Paper Info
Title
Exploring the Neural Pattern in EEG Records of Material-specific Memory Maintenance by Deep Learning.
Abstract
In this paper, a classification performance comparison among several machine learning algorithms is presented, utilizing different frequency features (theta, alpha, low beta, high beta, low gamma and high gamma) which are extracted from material-specific visual memory maintenance EEG records of 20 volunteers. Decision tree, Support Vector Machine(SVM), Convolutional Neural Network(CNN) and Deep Forest (DF) are used to classify. As shown in our results, deep forest improved the classification accuracy by 9.75% in low beta (14- 20Hz) while distinguishing Face from House and 1.25% with 50% standard deviation reduction in alpha (8-12Hz) while distinguishing Digit from Letter. Those results proved that deep learning has certain potential advantage over traditional machine learning in small-sample EEG analysis. Low beta and alpha band are found to be important features in material-specific memory maintenance in graph-specific and alphanumeric-specific conditions respectively.
Year
DOI
Venue
2018
10.1109/FSKD.2018.8687187
ICNC-FSKD
Field
DocType
Citations 
Decision tree,Pattern recognition,Convolutional neural network,Computer science,Support vector machine,Visual memory,Artificial intelligence,Beta (finance),Deep learning,Standard deviation,Machine learning,Electroencephalography
Conference
0
PageRank 
References 
Authors
0.34
0
6
Name
Order
Citations
PageRank
Zige Wang100.34
Hao Zhang220364.03
te hu322.07
Zifan Chen400.68
Steffen Gais500.34
Xiaoliang Gong600.68