Meditation-induced electroencephalography-based feature optimization technique using Aquila Optimizer algorithm

Meditation is a practice known for its ability to enhance mental clarity and emotional well-being. Recent years have witnessed a considerable growth in the application of nature-inspired algorithms for feature optimization. Existing research shows a scarcity of nature-inspired algorithms applied to EEG data in meditation contexts, prompting the need for effective methods to enhance classification performance. To address this gap a methodology which makes use of most recent optimization algorithm such as modified Aquila Optimizer (mAO) has been utilized for feature selection with XGBoost as classifier on rarely explored meditation dataset. This version is yet to be explored on EEG data. Initially the pre-processed EEG dataset is converted to the time-frequency images using efficient S-transforms. These images are fed to an auto-encoder to generate relevant feature vectors. These feature vectors are given as input to mAO for feature selection with XGBoost as a classifier. The Cauchy distribution function is used in place of the Levy distribution function in the modified version which offers superior convergence properties. In order to assess the performance of the model various metrics such as quantity of features, accuracy and execution duration were used to perform a comparative analysis with Aquila Optimizer (AO). It was observed that mAO outperforms AO in terms of duration of execution. This research demonstrates the efficacy of the proposed approach in contrast to cutting edge techniques in enhancing classification performance on EEG data for meditation analysis.