Pattern classification of time-series EEG signals using neural networks

Author

Fukuda, Osamu ; Tsuji, Toshio ; Kaneko, Makoto

Author_Institution

Fac. of Eng., Hiroshima Univ., Japan

fYear

1996

fDate

11-14 Nov 1996

Firstpage

217

Lastpage

222

Abstract

This paper proposes a pattern classification method of time-series EEG signals using neural networks. To achieve successful classification for non-stationary EEG signals, a new network structure that combines a probabilistic neural network and recurrent neural filters is used. This network is suitable for expressing statistical and time-varying characteristics of time-series EEG signals. In the experiments, two types of photic stimulation caused by eye opening/closing and by artificial light are used to measure the EEG data. It is shown that the proposed network can achieve high classification performance

Keywords

electroencephalography; learning (artificial intelligence); pattern classification; recurrent neural nets; time series; artificial light; eye opening/closing; neural networks; nonstationary EEG signals; pattern classification; probabilistic neural network; recurrent neural filters; time-series EEG signals; Back; Brain modeling; Electroencephalography; Filters; Medical diagnostic imaging; Neural networks; Pattern classification; Recurrent neural networks; Training data; Virtual reality;

fLanguage

English

Publisher

ieee

Conference_Titel

Robot and Human Communication, 1996., 5th IEEE International Workshop on

Conference_Location

Tsukuba

Print_ISBN

0-7803-3253-9

Type

conf

DOI

10.1109/ROMAN.1996.568822

Filename

568822