Sho Nakagome, Trieu Phat Luu, Yongtian He, Akshay Sujatha Ravindran & Jose L. Contreras-Vidal |

- 2020-07-22

- 18 M

**1**,**0**,**0**

Pre-processing pipelines for different offline experiments

are represented in Fig. 1. The base pipeline is selected such that they can easily be used in an online real-time

decoding scheme18. An H-infinity algorithm was used to specifically remove eye blinks, eye motions, amplitude

drifts and recording biases simultaneously23. The parameters of the H-infinity algorithms were kept the same as

the real-time decoding. Peripheral channels were removed as they typically contain many artifactual components.

The signals were then bandpass filtered using a 4th order butterworth filter. Although the frequency range was the

same, this is one of the differences compared to the real-time decoding as the real-time implementation utilized

finite impulse filter and the phase shift was expected. To this point, all processing was done through a MATLAB

script, which is also provided in the open-sourced repository. Additionally, before each experiment, the signals

were z-scored for each channel

1 : MATLAB ⺻

1. MATLAB ⺻ 003

1.1 MATLAB ϱ 003

â(command Window) Է 005

(Help) ̿ 007

1.2 Է 008

009

MATLAB ϱ 009

1.3 Ҵ 009

켱 011

Լ 012

1.4 ǥ 013

1.5 ó 015

̸ 015

clear ɾ 016

Ư 017

whos ɾ 017

1.6 Ϳ 018

018

023

ũ ° 024

1.7 (Random) Ҽ 025

025

Ҽ 027

1.8 ȣ ̿ 028

ȣĿ ġȯ 029

1.9 ڵ 030

ũƮ ڵ 030

ڸƮ ߰ 032

Լ ڵ 033

Լ 036

1.10 037

ezplot ̿ 037

plot ̿ 039

3 042

1.11 MATLAB (Excel) 043

ҷ 043

ɼ 046

ũƮ ɼ 049

Լ ɼ 049

Ϸ ϱ 050

2 :

An empirical comparison of neural networks and machine learning

algorithms for EEG gait decoding

1. Introduction 51

2. Methods 52

3. Results 56

4. Discussion 64

5. References 66