A Novel EEG-Based Emotion Recognition Framework Using DQN-DenseNet for Real-Time Applications

Emotion recognition using EEG signals has gained significant attention due to its potential applications in mental health diagnostics and human-computer interaction. However, existing algorithms often struggle with low accuracy and high computational time. This study introduces the DQN-DenseN et algorithm, which combines the sequential data processing capability of Deep Q-Learning Networks (DQN) with the efficient feature propagation of DenseNet. By leveraging EEG spectrograms processed with Fast Fourier Transform (FFT), the algorithm extracts and classifies features with enhanced accuracy. Experiments using the SEED dataset demonstrate that the proposed approach achieves a mean recognition accuracy of 99.04%, outperforming traditional CNN, DQN, and DenseNet models. The DQN-DenseNet algorithm offers improved adaptability and efficiency, paving the way for future advancements in EEG-based emotion recognition across dynamic real-world applications.