Performance improvement of P300-based home appliances control classification using convolution neural network

Abstract

P300 speller-based brain-computer interface (BCI) is an immediate correspondence from a human brain to a computer that depends on the translation of mind reactions produced by stimulus of a subject utilizing a P300 speller. No muscle movements are required for this communication. The present study helps the disabled people (viz.patients with spinal cord injury, spastic cerebral palsy, locomotive diseases, etc.) ease their lives by accessing light, fan, mobile device, door, television, electric heater, etc. A novel 2 × 3 matrix consisting of home appliances visuals has been proposed as a P300 paradigm. Once the matrix visualization is completed, auditory feedback was provided and the chosen command was executed. It enables six menus to be navigated for hadling six electronic systems with up to 30 control commands. The objective of this research is to develop a P300-based BCI system for operating different electronic appliances at home. The existing BCI-based P300 spellers with standard machine learning methods suffers from low information transfer rate (ITR) and poor classification accuracy. With the application of convolution neural network (CNN) for classification, the proposed system improves the performance in terms of P300 classification & target appliances detection. The proposed system is designed to be user-convenient & cost-effective in terms of hardware design. The experiments have been performed on the dataset acquired from 30 target item selections from nine subjects using 16 channel actiCAP Xpress EEG recorders. The experimental result demonstrates that the proposed CNN model achieved 90% classification accuracy. It also offers ITR of 22.3 bits per minute which is substantially greater than the existing methods. The presented approach shows its novelty in the real-time applications.