Kernel center adaptation in the reproducing kernel Hilbert space embedding method

Summary The performance of adaptive estimators that employ embedding in reproducing kernel Hilbert spaces (RKHS) depends on the choice location basis centers. Parameter convergence and error approximation rates depend where how centers are distributed state‐space. In this article, we develop theory relates parameter to position We criteria for choosing a specific class systems by exploiting fact state trajectory regularly visits neighborhood positive limit set. Two algorithms, based centroidal Voronoi tessellations Kohonen self‐organizing maps, derived choose RKHS method. Finally, implement these methods two practical examples test their effectiveness.