Analog VLSI Neural Network Testing

Abstract

As research works in analog VLSI have started evolving from research material to designing neural electronic applications. It is important, in view of testing complexities to think ahead on how to design testable neural circuits be it in discrete or VLSI level. For such applications to be reliable, there are steps necessary to understand. In other to deal with the issues of complexities in the hidden layer of the neural network. The methods proposed, considered utilities of analog design. Which are useful in both design and manufacturing of the analog VLSI artificial neural network (ANN). This work, therefore, revolves around design for testability with test coverage using ANN of the fault dictionary. The presented techniques show practical approach towards design and realization of reliable ANN.