I was not able to find much on this topic besides this article, and don't know if this is even something worth considering, but it has been stuck with me for a while.
Since any given set of data is linearly separable in some dimension, could there be some basis where all physical constants (i.e. plancks constant, speed of light in a vacuum) could be represented as integers? If so, how would one go about tackling this problem (machine learning model?) and what applications would a good outcome have?