Abstract:Motivated by the surge of interest in Koopman operator theory, we propose a machine-learning alternative based on a functional Bayesian perspective for operator-theoretic modeling of unknown, data-driven, nonlinear dynamical systems. This formulation is directly done in an infinite-dimensional space of linear operators or Hilbert space with universal approximation property. The theory of reproducing kernel Hilbert space (RKHS) allows the lifting of nonlinear dynamics to a potentially infinite-dimensional space via linear embeddings, where a general nonlinear function is represented as a set of linear functions or operators in the functional space. This allows us to apply classical linear Bayesian methods such as the Kalman filter directly in the Hilbert space, yielding nonlinear solutions in the original input space. This kernel perspective on the Koopman operator offers two compelling advantages. First, the Hilbert space can be constructed deterministically, agnostic to the nonlinear dynamics. The Gaussian kernel is universal, approximating uniformly an arbitrary continuous target function over any compact domain. Second, Bayesian filter is an adaptive, linear minimum-variance algorithm, allowing the system to update the Koopman operator and continuously track the changes across an extended period of time, ideally suited for modern data-driven applications such as real-time machine learning using streaming data. In this paper, we present several practical implementations to obtain a finite-dimensional approximation of the functional Bayesian filter (FBF). Due to the rapid decay of the Gaussian kernel, excellent approximation is obtained with a small dimension. We demonstrate that this practical approach can obtain accurate results and outperform finite-dimensional Koopman decomposition.

Operator-Valued Kernels, Machine Learning, and Dynamical Systems

Hilbert space valued Gaussian processes, their kernels, factorizations, and covariance structure

Canonical data-reconstructions via kernels, Hilbert space-valued Gaussian processes, and quantum states

Positive Definite Kernels, Algorithms, Frames, and Approximations

Gaussian processes in Non-commutative probability

Multiple Operator-valued Kernel Learning

Non-Euclidean Monotone Operator Theory and Applications

Minimax Optimal Kernel Operator Learning via Multilevel Training

Entangled Kernels -- Beyond Separability

Random Fourier Features for Operator-Valued Kernels

Kernel Operator-Theoretic Bayesian Filter for Nonlinear Dynamical Systems

Operator means of probability measures and generalized Karcher equations

Duality in RKHSs with Infinite Dimensional Outputs: Application to Robust Losses

Kernel theorems for operators on co-orbit spaces associated with localised frames

Multilinear operator-valued Calderón-Zygmund theory

Tropical Reproducing Kernels and Optimization

On the Expressive Power of Kernel Methods and the Efficiency of Kernel Learning by Association Schemes

Operator Learning: Algorithms and Analysis

Littlewood-Paley Operators with Variable Kernels

Auxiliary Functions as Koopman Observables: Data-Driven Analysis of Dynamical Systems via Polynomial Optimization

The Koopman Expectation: An Operator Theoretic Method for Efficient Analysis and Optimization of Uncertain Hybrid Dynamical Systems