Research On Feature Selection Method Based On Swarm Intelligence Algorithm

With the advent of the knowledge economy and big data era,innovation in data analysis and processing techniques is particularly important,and good data processing methods can achieve twice the result with half the effort when solving problems in areas such as data mining and machine learning.The purpose of feature selection is to remove redundant and irrelevant features before the data is trained,to reduce the number of features involved in training and to improve the accuracy of the classifier.In order to solve the optimization problem of feature selection,two different intelligent algorithms are improved in this paper.The experimental results show that the proposed algorithms perform well and solve the feature selection problem well.The research work in this paper is as follows:For the Android malicious application detection problem,a self-variant genetic algorithm is proposed.Firstly,the algorithm omits the crossover operator and balances the global and local search capabilities of the algorithm by changing the population size and mutation probability.Secondly,the fitness function of the algorithm is redefined while considering the relationship between classification accuracy and features and assigning weights to them.In addition,the algorithm is used on a classification model for Android malware detection.The model first extracts Android permissions as dataset features,binary encodes them and forms a new dataset,and uses a self-variant genetic algorithm to reduce the dimensionality of the dataset in order to solve the curse of dimensionality of the features.Finally,the algorithm is validated on several datasets,and the numerical experimental results show that the algorithm is effective.For the single-objective feature selection problem,a hybrid dual-population elite preservation strategy cuckoo search algorithm with uniform mutation is proposed.Firstly,the initialization of the algorithm uses a chaotic maps to increase population diversity and ergodicity.Secondly,during the iterative process,the individuals of the population are selected using a dual-population elite preservation strategy,and the selected individuals perform a uniform mutation operation,which can speed up the speed of finding the optimal solution and improve the local search capability of the algorithm.Finally,the Lévy flight is applied to expand the search region and improve the global search ability of the algorithm.This paper sets up comparison experiments between different strategies and validates the algorithm on several datasets.The experimental results show that the performance of the proposed algorithm is significantly better than that of the comparison algorithm.