Feature-level fusion of mental task’s brain signal for an efficient identification system

In this research, we have explored the canonical correlation analysis (CCA) to improve the performance of the identification system that involves multiple correlated modalities. In particular, we consider the electroencephalogram signal of different mental task performed by the subject such as breathing, mental mathematics, and geometric figure rotation, visual counting and mental letter composing. Our motivation based on the fusion of feature vector of mental task using canonical correlation analysis, where feature set extraction using empirical mode decomposition and information theoretic measure and statistical measurement. In order to classify the fused vector from different mental, we have used linear vector quantization (LVQ) neural network and its extension LVQ2. The results of the experiments testing the performance have been evaluated with two profiles of the database. We have observed canonical correlation-based fusion providing the better results in comparison with simple fusion rule. The novelty of this research is the new feature generation using fused feature of distinct mental task based on CCA.