Decoding Emotions From EEG Responses Elicited by Videos Using Machine Learning Techniques on Two Datasets.

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Current electroencephalogram (EEG)-based emotion identification methods often rely on complex deep learning models requiring substantial computational resources.
• There is a need for more accessible and less resource-intensive approaches for human emotion recognition using EEG signals.

Purpose of the Study:

• To propose and evaluate a simpler architectural method for human emotion recognition based on EEG data.
• To compare the performance of feature extraction directly from epochs versus decomposed brain rhythms.
• To investigate the effectiveness of different classifiers and feature combinations on public EEG datasets.

Main Methods:

• Utilized two public datasets: SEED and DEAP.
• Segmented EEG signals into 1-second epochs and decomposed them into brain rhythms.
• Computed features directly from epochs and from brain rhythms, examining various combinations.
• Employed different classifiers, including Support Vector Machine (SVM) and Multilayer Perceptron (MLP).
• Applied baseline feature correction for the DEAP dataset.

Main Results:

• Support Vector Machine (SVM) demonstrated superior performance on the DEAP dataset when baseline feature correction and epoch decomposition were combined.
• Achieved an average accuracy of 96.50% for high versus low valence and 96.71% for high versus low arousal classes on the DEAP dataset.
• Attained a best average accuracy of 86.89% on the SEED dataset using a Multilayer Perceptron (MLP) with two hidden layers.

Conclusions:

• A simpler EEG-based emotion recognition architecture can achieve high accuracy, offering a viable alternative to complex deep learning models.
• The proposed method, particularly with SVM on the DEAP dataset, shows significant potential for practical applications.
• This research could pave the way for developing automated mental health monitors for preliminary medical screening.