Non-Asymptotic Performance Guarantees for Neural Estimation of $\mathsf{f}$-Divergences

Open in new window