Hybrid WCA–PSO Optimized Ensemble Extreme Learning Machine and Wavelet Transform for Detection and Classification of Epileptic Seizure from EEG Signals

Epilepsy seizures are sudden, chaotic neurological functions. The complexity of the brain is revealed via electroencephalography (EEG). Visual examination-based EEG signal analysis is time-consuming, expensive, and difficult. Epilepsy-related mortality is a serious concern. In the diagnostic procedure, computer-assisted diagnosis approaches for precise and automatic detection and classification of epileptic seizures play a crucial role. Due to the classifier's high processing time requirements caused by its mathematical complexity and computational time, we propose a hybrid water cycle algorithm (WCA)–particle swarm optimization (PSO) optimized ensemble extreme learning machine (EELM) classification of seizures to improve the classification performance of the classifier. Firstly, we use feature extraction by utilizing the wavelet transform. The extracted features are aligned as input to the WCA–PSO–EELM for classification. The particle swarm optimization (PSO) algorithm is used to initialize the optimization variables of a WCA algorithm, and the WCA algorithm is used to optimize the input weight of the ELM (i.e., the WCA–PSO–ELM (WPELM)) for classification of seizure and non-seizure EEG signals. University of Bonn database is used for the experiment. The performance measures sensitivity, specificity, and accuracy are considered and achieved 98.78%, 99.23%, and 99.12%, that is, higher than those of other conventional algorithms. To validate the robustness of the WCA–PSO algorithm, three benchmark functions are considered for optimization. The comparison results are presented to visualize the uniqueness of the proposed WCA–PSO–EELM classifier. From the comparison results, it was observed that the proposed WCA–PSO–EELM model outperformed in classifying the seizure and non-seizure EEG signals.