Comment on Every Model Learned by Gradient Descent Is Approximately a Kernel MachineparentComments−api5ySo a "neural network" is actually a type of parameterized mathematical function that can be fit to any curve including higher dimensional surfaces, etc.?−xksteven5yThat's correct. See https://en.wikipedia.org/wiki/Universal_approximation_theore... for more details−laingc5yYes.
Comments
So a "neural network" is actually a type of parameterized mathematical function that can be fit to any curve including higher dimensional surfaces, etc.?
That's correct. See https://en.wikipedia.org/wiki/Universal_approximation_theore... for more details
Yes.