View article

Emotions encompass a wide range of feelings, thoughts, and behaviors, reflecting the complex output of the human brain. This interdisciplinary field draws from computer science, AI, neurology, healthcare, and more to study emotional experiences. Understanding and labeling one's emotions are crucial for mental health and well-being, especially in managing stress-related conditions. Emotion classification using electroencephalogram (EEG) signals has gained interest, particularly in affective computing. Developing an effective brain-computer interface (BCI) system for emotion recognition through EEG involves key components such as feature extraction and classifier selection. Deep learning methods, known for their superior performance, have recently garnered significant attention in this domain. Our paper introduces the Deep CNN for Emotion Recognition (DCNNER) framework, utilizing deep convolutional neural networks to accurately detect human emotions from EEG signals. To enhance the model's efficiency, we employ principal component analysis (PCA) for feature extraction (FE) and dimensionality reduction. By feeding only the selected features to various classifiers, we compare their performances on pre-processed and PCA-applied data. The proposed system outperforms existing approaches, achieving a remarkable model accuracy of 99% and a model loss of 0.3. The model employs a 3-dimensional representation, encompassing valence, arousal, and dominance for emotion detection. Our research showcases the potential of deep learning in EEG-based emotion recognition, promising advancements in affective computing and …