HIGHEST ACCURACY VERSION: 2_obj_data/all_data/EEG_11_normalized_std_conv -> 90% accuracy

We hypothesize that the Electroencephalography (EEG) bioelectrical sig-nals can be used as a reliable source to analyze and recognize people’smental states. We focus on using these signals to recognize a specific cur-rent and possibly predict the next mental states provoked by stimulus. Inthis paper we recognize the current state of a human using a machine learn-ing (ML) algorithm analyzing EEG. We present basic EEG concepts anduse a provided dataset (with 2 people and 6 electrodes) to conduct variousexperiments. We train ML models by choosing what features to use andextract from the data and tune the neural network’s architecture.

The ability to recognize and possibly forecast people’s mental states canbe helpful for the development of Brain Computer Interfaces and human-machine interaction. It can also be beneficial for psychology, neural mar-keting, and further development of the brain research field. As future work,we plan to expand our EEG dataset and explore correlations between theleft and right brain hemisphere.