1: \begin{abstract}

2: In simulation-based inferences for partially observed Markov process models (POMP), the by-product of the Monte Carlo 

3: filtering is an approximation of the log likelihood function. Recently, iterated filtering \cite{ionides06-pnas,ionides11} has originally been

4: introduced and it has been shown that the gradient of the log likelihood can also be approximated. 

5: Consequently, different stochastic optimization algorithm can be applied to estimate the parameters of the underlying models. 

6: As accelerated gradient is an efficient approach in the optimization literature, we show that we 

7: can accelerate iterated filtering in the same manner and inherit that high convergence rate while relaxing the restricted  

8: conditions of unbiased gradient approximation.  

9: We show that this novel algorithm can be applied to both convex and nonconvex log likelihood functions. 

10: In addition, this approach has substantially outperformed most of other previous approaches in a toy example and in a challenging scientific problem of modeling 

11: infectious diseases.  

12: \end{abstract}

13: