Synthesis of a class of artificial neural network using a CMOS current mode building block approach

An analogue current mode design methodology is presented for the implementation of binary associative memories. Thus by using the Hamming feedfoward network as an example of a binary associative memory, and efficiently mapping it into two CMOS-VLSI building block configurations a prototype ANN is developed. The IC building blocks utilizes the characteristics of the subthreshold region and converge to a classification in 'one pass' requiring no feedback; thereby resulting in an analogue network with the desirable features of high integration, low circuit complexity, and low power dissipation.<
>