重度運動障害者向けEEG信号の色分類
Color-based classification of EEG Signals for people with the severe locomotive disorder
単一電極EEGから注意機構付きLSTMで色を分類し、2色で93.5%、4色で65.75%の精度を達成した。
著者: Ankit Shrestha, Bikram Adhikari
分類: eess.SP, cs.HC, cs.LG, cs.RO, q-bio.NC
原文アブストラクト
The neurons in the brain produces electric signals and a collective firing of these electric signals gives rise to brainwaves. These brainwave signals are captured using EEG (Electroencephalogram) devices as micro voltages. These sequence of signals captured by EEG sensors have embedded features in them that can be used for classification. The signals can be used as an alternative input for people suffering from severe locomotive disorder.Classification of different colors can be mapped for many functions like directional movement. In this paper, raw EEG signals from NeuroSky Mindwave headset (a single electrode EEG sensor) have been classified with an attention based Deep Learning Network. Attention based LSTM Networks have been implemented for classification of two different colors and four different colors. An accuracy of 93.5\% was obtained for classification of two colors and an accuracy of 65.75\% was obtained for classifcation of four signals using the mentioned attention based LSTM network.
