Application of GA-BP Neural Network in MMS Index Prediction

MMS (Multimedia Messaging Service) index prediction plays an important role to the development of MMS business. Currently, MMS index prediction model based on back propagation (BP) neural network easily falls into local minimum because of the randomness of weights selection. So the genetic algorithm which is good at global search optimizes weights of BP neural network. Firstly, the genetic algorithm searches weights in the global scope and then BP algorithm further optimizes weights. Finally, the optimized algorithm achieves better prediction accuracy.