4 ms·
I think one of the caveats was that the function should be relatively smooth and continuous, a mandelbrot set isn't very smooth is it?
by 3v3rt 11y ago
I think one of the caveats was that the function should be relatively smooth and continuous, a mandelbrot set isn't very smooth is it?
- fnbr 11y agoSmoothness isn't necessary to prove the result, just continuity. You do need smoothness to prove a bound on the rate of convergence of the basis representation, and given that the boundary of the Mandelbrot doesn't have a closed form representation (as far as I'm aware), I think that the convergence of the neural network representation would be extremely slow.