Related Experiment Video
Updated: Dec 11, 2025
08:31
Conscious and Non-conscious Representations of Emotional Faces in Asperger's Syndrome
Published on: July 31, 2016
14.2K
An EEG Database and Its Initial Benchmark Emotion Classification Performance
Ayan Seal1,2, Puthi Prem Nivesh Reddy1, Pingali Chaithanya1
1PDPM Indian Institute of Information Technology, Design and Manufacturing, Jabalpur 482005, India.
Computational and Mathematical Methods in Medicine
|August 25, 2020
Summary
This study introduces a new database of electroencephalogram (EEG) signals from 44 volunteers to recognize human emotions. Using discrete wavelet transform and extreme learning machine (ELM), the method achieved 94.72% accuracy in classifying emotions.
Area of Science:
- Neuroscience
- Affective Computing
- Biomedical Engineering
Background:
- Human emotion recognition is crucial for academic and industrial applications.
- Current methods primarily use facial image analysis, limiting real emotion capture.
- A scarcity of publicly available electroencephalogram (EEG) signal databases hinders affective computing research.
Purpose of the Study:
- To present a novel database of EEG signals for emotion recognition.
- To establish an initial benchmark for EEG-based emotion classification.
- To make the database publicly available for future research.
Main Methods:
- Recorded 32-channel EEG data from 44 volunteers experiencing four emotional states (happy, fear, sad, neutral) induced by videos.
- Employed discrete wavelet transform for feature extraction and extreme learning machine (ELM) for classification.
- Utilized ELM for automated channel and subband selection.
Main Results:
- Achieved a peak classification accuracy of 94.72% for emotion recognition.
- Optimal performance was observed when features were extracted from the gamma subband of the FP1-F7 channel.
- The developed EEG database comprises data from 44 participants, with 23 females.
Conclusions:
- The presented EEG database and classification methodology offer a valuable resource for affective computing.
- EEG-based emotion recognition demonstrates high potential, surpassing traditional facial analysis in capturing genuine emotional states.
- The study highlights the effectiveness of discrete wavelet transform and ELM for robust emotion classification from EEG signals.
