Questions about the effect of pretraining.

[NorbertZheng]

I'm not sure about the effect of EEG pretraining. I have directly train a shallow convolution network on eeg_5_95_std.pth as used in your paper. And the test accuracy according to max validation is about 60% on average, which is better than the classification accuracy 45% mentioned in your paper.

And a parallel work NeuroImagen achieves better results without pretraining. So I wonder whether EEG pretraining is indeed meaningful???

[DILIU1]

好兄弟，可以加一个微信吗？我也在做这方面的工作，希望多多交流。

[zihang-xu]

I guess NeuroImagen may not use the same data, because the paper did not mention 5-95 hz filtering

[RomGai]

I guess NeuroImagen may not use the same data, because the paper did not mention 5-95 hz filtering

I think that's true. Without 5-95 hz filtering, this dataset is unreliable.

[TwoHuang]

好兄弟，可以加一个微信吗？我也在做这方面的工作，希望多多交流。

好兄弟,可以加个微信吗,我也开始做这方面工作了