A low power trainable analogue neural network classifier chip
Philip H. W. Leong, M.A. Jabri · 2002
The authors describe an analogue VLSI chip called Kakadu which implements a trainable (10, 6, 4) multilayer perceptron. Kakadu is a classifier designed for low-power applications and has a typical power consumption of 20 /spl mu/W. It was tested on many classification problems, including XOR, 4-b parity, character recognition, and arrhythmia classification.