Improvement And Application Of Ensemble Learning Algorithm

Neural network ensemble through training a number of neural network and the results of integration can significantly improve the generalization ability of learning systems. Due to its superior Performance and broad applicability, ensemble learning has become a hot topic of the machine learning community.In this paper,through doing some research on the theory of Bagging algorithm and cellular automata,we present a new Bagging method model based on cellular automata which use cellular automata state characteristics of synchronous converter to disturb sample thus increasing the individual network differences to enhance the generalization ability of learning systems.Through doing some research on the theory of neural network. We design and implement a reasonable RBF neural network as the base classification of Bagging algorithm. Then integrate the cellular automata technology into Bagging ensemble algorithm. In building a cellular automata model, we use genetic algorithms to find optimal cell transfer rules, which make each round of training sample set with the biggest differences.Finally, the improved Bagging algorithm based on cellular automata is applied to the Radar Emitter Recognition. Through feature extraction and normalization processing on Radar Emitter data, the Pretreatment radar simulation data is used as experimental data sets,then compare the improved algorithm and the traditional Baggig algorithm by using the radar data sets, experiment shows that the improved algorithm has better generalization ability.