search

KAZABANA / EEGMatch

MIT License
12 stars 2 forks source link

About Dataset Processing #2

Closed upeng10 closed 2 months ago

upeng10 commented 2 months ago

There is a line of code like this path='F:\zhourushuang\transfer_learning\feature_for_net_session'+str(session)+'_LDS_de' Could you explain how the file feature_for_net_session'+str(session)+'_LDS_de' is constructed? The seed and seed iv datasets do not provide this file directly.

KAZABANA commented 2 months ago

'F:\zhourushuang\transfer_learning\feature_for_net_session'+str(session)+'_LDS_de' is not a file. It is a folder containing LDS features from all the subjects within SEED or SEED-IV. For example, for the SEED dataset, Session 1, the folder might have 15 files. Each file contains the LDS features for each subject (a 3394 X 310 matrix, time X feature dim). For example, for subject 1 (session 1), you can construct the file by concatenating the LDS feature (a 62 X 5 -dim vector) from different trials.

upeng10 commented 2 months ago

Thank you for your answer. I tried it, but there may still be some minor problems and it didn't work. Could you please send me the files of this part directly? I'm very sorry for the trouble.

KAZABANA commented 2 months ago

We have uploaded a new script (data_prepare_seed.m) for the LDS-DE dataset construction (SEED). You can prepare the dataset for SEED-IV using a similar pipeline. Thanks

upeng10 commented 2 months ago

Thanks a lot for your work and I wish you good luck in your research!!!