A new data interpretation by a modified neural network model

This paper presents a novel approach of the use of neural networks. We are teaching the model with all patterns numerous times and use different information every time. The main idea is based on the fact that the basis training set contains information only about correlation of input information towards the output results. We made the following proposal: "If the input pattern belongs to a class of patterns C_i, then it does not belong to any other classes C_j for every j ≠ I". Here we have suggested a modified model of feed forward neural network with back propagation learning.