Platform for intrinsic evolution of analogue neural networks

Author

Rocke, Patrick ; Maher, John ; Morgan, Fearghal

Author_Institution

Dept. of Electron. Eng., National Univ. of Ireland, Galway

fYear

2005

fDate

28-30 Sept. 2005

Lastpage

11

Abstract

The goals of this research are to examine the possibility of intrinsically evolving analogue neural networks on a field programmable analogue array (FPAA). This paper discusses initial progress in this area by detailing the hardware implementation of three reconfigurable neuron models on an FPAA. Also discussed are possible techniques to allow dynamic reconfiguration of network connectivity and the plausibility of applying evolutionary algorithms to automate the discovery of neural network solutions

Keywords

evolutionary computation; field programmable analogue arrays; neural chips; analogue neural networks; dynamic reconfiguration; evolutionary algorithms; field programmable analogue array; hardware implementation; intrinsic evolution; network connectivity; reconfigurable neuron models; Artificial neural networks; Biological neural networks; Biological system modeling; Biology computing; Circuits; Evolution (biology); Field programmable analog arrays; Hardware; Neural networks; Neurons;

fLanguage

English

Publisher

ieee

Conference_Titel

Reconfigurable Computing and FPGAs, 2005. ReConFig 2005. International Conference on

Conference_Location

Puebla City

Print_ISBN

0-7695-2456-7

Type

conf

DOI

10.1109/RECONFIG.2005.29

Filename

1592493