Brain-computer interface based on machine learning of the EEG signals

• المؤلف الرئيسي: May Pwinnt Kyaw Thet
• مؤلفون آخرون: Jiang Xudong
• التنسيق: Final Year Project
• اللغة: English
• منشور في: Nanyang Technological University 2020
• الموضوعات: Engineering::Electrical and electronic engineering
• الوصول للمادة أونلاين: https://hdl.handle.net/10356/140567
• المؤسسة: Nanyang Technological University

A brain-computer interface (BCI) translates the human's brain signals to give a second chance to neuromuscular disabled people to be able to communicate, interact and function again. It is a communication tool which uses the brain activities(EEGs) by converting them into machine-learning languages. Although BCI is a matured and famous technology, EEGs used in BCI are non-stationary rhythms as the signals with low spatial resolution and high signal-to-noise (SNR) ratio. The signals obtained can be overlapped with artifacts. In order to recognize the motor function related information embedded in the rhythms, a machine learning method was introduced in order to extract the wanted features. This project describes a generalized common spatial patterns (CSPs) algorithm and its extension Regularized common spatial patterns (RCSP) for EEG features extractions. The features extracted are then compared between CSP Feature Selection (FS) and CSP Feature Weighting and Regularization (FWR) to rate the performances. EEG data from Berlin BCI Competition III, data set IVa as well as data set IIa from BCI Competition IV are used for training and testing the algorithms. Results obtained from all 4 tests are compared and analysed to determine the performance of these feature extraction methods. The experimental results prove that FWR give the better performance , accuracy comparing to FS in many of the tasks. RCSP results also outperform in majority of the tasks than CSP.