1: \begin{proof}

2: Under Assumption \ref{smoothness}, by applying Lemma \ref{vector_bound} and \ref{matrix_bound}, we can bound the moments of $\dot{M}_i(\theta_0)$ and $\ddot{M}_i(\theta_0) - \ddot{M}_0(\theta_0)$. Rigorously, for $1 \le q \le 8$, we have

3: \begin{eqnarray*}

4: \mathbb{E} \left[ \|\dot{M}_i(\theta_0)\|^q \right] &\le& \frac{C_v(q,d)}{n^{q/2}}G^q, \\

5: \mathbb{E} \left[ \vertiii{ \ddot{M}_i(\theta_0) - \ddot{M}_0(\theta_0)}^q \right] &\le& \frac{C_m(q,d)}{n^{q/2}}H^q.

6: \end{eqnarray*}

7: Therefore, by Markov's inequality, we have

8: \begin{eqnarray*}

9: & & \mbox{Pr}(E^c) = \mbox{Pr}(E_1^c \cup E_2^c  \cup E_3^c) \le \mbox{Pr}(E_1^c) + \mbox{Pr}(E_2^c) + \mbox{Pr}(E_3^c) \\

10: &\le& \frac{\mathbb{E} \left[ |\frac{1}{n}\sum_{j=1}^n L(x_j) - \mathbb{E}[L(x)]|^8 \right]}{L^8} + \frac{\mathbb{E} \left[ \vertiii{ \ddot{M}_i(\theta_0) - \ddot{M}_0(\theta_0)}^8 \right]}{(\lambda/4)^8} + \frac{\mathbb{E} \left[ \|\dot{M}_i(\theta_0)\|^8 \right]}{(\lambda \delta'/2)^8} \\

11: &\le& O(\frac{1}{n^4}) + O(\frac{1}{n^4}) + O(\frac{1}{n^4}) = O(n^{-4}).

12: \end{eqnarray*}

13: \end{proof}

14: