EVALUATING THE PERFORMANCE OF SOME DEEP LEARNING ARCHITECTURES IN THE PROBLEM OF EMOTION RECOGNITION FROM EEG SIGNALS

Electronics Engineering

The problem of emotion recognition from EEG signals is a topic of great interest in research worldwide. Effectively solving these problems will provide an efficient solution in building advanced intelligent HCI systems, applicable in various fields such as healthcare, education, entertainment, military, and society. The biggest challenge is finding suitable deep learning architectures that yield high recognition accuracy. In this paper, the author evaluates the effectiveness of three deep learning architectures (CNN, LSTM, a combination of CNN and LSTM) in recognizing emotions from the internationally recognized DEAP dataset. Experimental results show that the combined CNN and LSTM architecture achieves the highest efficiency with an accuracy of 92.26%, a loss of 0.1703, and an F1 score of 0.9055. This confirms the potential application of the combined CNN and LSTM deep learning architecture in real-world emotion recognition from EEG signals.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.