A Proposed Algorithm For Multivariate Artificial Neural Network

As some physiological presumptions of Brain Learning techniques led to develop Artificial Neural Network ( ANN ) to compete with the Brain’s Natural learning phenomenon. Various Networks and Algorithms have been proposed to enhance the machine learning process and to achieve some thing new. In this paper we have proposed a moderate Algorithm for Multivariate Artificial Neural Network. Introduction For last few decades the studies over brain learning system are being conducted and huge possibilities are being explored to design a competent Artificial Neural Network resembling the brain efficiencies. Also many algorithms in this direction have been proposed by various authors to meet the goal. Here we propose a moderate Algorithm for multivariate artificial neural network. Let i x is a neuron with weight i w so its strength is i w i x , and let we have i m neurons of type i x . Then the combined probabilistic strength of type i x will be Now let a neural network has n – units with m – external input lines with unified activation function f ( n i i m 1 ). The system output Y for the weight vector w = ( n w w ,..., 1 ) may be taken as ( 1 0 i w ), (Because in case of indefinite large numbers the probabilistic models are best suited for infinity ). Y = n