)
Download (109kB)
This item is available in: Library Catalogue

Abstract

This paper introduces compensatory learning and shows how it can lead to the formation of Cell Assemblies (CAs) that share neurons. The work starts with standard Hebbian learning then adds a compensatory modifier that forces the total synaptic weight leaving a neuron toward a constant. This enables overlapping CAs because neurons that are in multiple categories have reduced weight stopping Assemblies merging. This provides a biologically plausible learning rule that eases the ability of CAs to overlap. A more complex CA system based on this work, is being developed as part of a current EPSRC grant (EP/D059720/1).

Item Type: Article
Additional Information: Awaiting copyright clearance checking. JCA. 3.3.09.
Keywords (uncontrolled): Neural networks. Cell assemblies.
Research Areas: A. > School of Science and Technology > Computer Science
A. > School of Science and Technology > Computer Science > Artificial Intelligence group
ISI Impact: 8
Item ID: 1
Useful Links:
Depositing User: Mrs Sue Black
Date Deposited: 21 Aug 2008 16:12
Last Modified: 04 May 2015 08:53
URI: http://eprints.mdx.ac.uk/id/eprint/1

Actions (login required)

Edit Item Edit Item