recommendations for machine learning, mental learning, and reinforcement learning ？

YuminosukeSato

What are the recommended machine learning, mental learning, and reinforcement learning for classifying brain waves?
What are the recommendations for identifying the 7 types of EEG for cVEP, SSVEP, or MI images?

wjcroft

There is an MI tutorial here,

https://docs.openbci.com/Examples/EEGProjects/MotorImagery/

And we've already discussed MindAffect, which cVEP based, much superior to plain SSVEP.

There are past threads on ML subjects:

https://www.google.com/search?as_q=ml&as_sitesearch=openbci.com

Brainflow has some pre-trained classifiers:

https://brainflow.readthedocs.io/en/stable/Examples.html#python-eeg-metrics

William

YuminosukeSato

If brain waves are acquired using cVEP, what should MNE be used for preprocessing? Do you use CSP? Or is it better not to use MN?

wjcroft

The entire MindAffect package is self-contained and does not need MNE or other BCI platforms. Each cVEP system is different, MindAffect uses their own approach. Other cVEP based systems may use a different approach.

'cVEP' is code-based VEP. It sends unique pseudo-random on-off flash sequences for each unique on-screen button. Then in the MindAffect VEP analysis code, it de-noises the EEG and attempts to match which unique pattern (per button) is being looked at.

YuminosukeSato

Does it mean that brain waves are also fetched with BCI of mindaffect? Or does it mean that brain waves are taken with openBCI?

wjcroft

I thought you had done some of the steps of the MindAffect documentation / tutorial??

https://mindaffect-bci.readthedocs.io/en/latest/

Is that not the case? Their initial release used Ganglion, and fetching the data with Brainflow. But it should work with Cyton as well.

https://mindaffect-bci.readthedocs.io/en/latest/supported_hardware.html

re: "...brain waves are taken with OpenBCI?" The GUI is not being used, only the MindAffect Python code speaking with Brainflow library.