CMOS Implementation of Low Power Kohonen’s Neural Network for Medical Applications

Recently, artificial intelligence (AI) systems gain an increasing popularity and their development is fast. One of the most important fields of the AI system development are artificial neural networks (ANN’s) which, due to continuous progress in microelectronics and integrated circuits (IC) manufacturing, can be realized as large integrated blocks. Designers of such IC ANN’s take inspiration from biological networks. Networks of this type operate with analog signals, include a big number of neurons and a still bigger number of interconnections between them and, as a result, are able to process the information in parallel on a large scale. Recent advances in digital CMOS technologies enable implementation of different analog circuits in a VLSI integrated circuit form. The design and optimization efforts go towards reducing power dissipation, which is especially important in various diagnostic applications.