An FPGA Implementation of Kak's Instantaneously-Trained, Fast-Classification Neural Networks

Zhu, J. and Sutton, P. R. (2003). An FPGA Implementation of Kak's Instantaneously-Trained, Fast-Classification Neural Networks. In: K. Asada and M. Fujita, Proceedings of the 2003 IEEE International Conference on Field-Programmable Technology (FPT). The 2003 IEEE International Conference on Field-Programmable Technology (FPT), Tokyo, Japan, (126-133). 15-17 December, 2003.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
FPT2003ZhuFInalS.pdf FPT2003ZhuFInalS.pdf application/pdf 165.22KB 840
Author Zhu, J.
Sutton, P. R.
Title of paper An FPGA Implementation of Kak's Instantaneously-Trained, Fast-Classification Neural Networks
Conference Paper Type Fully Published Paper
Conference name The 2003 IEEE International Conference on Field-Programmable Technology (FPT)
Conference location Tokyo, Japan
Conference dates 15-17 December, 2003
Proceedings title Proceedings of the 2003 IEEE International Conference on Field-Programmable Technology (FPT)
Editor K. Asada
M. Fujita
Place published Tokyo, Japan
Publisher The University of Tokyo
Publication date 2003
Volume number 1
ISBN 0-7803-8320-6
Start page 126
End page 133
Total pages 8
Collection year 2003
Language eng
Abstract/Summary Motivated by a biologically plausible short-memory sketchpad, Kak's Fast Classification (FC) neural networks are instantaneously trained by using a prescriptive training scheme. Both weights and the topology for an FC network are specified with only two presentations of the training samples. Compared with iterative learning algorithms such as Backpropagation (which may require many thousands of presentations of the training data), the training of FC networks is extremely fast and learning convergence is always guaranteed. Thus FC networks are suitable for applications where real-time classification and adaptive filtering are needed. In this paper we show that FC networks are "hardware friendly" for implementation on FPGAs. Their unique prescriptive learning scheme can be integrated with the hardware design of the FC network through parameterization and compile-time constant folding.
Subjects 280212 Neural Networks, Genetic Alogrithms and Fuzzy Logic
291600 Computer Hardware
E1
291699 Computer Hardware not elsewhere classified
671299 Computer hardware and electronic equipment not elsewhere classified
Keyword esgweb-research-jihan
neural networks
Q-Index Code E1

 
Versions
Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 2608 Abstract Views, 840 File Downloads  -  Detailed Statistics
Created: Fri, 20 Feb 2004, 10:00:00 EST by Jihan Zhu on behalf of School of Information Technol and Elec Engineering