Emotional states recognition, implementing a low computational complexity strategy.

Area of Science:

• Neuroscience
• Computer Science
• Biomedical Engineering

Background:

• Recognizing emotional states from electroencephalography (EEG) signals is crucial for various applications.
• High computational burden often limits the practical implementation of EEG-based emotion recognition systems.
• Efficient data processing and pattern recognition are key challenges in this field.

Purpose of the Study:

• To develop a computationally efficient methodology for emotion recognition using EEG signals.
• To reduce the data processing requirements by optimizing electrode selection based on Brodmann regions.
• To present design suggestions for low-complexity neural networks and support vector machines for pattern recognition.

Main Methods:

• Establishing relationships between EEG electrodes and Brodmann regions to identify and discard irrelevant electrodes.
• Implementing data reduction strategies to minimize computational load.
• Utilizing low computational complexity neural networks and support vector machines for pattern recognition.

Main Results:

• A methodology for emotion recognition from EEG signals with reduced computational burden was developed.
• The approach successfully identified relevant electrodes and discarded non-informative ones.
• The proposed pattern recognition models achieved a mean recognition rate of up to 90.2%.

Conclusions:

• The developed methodology offers an efficient approach to EEG-based emotion recognition.
• Reduced data processing and optimized algorithms significantly lower computational complexity.
• The findings suggest a viable path for practical, low-resource emotion recognition systems.