Neuro signals: A future biomertic approach towards user identification

Electroencephalography (EEG) have been receiving a lot of attention due to its recent use in the field of biometrics. Signals traced from the different parts of the brain has become an upsurge area of interest for the researchers. Evidences have been provided by the research communities where the uniqueness of neuro-signals can possibly be used for building a robust biometric identification system. In this paper, we investigate the robustness of EEG signals in two different scenario of data collection, namely, Eyes Open (EO) and Eyes Closed (EC) for building a person identification system. For this, a publicly available EEG signals dataset of 109 users have been used. The EEG signals have been modeled using two different classifier, namely, Support Vector Machine (SVM) and Random Forest (RF). Next, a feature selection approach has been applied to reduce the number of features and results have been computed to find optimal feature dimension. From experiments, person identification rates of 97.64% (EO) and 96.02% (EC) using SVM, and 98.16% (EO) and 97.30% (EC) have been recorded using RF classifiers.