Hello Dani,
I understand you are facing difficulty in understanding the assumptions behind AMICA and binica. Both AMICA and binica assume non-Gaussian ICs, and neither assumes Gaussian ICs.
Here's a simplified breakdown:
- AMICA (Adaptive Mixture ICA): AMICA does assume that the independent components (ICs) in the signal are non-Gaussian. It's designed to handle non-Gaussian ICs and is more computationally demanding due to its adaptive nature. For more information on AMICA, you can refer to the following link: https://github.com/sccn/amica/wiki
- binica (binary ICA): Contrary to what you've heard, binica also assumes that the ICs in the signal are non-Gaussian, similar to AMICA. It's called "binary" because it uses a binary approach to separate the ICs. For additional details on binica, check out the following link: https://github.com/sccn/binica/wiki
For more functions and examples, consider referring to the EEGLAB documentation, which provides additional insights and practical examples for using ICA algorithms in the context of electrophysiological data analysis using the below link.
I hope this clears your doubt.
Best Regards,
Asim Asrar
