Simulation Based Inference (SBI)

In this mini-course we will look at Simulation Based Inference (SBI), also known as Approximate Bayesian Computation (ABC) and Likelihood Free Inference (LFI): a way to conduct inference (either Frequentist or Bayesian) when we cannot compute the likelihood P(X|θ) analytically, though we can still generate samples. In addition to understanding the problem and the different solutions, we will look at examples using both R and Python. The algorithms that we will look into include Rejection ABC, MCMC ABC, Regression Adjustment, SNPE, SNLE and SNRE.