Research On Classification Algorithm Based On Artificial Bee Colony Algorithm

In recent years,more and more people research on high-dimension data in data mining.Classification is an effective method for data analysis.However,the ever-increasing dimension of the data including some irrelevant or redundant features increase the difficulty of classification.So,the selection of feature subsets and optimization of parameters of classifier are two important factors to improve the classification performance for the classification problem.As a new algorithm,intelligent optimization algorithm can be used to solve the complex practical problems by simulating the group behavior.For instance,artificial bee colony algorithm can be used to solve the problems by simulating the intelligent foraging behavior of honeybee.The algorithm not only makes the local searching efficient but also has the ability of global optimization with the help of heuristic search strategy.Based on the advantages of simple implementation and strong robustness,it has been used in pattern recognition and intelligent control.Therefore,the feature selection and optimization of support vector machine parameters based on artificial bee colony algorithm is proposed.The goal of algorithm is to improve the classification rate and reduce the number of features.The proposed algorithm is validated on the UCI database.The result show that the proposed algorithm has the ability of feature selection and it can improve the classification rate obviously.Compared with other algorithms,the proposed algorithm is more efficient in feature selection and classification.Considered the randomness of the initial solution of artificial bee colony algorithm,the concept of integer tents is introduced to improve the initial stage and scout bee stage.And the performance of the improved algorithm is verified by the standard test function.The results show that the improved algorithm not only accelerates the convergence speed but also improves the algorithm precision.The proposed algorithm is also used for feature selection and parameter optimization.By using the same data set to verify the results,it is found that the improved algorithm has higher classification rate and fewer features.Finally,multi-objective artificial bee colony algorithm is used to design the classifier with the classification rate and feature number as their goals,without setting the weight.And the algorithm is used for data classification,the classification result have improved compared with the previous methods.We can make a decision on how to choose the result of high classification rate or fewer features.