Spectral Blind Source Separation Using Recurrent Neural Network #11
Replies: 2 comments 15 replies
|
I would be curious to see how this performs with different referencing methods (CAR, REST, Surface Laplacian). I could help with that part. |
|
Great! Thank you so much for offering to help. I will be interested to learn how different referencing methods can influence EEG power spectra. I will try applying spectral BSS to the mne-python ssvep dataset, and get in touch with you to share some results with code that can be modified to implement different referencing methods. |
|
Hi @Ajarn-Jamie, On Feb 13 at 11 am EST, we will have an online meeting for an interim report. Would you be willing to present your idea in a 10-min slot? Please let me know. |
|
I have just shared a set of scripts with you through github - the first script prepares the data for training RNNs, so this would be where to re-reference the data (see comment). I'm using python 3.10.5. Let me know how it goes and if you have any questions |
|
@amisepa how's it going? Did you get a chance to try re-referencing? |
|
Hi @Ajarn-Jamie! |
|
@Ajarn-Jamie Hi! The invitation you sent expired, sorry for being so slow. Can you send again and I'll give a try soon? |
|
Hi @amisepa OK, no worries - I just resent the invitation. Please let me know how it goes |
Uh oh!
There was an error while loading. Please reload this page.
Hi all,
I am thinking of starting a subproject related to frequency-based blind source separation (BSS) using recurrent neural network (RNN). The rationale is that applying spectral BSS before one-over-f/FOOOF analyses could reduce the impact of source signal mixing through volume conduction and improve analyses.
If you would like to get involved with this subproject, please get in touch to share how you wish to contribute.
Also, please feel free to share any comments/suggestions you have regarding this potential subproject.
All the best,
Jamie
All reactions