Artificial Neural Network-Based Framework for Improved Classification of Tensor-Recovered EEG Data

Electroencephalography (EEG) signals are usually affected by presence of missing data because of various reasons. Depending on the percentage of missing data, it affects significantly the classification accuracy which in turn affects the performance of prosthesis. Moreover, it has also been observed that there exists no universal classifier which performs best in all types of data. Therefore, in this paper, we propose a framework to employ tensor-based Canonical/Polyadiac Weighted-optmization (CP-WOPT) and Artificial Neural Network (ANN) to recover missing data and perform classification, respectively. The results of classification have been compared with established methods such as Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), Logistic regression, Boosted trees, Bagged trees and k-nearest neighbor (kNN). The results indicate significant improvement in classification accuracy on complete, missing and recovered data when ANN is employed. Classifiers are applied on complete, missing and recovered data explicitly to test the performance of our framework. Results show that mean classification accuracy on complete data, missing data and recovered data was 88%, 62% and 81% respectively which shows applicability of our framework.