1: \begin{abstract}

2: 		

3: 		Intelligent real-world systems critically depend on expressive information about their system state and changing operation conditions, \eg due to variation in temperature, location, wear, or aging. 

4: 		To provide this information, online inference and learning attempts to perform state estimation and (partial) system identification simultaneously. Current works combine tailored estimation schemes with flexible learning-based models but suffer from convergence problems and computational complexity due to many degrees of freedom in the inference problem (\ie parameters to determine). To resolve these issues, we propose a procedure for data-driven offline conditioning of a highly flexible \ac{gp} formulation such that \textit{online} learning is restricted to a subspace, spanned by expressive basis functions.  

5: 		Due to the simplicity of the transformed problem, a standard particle filter can be employed for Bayesian inference. In contrast to most existing works, the proposed method enables online learning of target functions that are nested nonlinearly inside a first-principles model. Moreover, we provide a theoretical quantification of the error, introduced by restricting learning to a subspace. 

6: 		A Monte-Carlo simulation study with a nonlinear battery model shows that the proposed approach enables rapid convergence with significantly fewer particles compared to a baseline and a state-of-the-art method.

7: 	\end{abstract}

8: