EEG-Based Classification of Imagined Fists Movements using Machine Learning and Wavelet Transform Analysis

ALI M., BANIYOUNES and EMAD, A. AWADA and MOHAMMAD, H. ALOMARI (2014) EEG-Based Classification of Imagined Fists Movements using Machine Learning and Wavelet Transform Analysis. In: International Conference on Advances in Computing, Communication and Information Technology CCIT 2014, 01 - 02 June,2014, London, UK.

[img]
Preview
Text
20140709_122719.pdf - Published Version

Download (745kB) | Preview
Official URL: https://www.seekdl.org/conferences/paper/details/3...

Abstract

Electroencephalography (EEG) signals represent the brain activity by the electrical voltage fluctuations along the scalp. In this paper, we propose a system that enables the differentiation between imagined left or right fist movements for the purpose of controlling computer applications via imagination of fist movements. EEG signals were filtered and processed using a hybrid system that uses wavelet transform analysis and machine learning algorithms. Many Daubechies orthogonal wavelets were used to analyze the extracted events. Then, the Root Mean Square (RMS) and the Mean Absolute Value (MAV) were calculated to the wavelet coefficients in two detail levels. Support Vector Machines (SVMs) and Neural Networks (NNs) were applied to the feature vectors and optimized by carrying out an intensive learning and testing experiments. Optimum classification performances of 84.5% and 82.1% were obtained with SVMs and NNs, respectively. Compared with the related research work reported in the literature, our system showed a good performance for the classification of fists movements, which enables the control of many computer applications via imagination.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: BCI, Data Mining, Machine Learning, SVMs, NNs, DWT, Feature Extraction
Depositing User: Mr. John Steve
Date Deposited: 18 May 2019 12:20
Last Modified: 18 May 2019 12:20
URI: http://publications.theired.org/id/eprint/2462

Actions (login required)

View Item View Item