GANGLION-a fast hardware implementation of a connectionist classifier

Author

Cox, Charles E. ; Blanz, W. Ekkehard

Author_Institution

IBM Almaden Res. Center, San Jose, CA, USA

fYear

1991

fDate

12-15 May 1991

Abstract

The authors describe the implementation of GANGLION, a fully interconnected, digital, feedforward connectionist classifier with one hidden layer capable of 4.48 billion interconnections per second. The single card classifier, built using only off-the-shelf components, relies heavily on field-programmable gate arrays. The high speed of the classifier makes it especially suited for applications in machine vision, and in particular in industrial inspection tasks, where sensor speeds beyond the speed of ordinary TV cameras are common

Keywords

automatic optical inspection; computer vision; computerised pattern recognition; computerised picture processing; logic arrays; neural nets; FPGA; GANGLION; VME card; artificial neural networks; connectionist classifier; digital type; fast hardware implementation; feedforward; field-programmable gate arrays; industrial inspection; machine vision; single card classifier; Adaptive arrays; Artificial neural networks; Costs; Feeds; Hardware; Logic arrays; Pattern recognition; Production; Programmable logic arrays; Prototypes;

fLanguage

English

Publisher

ieee

Conference_Titel

Custom Integrated Circuits Conference, 1991., Proceedings of the IEEE 1991

Conference_Location

San Diego, CA

Print_ISBN

0-7803-0015-7

Type

conf

DOI

10.1109/CICC.1991.164044

Filename

164044