Not to be too glib, but as a non-ML person with decent numerics ability:
... did this post just spend a ton of boring math explaining that a neural network has the ability to detect whether an image is bumpy or smooth?
I mean, Fourier analysis has a very straightforward "intuitive" interpretation. That's the point of its value in applied domains, after all. Straightforward intuitive things are "neural" by their nature, no?
... did this post just spend a ton of boring math explaining that a neural network has the ability to detect whether an image is bumpy or smooth?
I mean, Fourier analysis has a very straightforward "intuitive" interpretation. That's the point of its value in applied domains, after all. Straightforward intuitive things are "neural" by their nature, no?