Automatic Non-linear Feature Selection Framework for Epileptic Seizure Detection

This paper presents a qualitative automatic feature selection framework. Feature plays very important role in selecting those features which provides the best results terms of accuracy. The research work is aimed for depth analysis non-linear parameters using EEG signals. also comprehensive study and their interpretations characterizing epileptic seizures. We examine quality each independently classification performance metrics to provide meaningful information about features. Optimally setting combination leads high accuracy can further improve by combining some other with optimal combination. Experimental on two data sets shows that Hjorth (HjPm) + approximate entropy (ApEn), HjPm ApEn Higuchi fractal dimension (HFrDm) give nearly same However, combined statistical gives more than 99% most cases. providing computationally inexpensive solution be deployed low-cost hardware.