Answers are grounded in this paper and cite their sources.
Estimates of the generalization error are proved for a residual neural network with L random Fourier features layers $ z₊₁= z_ + Re∑ ₖ₌₁^K bₖ\,e^{iω ₖ z_ }+ Re∑ ₖ₌₁^K cₖ\,e^{iω ^{}ₖ· x}$. An optimal distribution for the frequencies $(ω ₖ,ω ^{}ₖ)$ of the random Fourier features $e^{iω ₖ z_ }$ and $e^{iω ^{}ₖ· x}$ is derived. This derivation is based on the corresponding generalization error for the approximation of the function values $f(x)$. The generalization error turns out to be smaller than the estimate ${\| f\|^2_{L^1({R}^d)}}/{(KL)}$ of the generalization error for random Fourier features, with one hidden layer and the same total number of nodes $KL$, in the case of the $L^∞ $-norm of $f$ is much less than the $L^1$-norm of its Fourier transform $ f$. This understanding of an optimal distribution for random features is used to construct a new training method for a deep residual network. Promising performance of the proposed new algorithm is demonstrated in computational experiments.
No takes yet. Share an insight, caveat, or question.
Kammonen et al. (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: