3 ms·
Isn't this similar to infinite monkey theorem: "Given an infinite length of time, a chimpanzee punching at random on a typewriter would almost surely type out a
by bluesign 11y ago
Isn't this similar to infinite monkey theorem: "Given an infinite length of time, a chimpanzee punching at random on a typewriter would almost surely type out all of Shakespeare's plays."
Given that we can have infinite big neural network and train it in an infinite length of time, it can compute any function.
- tfgg 11y agoI don't think so, since the former is about the occurrence of sequences in stochastic, ergodic processes, and the latter is about approximating deterministic functions. It's more like given an infinite number of parameters, you have enough degrees of freedom to describe the vector space a function lives in with some set of basis functions, and a neural net supplies that.