A Combining Method with Many Sets of Weights and Biases in Pattern Recognition Neural Network

Abstract

In Supervised training method, the training data is a pair consisting of an input object (typically a vector) and a desired output value (also called supervisory signal). Pattern recognition systems are in trained from labeled ‘training’ data, but when no labeled data are available, other algorithms can be used to discover previously unknown patterns. Back propagation requires a known, desired output for each input value in order to calculate the loss function gradient. It is therefore usually considered to be a supervised learning method, although it is also used in some unsupervised networks such as auto encoders. In this paper, we proposed a training method to classify all the training patterns using designed neural network pattern recognition. For more classification, the neural network designed as training data set has to be separated with a reject output. Neural network will find not only one but many sets of weights and biases with the help of training method being classify all the training patterns , control the recognized rejection and error rate is reduced. The proposed method can reduce the neural network size and to design fast smart sensors for the robots, it can be implemented on a Field Programmable gate array chip.