论文部分内容阅读
We consider machine learning techniques for significantly accelerating simulator-based approximate inference.The first strategy introduces classification for discrimination between representative and poor simulated data sets,such that consistent estimators can be obtained.