Uncertainty quantification has received increasing attention in machine learning in the recent past. In particular, a distinction between aleatoric and epistemic uncertainty has been found useful in this regard. The latter refers to the learner's (lack of) knowledge and appears to be especially difficult to measure and quantify. In this paper, we analyse a recent proposal based on the idea of a second-order learner, which yields predictions in the form of distributions over probability distributions. While standard (first-order) learners can be trained to predict accurate probabilities, namely by minimising suitable loss functions on sample data, we show that loss minimisation does not work for second-order predictors: The loss functions proposed for inducing such predictors do not incentivise the learner to represent its epistemic uncertainty in a faithful way.

Minorquestions: Numberofstateswere6 selected based on the BIC criterion, and the selected number conformed with existing clinical guidelines (please7 refertoresponse "Disease phenotypes" forReviewer3).

Learning the structure of a Bayesian network from observational data is a well knownbutanincredibly difficult problem tosolveinthemachine learning community. Duetoits popularity and applications, a considerable amount of work has been done in this field.

Neural Information Processing Systems http://nips.cc/

Neural Information Processing Systems http://nips.cc/