Emotion Classification of EEG signals using Logistic Regression classification

• Published in: 2024 3rd International Conference for Innovation in Technology (INOCON) pp. 1 - 5
• Main Authors: Singh, Anmol Rattan, Singh, Gurjinder, Saluja, Nitin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.03.2024
• Subjects: Brain; Brain modeling; Deep Analysis; EEG signals; Feature extraction; Human computer interaction; Logistic regression; Logistics regression (LR); Machine learning; Medical services; Neuroscience; Python
• DOI: 10.1109/INOCON60754.2024.10511417

The exploration of emotion classification through the analysis of EEG signals presents a multifaceted research area that merges neuroscience, psychology, and machine learning. In this project, our focus shifts to using Logistic Regression (LR), another potent technique in machine learning, known for its proficiency in dealing with binary and multiclass classification problems and LR estimates probabilities using a logistic function, which is particularly effective for categorizing data into distinct groups. This project involves analyzing electroencephalogram (EEG) signals, the intricate reflections of the brain's activities, influenced by various emotional states. To utilize LR, the initial steps mirror those in SVM-based approaches. We collect EEG data from subjects experiencing a range of emotions, induced through different stimuli such as images or audio clips. This data is inherently complex and multi-dimensional, necessitating thorough preprocessing to isolate pertinent features. Once the features are extracted, they are input into the LR model. This model is adept at handling binary classifications (e.g., happy vs. not happy) and can be extended to multiclass problems (e.g., distinguishing between happiness, sadness, and stress) using techniques like one-vs-rest (OvR) or multinomial logistic regression. The model learns to associate specific EEG patterns with particular emotional states, assigning probabilities to these classifications. The implications of this research are profound, especially in the realms of human-computer interaction and healthcare. In technology, it could lead to AI systems that better understand and respond to human emotions, enhancing user experience. In healthcare, it offers a novel approach to diagnosing and managing emotional disorders, providing insights into the brain's response to different emotional states. This project, therefore, is not only a step forward in machine learning applications but also a significant contribution to our understanding of the complex interplay between the brain and emotions.