Using Mixed Linear Programming to Classify BCI Signals Submission

Abstract

There are a lot of possibilities to control devices remotely nowadays. Doing this with brain waves is not a science fiction anymore. The first steps for device control by brain are already made. The main challenge is identification and classification of the human waves corresponding to the movement, emotions and so on. Brain-Computer Interfaces (BCI) devices provide a connection between the user's brain and the computer. These can be observed in our daily life, such as Medical Applications, Neuroergonomics and Smart Environment, Neuromarketing and Advertising, Education and Self-Regulation, Games and Entertainment, Security and Authentication, etc. To reach the best result a reliable classification of BCI signals is needed. Controlling machines and making science fiction with the power of our mind is a reality today. This study deals with increasing accuracy of the processing and classification of the Brain Wave Signals. This approach is based on pre-determination of the channels with useful information and consequently precise classification of the signals. In this experiment, the BCI device Emotiv 14-Epoc is used for collecting brain signals. They are received from BCI devices. The signals classified are based on the Mixed Linear Programming (MLP). This is an approximation to reach the goal- precise identification of brain signals. The results indicate the necessity of approaches that enhance the characteristics: capabilities and quality. The main results of the presented work show the accuracy of signals identification is improved from 60% to 90% when preliminary determination of the channels is done.