Chapter 9 is a particularly interesting review of the recent machine learning research making the point that, absent knowledge of a problem's specific domain, no one classifier is better that any other.This point, it seems to me, has interesting implications for matters such as handwriting identification -- and also for the question of the nature of uncertain factual inference and statistical inference in general. No? But cf.(?) Judea Pearl, who maintains that all interesting inference involves causality.
Well, I will have to buy the book by Duda et al.