Related Experiment Video
Updated: Jun 28, 2026
09:32
Cortical Source Analysis of High-Density EEG Recordings in Children
Published on: June 30, 2014
21.3K
REGEEG: A Regression-Based EEG Signal Processing in Emotion Recognition.
IEEE Journal of Biomedical and Health Informatics
|March 10, 2025
Summary
Electroencephalograms (EEG) show promise for AI emotion recognition. A novel REGEEG method with K-Nearest Neighbors achieved over 95% accuracy in classifying emotions during gameplay.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Machine Learning
Background:
- Electroencephalograms (EEG) are crucial for non-invasive study of emotion recognition.
- Developing AI models to understand human behavior and decision-making is a key research area.
Purpose of the Study:
- To develop an accurate emotion recognition model using EEG signals during gameplay.
- To test various machine learning classification kernels for optimal performance.
Main Methods:
- Utilized the publicly available GAMEEMO database for EEG-based emotion recognition.
- Developed a novel signal processing method called Regression EEG (REGEEG) with an electrode pairing selector.
- Evaluated 28 machine learning kernels, including K-Nearest Neighbors (k-NN), using statistical and polynomial feature extraction.
Main Results:
- Five kernels achieved over 80% classification performance.
- The K-Nearest Neighbors (k-NN) model exceeded 95% accuracy, F1-Score, and kappa-score.
- REGEEG demonstrated robust performance across 30-fold Cross-Validation and Leave-one Subject-out techniques.
Conclusions:
- The REGEEG method and EEG electrode pair channel selection are effective for emotion recognition.
- The study highlights the potential of EEG-based AI for understanding human emotions during interactive tasks.
