Optimizing 1D-CNN-Based Emotion Recognition Process through Channel and Feature Selection from EEG Signals.

Area of Science:

• Affective computing and human-computer interaction
• Biomedical signal processing and machine learning

Background:

• Electroencephalogram (EEG)-based emotion recognition is crucial for applications like mental health monitoring and personalized interventions.
• Existing methods often face challenges with data efficiency and model complexity, limiting real-time applications.

Purpose of the Study:

• To improve the efficiency and accuracy of deep learning models for EEG-based emotion recognition.
• To explore unique EEG channel and feature selection methods for optimized data processing.
• To develop a lightweight deep learning approach for real-time emotion classification.

Main Methods:

• Utilized one-dimensional convolutional neural networks (1D-CNN) for analyzing EEG signals and classifying emotional states.
• Implemented novel EEG channel and feature selection techniques to reduce data dimensionality and enhance model efficiency.
• Employed data augmentation strategies to increase dataset size and improve model robustness.

Main Results:

• Achieved high mean accuracies: 97.6% on MAHNOB-HCI, 95.3% on SEED, and 89.0% on DEAP datasets.
• The 1D-CNN model effectively captured intricate patterns for distinguishing emotional states (High Valence/Low Valence, High Arousal/Low Arousal).
• Optimized models demonstrated significant improvements in memory, processing time, and accuracy.

Conclusions:

• The proposed method offers a highly efficient and accurate approach for EEG-based emotion recognition.
• Results indicate strong potential for developing cost-effective IoT devices for real-time EEG data collection and emotion analysis.
• This research enhances the feasibility and applicability of emotion-aware systems in various real-world scenarios.