1 problem

Let $f\colon \mathbb{R}^n \to \mathbb{R}$ be the output of a fully connected neural network with $\ell$ hidden layers of widths $n_1,\ldots,n_\ell$ and a Pfaffian activation functi…