Abstract: This paper discusses in detail the various advanced neural network algorithms, which are used to solve critical prediction applications such as, predict the traffic on highways or predict number of calls in the call center. Application of neural network for the various prediction application methods to the real life data is challenging due to very large size of the data, high dimensionality and presence of seasonal variations. Predicting the call center traffic or number of vehicles on highways for each time slot for forthcoming day (s) is of immense importance for planning sufficient resources to provide satisfactory and quick customer service. The paper discusses the use of back propagation algorithm for training the neural network using the historic data of the application. Then it discusses the need for improvement in the back propagation algorithm and the use of another algorithm, which is called as back propagation with momentum (BPM) algorithm, which converges faster than back propagation algorithm. The paper also discusses the use of another algorithm called as super self-adapting back propagation (SUPERSAB) as it shows more accurate results than other algorithms for the prediction applications.
Keywords: Neural Network, Back Propagation, Momentum, Prediction, SuperSAB