In this paper, A CNN-based ensemble model for epileptic seizure detection is proposed. The proposed model improves seizure detection performance through a structure that merges the training results of AlexNet, VGG16, VGG19 models and retrains the merg...
In this paper, A CNN-based ensemble model for epileptic seizure detection is proposed. The proposed model improves seizure detection performance through a structure that merges the training results of AlexNet, VGG16, VGG19 models and retrains the merged data into the MLP model. In addition, the proposed model rearrange the learning results of the three models used in the merge phase into one-dimensional data, learn the merged data in the re-learning phase into an MLP model with a fully connected layer, and derive the final results through the softmax function. As a result of the CPSM experiment using the CHB-MIT Scalp EEG Database with the proposed model, the average sensitivity of 92% and the FPR of 0.36 were obtained.