6 ms·
With the universal approximation theorem, we can use artificial neutral networks with ReLUs to accurately approximate a function. But we just get weights out of
by knicholes 3y ago
With the universal approximation theorem, we can use artificial neutral networks with ReLUs to accurately approximate a function. But we just get weights out of it at the end. This approach feels similar, but provides code in the end.