Classification Of Amanita Model Based On Support Vector Machine With Mixed Kernel Function

With the development of modern agriculture,an increasing number of computer technologies are combined with agricultural products industry,so it is an inevitable trend to combine the research of Amanita with machine learning algorithm.The fungi,Basidiomycota,Basidiomycetes,Agaricales,Amanitaceae,and Amanita species are collectively called amanita,and Amanita will be applied widely in future,as well as economic and ecological value.Especially in the fields of biological control,and development of new medicine,Amanita has a good performance.Amanita is widely distributed in all continents of the world;in addition to some edible strains,there are some highly toxic strains.Therefore,effective classification can improve the safety of food,which is indispensable to research and establish relevant classifiable models.Many machine learning methods can perform excellent processing on data and establish a classification model.Based on the characteristic data of amanita,the SVM(Support Vector Machine)method is more suitable,for dealing with the small size of samples has a strong advantage and has a dazzling performance in the classification method.In this paper,aiming at the current problem that Amanita classification requires a lot of manpower,complicated operation and slow speed,starting from the characteristic morphological data of Amanita genus,and a classification model based on machine learning algorithm is proposed.(1)First of all,the model uses the most normalized method to preprocess the data,and uses a support vector machine as a classifier,verifies the characteristic data of Amanita,using a cross-validation algorithm to find the optimal penalty coefficient C,by comparing the classification results of kernel function support vector machine,and combined with the detailed data indicators of confusion matrix analysis,as well as the errors generated by the classification,the test results show that the linear kernel support vector machine classifier is the best for the test of amanita feature data.(2)Secondly,using the idea of kernel function fusion,the better kernel function in the experiment was improved and merged,because considering the multi-parameter problem,particle swarm optimization algorithm is used to optimize the relevant parameters,observe the final test results,and select the optimal mixed kernel function as the final improvement to be used in SVM classification model.Under the verification of the characteristic data of Amanita,the final accuracy of the improved model reached the expected target.(3)Finally,by testing and comparing the test results of mixed kernel function and single kernel function,as well as Random Forest,GBDT,SVM + Random Forest,the model of morphological feature classification of Amanita fungi based on support vector machine is feasible,and the accuracy of the final test set is 92.85%,the improved classification model of support vector machine with mixed kernel function in lays the foundation of application in future,the accuracy of the final test set is 99.43%,and the overall improvement is 6.58%.