Improving the Feature Stability and Classification Performance of Bimodal Brain and Heart Biometrics

Abstract

Electrical activities from brain (electroencephalogram, EEG) and heart (electrocardiogram, ECG) have been proposed as biometric modalities but the combined use of these signals appear not to have been studied thoroughly. Also, the feature stability of these signals has been a limiting factor for biometric usage. This paper presents results from a pilot study that reveal the combined use of brain and heart modalities provide improved classification performance and furthermore, an improvement in the stability of the features over time through the use of binaural brain entrainment. The classification rate was increased, for the case of the neural network classifier from 92.4% to 95.1% and for the case of LDA, from 98.6% to 99.8%. The average standard deviation with binaural brain entrainment using all the inter-session features (from all the subjects) was 1.09, as compared to 1.26 without entrainment. This result suggests the improved stability of both the EEG and ECG features over time and hence resulting in higher classification performance. Overall, the results indicate that combining ECG and EEG gives improved classification performance and that through the use of binaural brain entrainment, both the ECG and EEG features are more stable over time.

A Survey on Brain Biometrics

“…It estimates the parameters using the Levinson-Durbin algorithm based on the last autoregressive-parameter estimated from each model order p by minimizing the forward and backward prediction error. In order to determine the optimal order of the AR model, there are generally three methods: minimizing the error of the predictor equation through experimental results with different orders, minimizing the Akaike Information Criterion (AIC), and based on the eigenvalues of the matrixR in the Yule-Walker equations [5,54,69,118].…”

“…Several papers using NNs to make decisions in brain biometric recognition are listed in Table 8. Many studies used the 112:21 one-hidden layer NN, which was constructed using one input layer, one hidden layer, and one output layer [43,52,53,55,60,105,116,118,135,153,162]. Besides the conventional one-hiddenlayer NN, the Deep Neural Network (DNN) with two or more hidden layers, was also investigated.…”

“…Table 9 summarizes studies that adopted DA methods. In addition to LDA [2,24,61,73,74,80,84,106,118,121,147,179], other DA variants are available, such as Diagonal Linear Discriminant Analysis (D-LDA), which is similar to LDA but with diagonal covariance matrices; Quadratic Discriminant Analysis (QDA), which fits multivariate normal density with covariance that is not identical for each class; and Diagonal Quadratic Discriminant Analysis (D-QDA), which is similar to QDA but with diagonal covariance matrices [73,138,147].…”

“…Table 12 lists the papers which used multiple traits for authentication. EEG was combined with eye blinks [2,178], ECG [9,10,118,155], face [135,172], iris [66], hand kinematic synergies [124], fingerprint [157], ID and gait [185], EOG, GSR, and a Respiration Belt (Rb) [86,103], and performance improved.…”

Effect of tDCS stimulation of motor cortex and cerebellum on EEG classification of motor imagery and sensorimotor band power

“…This is a limitation of the study because the ERS analysis cannot describe satisfactorily the EEG changes that may lead to accuracy improvement. In this case, evaluation of the montage during BCI training would be useful for the reason that brain entrainment is suspected to minimize feature variability over the experimental sessions [ 70 ], which could facilitate ERS analysis. It should be noted that not all subjects have the skills for performing detectable desynchronization patterns in their EEG during MI without training [ 71 ], so brain entrainment would offer more stable EEG classification features and the possibility of evaluating the montage in a learning context that represents better the conditions in which the tDCS montage was expected to exhibit its hypothesized behavior.…”