Research On Improvement And Extension Of Fruit Fly Optimization Algorithm

Traditional optimization algorithms have unique advantages for solving multi-dimensional complex problems,and group intelligence algorithms have unique advantages in solving such optimization problems.As an intelligent optimization algorithm that mimics the predation of fruit flies,the fruit fly optimization algorithm(FOA)has received more and more attention because of its easy implementation,strong local search ability and easy to understand operating principles.At the same time,with the Drosophila optimization algorithm as an example,the group intelligence algorithm is more and more widely used in various fields,The premature algorithm and the low precision of the final solution have become a problem that must be paid attention to and tried to solve.In this paper,the problem is researched and improved,and it is applied to solve the problem of text classification.Specifically,it includes the following three aspects:Firstly,FOA has the problem of easy to fall into local optimum and solve the complex problem with low precision.An adaptive modified Chaotic Map Fruit Fly Optimization(AMCFOA)algorithm is proposed for this problem.The algorithm uses the new candidate solution mechanism and adaptive step size to solve the problem that the algorithm solves complex problems with low precision and premature convergence.When the aggregation degree becomes larger in the later stage of the algorithm,the chaotic map is used to try to escape the local optimal solution.The simulation results of six classical test functions show that AMCFOA can effectively avoid premature,improve the accuracy of the solution and speed up the convergence.Secondly,Multi-swarm Wavelet Transform Fruit Fly Optimization Algorithm(MWTFOA)is proposed to solve the problem that FOA is easy to fall into local optimum,easy to prematurely converge and solve the problem of complex problems.The algorithm changes the original single population into three groups,each using a different step size strategy.In this way,the global search and local search capabilities of the algorithm are enhanced.When the degree of aggregation in the later stage of the algorithm becomes larger,the diversity of the population becomes smaller.At this time,the wavelet transform is used to increase the diversity of the population,so as to escape the local optimal solution.The simulation results of six classical test functions show that MWTFOA can effectively avoid premature,improve the accuracy of the solution and speed up the convergence.Thirdly,When the number of training samples is large,the KNN(K-Nearest Neighbor)classification efficiency is low.At the same time,when the training samples are not balanced,the classification performance of the algorithm will also be affected.For these two problems,this paper first proposes a weighted KNN text classification algorithm with variable precision rough sets.AMCFOA and MWTFOA are then applied to the improved algorithm.Experimental results show that the improved algorithm can further improve the performance and efficiency of classification.It proves that AMCFOA and MWTFOA algorithms have high application value.