What Keeps a Bayesian Awake at Night
The Cambridge Machine Learning Group is launching a blog, featuring a first two-part post about what keeps a Bayesian awake at night. In the first part, during day time, we lay out the standard arguments that many use to support Bayesian inference, ranging from more fundamental theorems, like Cox’s theorem, to unit tests, like Wald’s theorem. In the second part, at night time, we take a closer look at these standard arguments and identify the weaknesses which cause Bayesians to lose sleep at night: the standard justifications have problems, modelling is hard and sensitive to innocolous details, and—worst of all—one typically must resort to approximate inference. Check it out!