EEG-based stress recognition using deep learning techniques

• Main Author: Syabil Kwajah
• Other Authors: Arokiaswami Alphones
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Simulation and modeling; Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/157417
• Institution: Nanyang Technological University

Stress recognition using Electroencephalogram (EEG) based signal is a promising area of study. There had been multiple studies about stress recognition however, the aim of this project was to implement Deep Learning for stress detection. As EEG is the least intrusive data collecting method and experiment can be conducted in a controlled environment, more studies are discovering the potential EEG has. Being a relatively new field, Deep Learning has a broad spectrum of possibilities to be explored to improve the efficiency as well as accuracy for stress detection. An optimum algorithm has yet to be identified for Stress Recognition with Deep Learning Techniques. The aim of this paper is to implement a well-known Neural Network called Convolutional Neural Network (CNN) along with Recurrent Neural Network (RNN) called R-CNN. CNN handling spatial information and the RNN handling temporal information. The RNN used will be the LSTM architecture. Statistical Features will be extracted during the feature extraction portion before moving on to the deep learning techniques. As a baseline comparison, a paper with similar parameters was used with a different classifier. The accuracy obtained from the paper was 67.08% using K-Nearest Neighbour (KNN) with 2 classes. Dataset used from both paper is from a Dataset for Emotion Analysis using EEG, Physiological and video signals (DEAP). With the completion of this paper, hopefully more insights and research will be done within the ever-growing field.