1: \begin{abstract}

2: Deep nonparametric regression, characterized by the utilization of deep neural networks to learn target functions, has emerged as a focus of research attention in recent years. Despite considerable progress in understanding convergence rates, the absence of asymptotic properties hinders rigorous statistical inference. 

3: To address this gap, we propose a novel framework that transforms the deep estimation paradigm into a platform conducive to conditional mean estimation, leveraging the conditional diffusion model. 

4: Theoretically, we develop an

5: end-to-end convergence rate for the conditional diffusion model

6: and establish the asymptotic normality of the generated samples. Consequently, we are equipped to construct confidence regions, facilitating robust statistical inference. 

7: Furthermore, through  numerical experiments, we empirically validate the efficacy of our proposed methodology.

8: 

9: 

10: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

11: \vspace{0.5cm} \noindent{\bf KEY WORDS}:

12: Statistical inference,

13: Conditional diffusion model, 

14: End-to-End error analysis, 

15: Deep nonparametric regression.

16: \end{abstract}

17: