A CALIBRATION-FREE PATTERN-BASED BRAIN MACHINE INTERFACES USING ARTIFICAL NEURAL NETWORKS

Shreyas J+ ,Shabreen N , Dharamendra Chouhan , Udayaprasad P K, Dilip Kumar S M

Shreyas J+ ,Shabreen N , Dharamendra Chouhan , Udayaprasad P K, Dilip Kumar S M

Abstract

A Calibration technique is a method used brain computer interface (BCI) system, which requires a time period of 20-30 minutes. The procedure of calibration is problematic and unfeasible for building the reliable decoder. To overcome the drawback of existing system, a spectral- spatial algorithm is proposed. The data set of motor imagery (MI) which consists of 14 subjects and 15 electroencephalography (EEG) signals is taken into considerations. The two modules are constructed for data preprocessing and feature extraction. The proposed spectral- spatial algorithm is independently trained and test through artificial neural network (ANN). Based on that classification is done using several machine learning approaches like random forest (RF), neural network (NN), XGboost and given as incoming to hidden layer (Lth layer). The obtained results indicates 2% of improvement in comparison with existing methodology.