Hi all,
Wondering if anyone has any previous experience with augmenting connectomes to increase dataset size. It seems that the literature on this is quite sparse and people seem to choose their methods adhoc.
The following paper is one of the only ones available, and performs a newtonian blurring approach:
Another approach we’ve though of is performing SMOTE on the vectorized lower triangular section of the connectome.
If anyone has any more literature or prior experience it would be greatly appreciated.
Many thanks,
Chris