Anthony on Nostr: Are you familiar with random kitchen sinks? This sounds like a similar idea: creating ...

Are you familiar with random kitchen sinks? This sounds like a similar idea: creating a low-rank approximation of a "kernel":
> Using tools from probability theory on Banach
spaces, we show that with high probability, a fixed target function in a Reproducing Kernel Hilbert Space can be approximated well in the L∞ and the L2 sense as a linear combination of a few randomly chosen basis functions. For the class of functions we consider, the approximation rate turns out to be the same as that obtain [sic] by choosing the basis optimally.

(From Rahimi, A and Recht, B (2007). *Uniform Approximation of Functions with Random Bases*)

I honestly wish ML people would do better connecting the dots in their own field so we're not left wondering what the relationships are among these techniques. EC is terrible about this too.