Neural Networks Training Based on Sequential Extended Kalman Filtering for Single Trial EEG Classification

2010 ◽  
Author(s):  
Arjon Turnip ◽  
Keum-Shik Hong ◽  
Shuzhi Sam Ge ◽  
Myung Yung Jeong
Keyword(s):  
Neural Networks ◽  
Kalman Filtering ◽  
Single Trial ◽  
Extended Kalman Filtering ◽  
Eeg Classification

Robust soft sensors for SBR monitoring

2001 ◽  
Vol 43 (3) ◽  
pp. 101-105 ◽  
Author(s):  
D. Zyngier ◽  
O. Q. Araújo ◽  
M. A. Coelho ◽  
E. L. Lima
Keyword(s):  
Neural Networks ◽  
Wastewater Treatment ◽  
Kalman Filtering ◽  
Ammonium Ion ◽  
Carbonaceous Matter ◽  
Process Variables ◽  
Extended Kalman Filtering ◽  
Soft Sensors ◽  
Nitrate Ion

In wastewater treatment some process variables may be very difficult to measure directly because of the non-availability or excessive cost of sensors. An alternative is to use ‘soft sensors” that provide online estimates of these inacessible variables by calculations based on auxiliary measurable variables. Two such sensors are proposed based on extended Kalman filtering or neural networks that could enable this monitoring of nitrate ion, ammonium ion and carbonaceous matter concentrations during nitrification of wastewater.

scholarly journals Decoding Visual Motions from EEG Using Attention-Based RNN

2020 ◽  
Vol 10 (16) ◽  
pp. 5662
Author(s):  
Dongxu Yang ◽  
Yadong Liu ◽  
Zongtan Zhou ◽  
Yang Yu ◽  
Xinbin Liang
Keyword(s):  
Neural Networks ◽  
Visual Motion ◽  
Data Augmentation ◽  
Training Data ◽  
Single Trial ◽  
Online Data ◽  
Classification Rate ◽  
Visual Motion Perception ◽  
Eeg Classification ◽  
Feature Representations

The main objective of this paper is to use deep neural networks to decode the electroencephalography (EEG) signals evoked when individuals perceive four types of motion stimuli (contraction, expansion, rotation, and translation). Methods for single-trial and multi-trial EEG classification are both investigated in this study. Attention mechanisms and a variant of recurrent neural networks (RNNs) are incorporated as the decoding model. Attention mechanisms emphasize task-related responses and reduce redundant information of EEG, whereas RNN learns feature representations for classification from the processed EEG data. To promote generalization of the decoding model, a novel online data augmentation method that randomly averages EEG sequences to generate artificial signals is proposed for single-trial EEG. For our dataset, the data augmentation method improves the accuracy of our model (based on RNN) and two benchmark models (based on convolutional neural networks) by 5.60%, 3.92%, and 3.02%, respectively. The attention-based RNN reaches mean accuracies of 67.18% for single-trial EEG decoding with data augmentation. When performing multi-trial EEG classification, the amount of training data decreases linearly after averaging, which may result in poor generalization. To address this deficiency, we devised three schemes to randomly combine data for network training. Accordingly, the results indicate that the proposed strategies effectively prevent overfitting and improve the correct classification rate compared with averaging EEG fixedly (by up to 19.20%). The highest accuracy of the three strategies for multi-trial EEG classification achieves 82.92%. The decoding performance for the methods proposed in this work indicates they have application potential in the brain–computer interface (BCI) system based on visual motion perception.

scholarly journals Interpretable deep neural networks for single-trial EEG classification

2016 ◽  
Vol 274 ◽  
pp. 141-145 ◽  
Author(s):  
Irene Sturm ◽  
Sebastian Lapuschkin ◽  
Wojciech Samek ◽  
Klaus-Robert Müller
Keyword(s):  
Neural Networks ◽  
Deep Neural Networks ◽  
Single Trial ◽  
Eeg Classification
Sign in / Sign up

Export Citation Format

Share Document