DEEMD-SPP: A Novel Framework for Emotion Recognition Based on EEG Signals.

Area of Science:

• Neuroscience
• Affective Computing
• Signal Processing

Background:

• Electroencephalography (EEG) is crucial for studying brain activity, cognitive diseases, and affective disorders.
• Accurate emotion recognition from EEG signals is a key challenge in affective computing.
• Existing methods like ensemble empirical mode decomposition (EEMD) face issues with residual noise and feature dimension unification.

Purpose of the Study:

• To propose a novel framework, DEEMD-SPP, for improved emotion recognition from EEG signals.
• To address the limitations of EEMD in noise reduction and feature selection.
• To develop a robust method for unifying feature dimensions in EEG-based emotion recognition.

Main Methods:

• Denoising Ensemble Empirical Mode Decomposition (DEEMD) was developed to decompose EEG signals, eliminate noise, and select relevant intrinsic mode functions (IMFs).
• Time-domain and frequency-domain features were extracted from the selected IMFs.
• Spatial Pyramid Pooling Network (SPP-Net) was utilized as a classifier to handle variable-sized feature maps and output fixed-size vectors.

Main Results:

• The DEEMD-SPP framework effectively reduced the impact of white noise in EEG signals.
• Accurate extraction of relevant EEG features was achieved through the proposed decomposition and selection process.
• Significant improvements in the performance of emotion recognition were demonstrated.

Conclusions:

• The DEEMD-SPP framework offers an effective solution for noise reduction and feature extraction in EEG signal processing.
• This approach enhances the accuracy and effectiveness of emotion recognition from EEG data.
• The study highlights the potential of combining DEEMD and SPP-Net for advanced affective computing applications.