8
38

Kernel Methods are Competitive for Operator Learning

Abstract

We present a general kernel-based framework for learning operators between Banach spaces along with a priori error analysis and comprehensive numerical comparisons with popular neural net (NN) approaches such as Deep Operator Net (DeepONet) [Lu et al.] and Fourier Neural Operator (FNO) [Li et al.]. We consider the setting where the input/output spaces of target operator G:UV\mathcal{G}^\dagger\,:\, \mathcal{U}\to \mathcal{V} are reproducing kernel Hilbert spaces (RKHS), the data comes in the form of partial observations ϕ(ui),φ(vi)\phi(u_i), \varphi(v_i) of input/output functions vi=G(ui)v_i=\mathcal{G}^\dagger(u_i) (i=1,,Ni=1,\ldots,N), and the measurement operators ϕ:URn\phi\,:\, \mathcal{U}\to \mathbb{R}^n and φ:VRm\varphi\,:\, \mathcal{V} \to \mathbb{R}^m are linear. Writing ψ:RnU\psi\,:\, \mathbb{R}^n \to \mathcal{U} and χ:RmV\chi\,:\, \mathbb{R}^m \to \mathcal{V} for the optimal recovery maps associated with ϕ\phi and φ\varphi, we approximate G\mathcal{G}^\dagger with Gˉ=χfˉϕ\bar{\mathcal{G}}=\chi \circ \bar{f} \circ \phi where fˉ\bar{f} is an optimal recovery approximation of f:=φGψ:RnRmf^\dagger:=\varphi \circ \mathcal{G}^\dagger \circ \psi\,:\,\mathbb{R}^n \to \mathbb{R}^m. We show that, even when using vanilla kernels (e.g., linear or Mat\'{e}rn), our approach is competitive in terms of cost-accuracy trade-off and either matches or beats the performance of NN methods on a majority of benchmarks. Additionally, our framework offers several advantages inherited from kernel methods: simplicity, interpretability, convergence guarantees, a priori error estimates, and Bayesian uncertainty quantification. As such, it can serve as a natural benchmark for operator learning.

View on arXiv
Comments on this paper