5:3 compressor based neural network with LM Algorithm in Multiplier as Application

Abstract

Abstract In conventional approach for VLSI, implementing large numbers of operations in parallel is possible with NN. The fundamental task of a neural network hardware is not dependent on the implementation technology and are quite constructive in simulating the digital circuits. These NN representations can be incorporated in various applications where in the behaviour of these circuits are essential to get a solution for discrete problems. Therefore, supervised learning algorithms are used to train the computer software. One such algorithm is the L-M method which is one of the powerful and widely used approaches. This paper accounts to the designing and training of a neural network model as computing technique in 5:3 compressor with Lavenberg-Marquardt Algorithm. The performance parameters are significantly remarkable for real time implementation, are presented in the paper.