تشخيص ميزان پيچيدگي فعاليت هاي ذهني بوسيله سيگنالهاي مغزي با استفاده از شبكه هاي عصبي

EEG-based Detection of Mental Workload by Using Artificial Neural Networks

It has been shown that the human brain activity during performance of the mental tasks are reflected by the electroencephalographic recordings. These suggest the possibility of using the event-related EEG signals to observe the brain processes during mental activity. This paper concerns with the classification of four different mental tasks: a baseline task, three different mental multiplication. During baseline, subject did not perform a specific task. During mental multiplication, three different levels of mental arithmetic were considered. The mean absolute value (MAV), variance, the normalized power of alpha, teta, and beta band, the relative power of the beta to the alpha band, and the relative power of the teta to the alpha band constitute the feature vector. The feature vectors were fed into classifiers. We employed the multilayer perceptron (MLP) for discriminating different patterns of the EEG signal. The results of this analysis show that 92% of the novel data were classified correctly. This suggests the possibility of using brain potentials to quantify the mental workload.