The most general change for the fifth edition was the extension of the material related to the
stochastic approaches to AI. To accomplish we added a completely new
Chapter 5 that introduces the stochastic methodology.
From the basic
foundations of set theory and counting we develop the notions of
probabilities, random variables, and independence. We present and nse
Bayes' theorem first with one symptom and one disease and then in its
full general form.
We
examine the hypotheses that underlie the use of Bayes and then present
the argmax and naive Bayes approaches. We present examples of stochastic
reasoning, including several from work in the analysis of language
phenomena. We also introduce the idea of conditional independence that
leads to our presentation of Bayesian belief networks (BBNs) and
d-separation in Chapter 9.
