Backpropagation learning in analog T-Model neural network hardware
Zheng Tang, Okihiko Ishizuka, Hiroki Matsumoto · 2005
In this paper, we describe VLSI implementation of a modified backpropagation learning in the T-Model neural networks. A digitally-controlled synapse circuit and an adaptation rule circuit with a R-2R ladder network, a simple control logic circuit and an UP/DOWN counter are implemented to realize the modified backpropagation of error technique. We also present the adaptive learning using digitally-controlled synapse to the T-Model networks for several examples in order to study the learning capabilities of the analog T-Model neural hardware. These experiments show that the T-Model adaptive neural networks using the modified backpropagation can perform learning procedure quite well.