Recognition efficiency enhancement of control chart pattern using ensemble MLP neural network

chart pattern for six different fundamental patterns have been proposed. The multilayer perceptron based neural network has applied as an entity to form the ensemble. The loss of generalization property of an individual neural network has been fulfilled by providing the diversity in the training data for the individual entity of ensemble. The outcome of individual recognizer has combined over their decision value to include the outcome knowledge of each and every entity instead of majority voting of decision. The proposed method has been verified over a large set of data experimentally and observed that there is significant improvement with proposed ensemble approach in compare to individual recognizer.