Header menu link for other important links
X
Single trial P300 classification using convolutional LSTM and deep learning ensembles method
, Joshi R., Goel P.,
Published in Springer Verlag
2018
Volume: 11278 LNCS
   
Pages: 3 - 15
Abstract
The odd ball paradigm is a commonly used approach to develop Brain Computer Interfaces (BCIs). EEG signals have shown to elicit a positive deflection known as the P300 event related potential during odd ball experiments. BCIs based on these experiments rely on detection of the P300 potential. EEG signals are noisy, and therefore P300 detection is performed on an average of multiple trials, thus making them inappropriate for BCI applications. We propose a neural network model based on Convolutional Long Short Term Memory (ConvLSTM) for single trial P300 classification. EEG data encodes both spatial and temporal information using multiple EEG sensors. Convolutional neural networks (CNNs) have been known to capture spatial information whereas LSTMs are known to capture temporal information. Our experiments show that the proposed method outperforms previous CNN based approaches on raw EEG signals. The approaches were evaluated on publicly available dataset II of BCI competition III. Another dataset was recorded locally using audio beeps as stimuli to validate these approaches. The ensemble models based on CNNs and ConvLSTM are also proposed. These models perform better than individual architectures. © Springer Nature Switzerland AG 2018.
About the journal
JournalData powered by TypesetLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherData powered by TypesetSpringer Verlag
ISSN03029743
Open AccessNo