Motor imagery EEG signal classification using deep learning for brain computer interfaces

Download

index.pdf

Date

2017

Author

Rezaaei Tabar, Yousef

Metadata

Show full item record

Item Usage Stats

217
views

78
downloads

In this thesis we proposed a novel method for classification of Motor Imagery (MI) EEG signals based on deep learning. Convolutional Neural Networks (CNN) and Stacked Autoencoders (SAE) networks were investigated for MI EEG classification. A new form of input is introduced to combine time, frequency and location information extracted from EEG signal and also a new deep network combining CNN and SAE is proposed in this thesis. In the proposed network, the features that are extracted by CNN are classified through the deep SAE network. The results obtained on public datasets revealed that the proposed method provides better classification performance compared to other state of art approaches. Furthermore, four different experiments were conducted on totally 16 subjects to collect MI EEG signals in addition to public datasets. Two of these experiments were used to record motor imagery data and perform offline classification. Online feedback was provided in the other two sets of experiments, where EEG signal was classified in real time and appropriate visual feedback was presented to the subject during the experiment. Our proposed signal processing method was applied to the recorded data as well as public datasets and compared with present state of art methods.

Subject Keywords

Electrodiagnosis., Brain-computer interfaces., Imaging systems in medicine., Convolutional Neural Networks

URI

http://etd.lib.metu.edu.tr/upload/12621596/index.pdf
https://hdl.handle.net/11511/27015

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Cognitive Learner: An Ensemble Learning Architecture for Cognitive State Classification Moğultay, Hazal (2017-05-18) In this study, we propose an ensemble learning architecture called "Cognitive Learner", for classification of cognitive states from functional magnetic resonance imaging (fMRI). Proposed architecture consists of a two-layer hierarchy. In the first layer, called voxel layer, we model the connectivity among the voxel time series to represent the detailed information about the experiment. In the second layer, we cluster the voxel time series by using functional similarity measure, to partition the brain volume...
Neural networks with piecewise constant argument and impact activation Yılmaz, Enes; Akhmet, Marat; Department of Scientific Computing (2011) This dissertation addresses the new models in mathematical neuroscience: artificial neural networks, which have many similarities with the structure of human brain and the functions of cells by electronic circuits. The networks have been investigated due to their extensive applications in classification of patterns, associative memories, image processing, artificial intelligence, signal processing and optimization problems. These applications depend crucially on the dynamical behaviors of the networks. In t...
Representation of human brain by mesh networks Önal Ertuğrul, Itır; Yarman Vural, Fatoş Tunay; Department of Computer Engineering (2017) In this thesis, we propose novel representations to extract discriminative information in functional Magnetic Resonance Imaging (fMRI) data for cognitive state and gender classification. First, we model the local relationship among a set of fMRI time series within a neighborhood by considering temporal information obtained from all measurements in time series. The estimated local relationships, called Mesh Arc Descriptors (MADs), are employed to represent information in fMRI data. Second, we adapt encoding ...
Neural network based online estimation of maneuvering steady states and control limits Gürsoy, Gönenç; Yavrucuk, İlkay; Department of Aerospace Engineering (2010) This thesis concerns the design and development of neural network based predictive algorithms to predict approaching aircraft limits. Therefore, approximate dynamics of flight envelope parameters such as angle of attack and load factor are constructed using neural network augmented dynamic models. Then, constructed models are used to predict steady state responses. By inverting the models and solving for critical controls at the known envelope limits, critical control inputs are calculated as well. The perf...
Frequency invariant beamforming and its application to wideband direction of arrival estimation Babataş, Eren; Candan, Çağatay; Department of Electrical and Electronics Engineering (2008) In this thesis the direction of arrival estimation of wideband signals using frequency invariant beamforming method is examined. The difficulty with the direction of arrival estimation of wideband signals is that it is not possible to obtain a single covariance matrix valid for the whole frequency spectrum of the signal. There are various methods proposed in the literature to overcome this difficulty. The common aim of all the methods is to obtain a composite covariance matrix for the overall band of the si...

Citation Formats

Y. Rezaaei Tabar, “Motor imagery EEG signal classification using deep learning for brain computer interfaces,” Ph.D. - Doctoral Program, Middle East Technical University, 2017.