A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Data Science and Artificial Intelligence, School of Engineering and Technology

Thesis (M. Eng.) - Asian Institute of Technology, 2023

The classification of motor imagery through electroencephalogram signals is a significant area of research that has been thoroughly explored in the domain of brain-computer interfaces (BCIs). EEG-based classification often faces the issue of overfitting due to the scarcity of data. Data augmentation techniques have been proposed as a solution to address the issue by increasing the size of the training data set. This research paper presents a novel approach for enhancing motor imagery classification in EEG signals through the application of diffu sion models as a data augmentation technique. The utilization of diffusion models involves the introduction of Gaussian noise to the initial EEG signals, resulting in the production of novel samples. The proposed method is evaluated on a publicly available EEG dataset for the purpose of motor imagery classification. A comparison is made between this method and various other state-of-the-art data augmentation techniques. The study{u2019}s findings indicate that the proposed method exhibits superior performance in classification accuracy compared to alternative methods. These results suggest that the proposed method has the potential to serve as a viable data augmentation technique for classifying EEG-based motor imagery.