Robust and scalable inference for simulator-based models

Simulator-based models are becoming increasingly popular in many research
domains across academia and industry. Calibration of such models in the light of
data and quantification of uncertainty about model parameters are key challenges
for practical applications and the topic has received increasing attention
during the past decade. We will discuss various inference techniques for
simulator-based models that improve computational feasibility by adopting
techniques from machine learning to build surrogate models for the approximate
likelihood or posterior. Asymptotic properties of Jensen-Shannon Divergence based inference will also be briefly
considered. Several of the discussed approaches are available in the
open-source software platform Engine for Likelihood-Free Inference (ELFI): http://elfi.ai.