Title :
Fully digital neural network implementation based on pulse density modulation
Author :
Tomberg, Jouni ; Ritoniemi, Tapani ; Kaski, Kimmo ; Tenhunen, Hannu
Abstract :
An efficient implementation of a Hopfield-type fully connected neural-network architecture is presented that is based on a pulse-density modulation technique implemented by using fully digital structures. The synaptic weights are programmable, and thus the area of one synapse and the entire network depends on the resolution of the weight. Advantages of the design are its modularity and expandability
Keywords :
VLSI; digital integrated circuits; neural nets; parallel architectures; Hopfield-type fully connected neural-network architecture; VLSI; digital neural network implementation; expandability; fully digital structures; modularity; programmable synaptic weights; pulse density modulation;
Conference_Titel :
Custom Integrated Circuits Conference, 1989., Proceedings of the IEEE 1989
Conference_Location :
San Diego, CA, USA
DOI :
10.1109/CICC.1989.56744
