Dynamic Classification Based Brain Emotional Learning for EEG Signal Processing in P300-based Brain and Computer Interface

Aims/ Objectives: Today, the interest in brain and computer interfaces has rapidly grown owing to the possibility of providing disabled subjects with new communication channels. Despite these interests, there are some obstacles in providing applicable BCIs. One of these obstacles is the non-stationary nature of brain signals varying from trial-to-trial and subject-to-subject. To overcome this problem, we need to design dynamic systems to adapt them to this data.
Methodology: In this paper, we propose a dynamic classifier-based brain emotional learning (DCBEL) for P300 based BCIs. This algorithm, by inspiration of brain emotional learning system, provides a dynamic system which is able to deal with non-stationary nature of brain signals. The application of the proposed method in P300 based BCIs is done for the first time. We test this system, on 4 able-bodied and 4 disabled subjects.
Results: The results showed classification accuracy of 95.39 for disabled and 93.27 for able-bodied.
Conclusion: The comparison of our results with two other algorithms multilayer perceptron and fuzzy inference system proves the superiority of our proposed algorithm.