We explore the ability of DWT-based evenlope analysis in the feature extraction of epilepsy signals.

A novel network structure is proposed for 3-class EEG detection based on the new concept.

In order to obtain more reliable results, different feature extracting techniques and classifiers have been considered in this work.

Our framework has achieved a satisfied classification accuracy in comparison with the existing methods.

The proposed scheme can be a suitable candidate for extensive clinical application in terms of the classification robustness and stability.