Hi,
I'm currently trying to run cudaica after using pca to reduce the number of dimensions using the following process:
- remove mean
- define number of components explaining 90% of the variance
- whiten
- ica using whitened signal
My problem now comes with trying to compute the projections from the output, and then how to reconstruct the data with the original numbers of channels.
Dear friend
Once you have the components of the original data points, you can plot the data using n-dimensional space plots.
That will give you a clear picture.
Cheers
Hi,
I am not sure if applying ICA on the results out of PCA to be a good idea.
PCA guarantees that the resulted axes are orthogonal with respect to one another (statistically independent). So, how are you planning to apply ICA on the resultant?
Cheers
Peyman
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue