Application of SVM Based on Improved Particle Swarm Optimization Algorithm in Epileptic Seizure Detection

In order to solve the problem of low classification accuracy of epilepsy Electroencephalography(EEC) signals in medical diagnosis, this paper proposes the CPSO-SVM model based on the classification particle swarm optimization (CPSO) algorithm and support vector machine. In order to solve the problem that the inertia weight cannot be dynamically evaluated in the traditional Particle Swarm Optimization(PSO), a CPSO algorithm based on particle fitness classification is proposed to realize the adaptive inertia weight. In this paper, the proposed algorithm is applied to the paranicter selection of support rector machines to obtain the optimal combination of kernel function parameter g and penalty factor C for the optimal classification performance of support vector machines. Firstly, the preprocessed sample data are decomposed by wavelet packet to obtain the frequency band energy ratio characteristics. Secondly, the CPSO algorithm is used to set the optimal parameters of SVM. Finally, the CPSO-SVM classification model is obtained by inputting the training set. The simulation results show that the proposed CPSO algorithms has significantly improved optimization effect compared with traditional PSO algorithm, and the classification accuracy of CPSO SVM model is better than PSO-SVM and typical support vector machine.