search

Altaheri / EEG-ATCNet

Attention temporal convolutional network for EEG-based motor imagery classification
https://ieeexplore.ieee.org/document/9852687
Apache License 2.0
173 stars 22 forks source link

I want to add a program #5

Closed hljuniversity closed 1 year ago

hljuniversity commented 1 year ago

Hello, thank you for sharing the code，I added a piece of de-noising code to the ATCNet model, which reduced the accuracy of the verification set. I don't understand why. Do you understand?

Altaheri commented 1 year ago

Hi, It depends on which method you used. Some key features may be lost during the denoising process. In deep learning (DL), noise may not be an issue. DL can learn very well from noisy data. Also, adding noise to the EEG data is one way to augment the data and help the model to generalize well. I haven't worked with noise removal in the EEG data; However, you may refer to the following review article which discusses the effect of pre-processing on MI-EEG data (including denoising) for DL-based techniques.

https://link.springer.com/article/10.1007/s00521-021-06352-5