ABSTRACT
This chapter explains that whole swathes of the fields of Bayesian and non-Bayesian statistics are concerned with the supervised learning problem. The field of "supervised learning" goes by many other names: inductive inference, statistical inference, machine learning, regression and classification. Without a shared culture, even given a single shared theoretical superstructure, supervised learning is more a set of engineering techniques and tools for working with them than it is a bona fide science. Given the far-ranging nature of the supervised-learning problem, it should come as no surprise that theoretical investigations of supervised learning have been conducted from a number of different vantage points. The fact that all of the issues are so open-ended suggests that as a science supervised learning is immature. Corroborating evidence for the view arises if one views supervised learning from the perspective of physics in particular and the natural sciences in general.
