1: \begin{abstract}

2: We provide a theoretical foundation for non-parametric estimation of functions of random variables using kernel mean embeddings. We show that for any continuous function $f$, consistent estimators of the mean embedding of a random variable $X$ lead to consistent estimators of the mean embedding of $f(X)$.

3: For Mat\'ern kernels and sufficiently smooth functions we also provide rates of convergence.

4: 

5: Our results extend to functions of multiple random variables. If the variables are dependent, we require an estimator of the mean embedding of their joint distribution as a starting point; if they are independent, it is sufficient to have separate estimators of the mean embeddings of their marginal distributions. In either case, our results cover both mean embeddings based on i.i.d.\:samples as well as ``reduced set'' expansions in terms of dependent expansion points. The latter serves as a justification for using such expansions to limit memory resources when applying the approach as a basis for probabilistic programming.

6: 

7: %\Adam{I would finish with ``as long as their joint embedding can be estimated consistently'' - I'm not sure what sort of application you had in mind.}

8: 

9: %  We provide a theoretical foundation for estimating functions of random variables using kernel mean embeddings. We show that for any continuous function an arbitrary consistent estimator of a mean embedding of a random variable leads to a consistent estimator of the mean embedding of a function of this variable. For Gaussian kernels and sufficiently smooth functions we also provide finite sample guarantees. Finally, we provide a discussion of why those results are important for practical applications.

10: \end{abstract}

11: