Two-step Approximate Bayesian Computation for inferring network models using Copula

In this thesis we first give a brief review of the ABC method. Then we propose a two-step method for parameter inference based on Rejection ABC and MCMC ABC algorithms with Gaussian copula for early rejection. Moreover, we present several examples of network models to compare our strategy with conventional ABC methods. The first proposed algorithm based on Rejection ABC algorithm is evaluated by an exact model, a deterministic model and two stochastic models. The second proposed algorithm based on MCMC ABC is evaluated by an exact model and two deterministic models. Numerical results show that our methods are efficient.