This algorithm trains a given multilayer neural network for a given set of input patterns with known classifications. For each entry of the sample set, the network examines its ouput response to the sample input pattern. Then, the output response is compared to the desired and known output and the error is computed. Based on this error there is a weight adjustment process through the mean square error of the output response to the sample input. Hence, the set of these sample patterns are repeatedly presented to the network until the error value is minimized.
