4 ms·
Does anybody know if this is true for other machine learning techniques?
by pesenti 12y ago
Does anybody know if this is true for other machine learning techniques?
- mturmon 12y agoThe same arguments as in the original Cybenko paper, or the Stone-Weierstrass theorem, lend support to the idea that SVMs are universal approximators (with most typical kernels). This has been proven by a couple of authors. I'm not aware of universal approximation results for random forests, but since they have the same general construction, this would not be surprising.