
I'm working late tonight. There are two things I have to do: Preparing for a meeting next week (boring), and preparing a lecture for tomorrow morning (fun).
The subject I'm gonna teach tomorrow is Bayesian inversion. It's based on the so-called Bayes' rule of mathematical statistics. Bayes' rule is (of course) named after Thomas Bayes (1702-1761) who discovered this relation.
Thomas Bayes was a theologist and mathematician. During his lifetime, he published two articles, one on theology, and one on mathematics. Bayes rule, however, was published after his death. He never got the chance to appreciate the fame it earned him.
Bayes rule is about how a probability distribution is changed if we add some extra information. I'll try to explain. Consider to random variables:
(A): Kids like the dinner and don't complain
(B): We have pizza for dinner
The probability of (A), kids like the dinner is maybe 60% in our family. However, given (B), we have pizza for dinner, the probability is a lot higher. That's easy.
The opposite (or inverse) question is harder. If (A), kids like the dinner, what's the probability of (B), we have pizza for dinner? The kids like other things too, for instance taco and pasta and steak. This question does not have a unique answer.
But if we add the extra information that we have Taco every Friday, pizza every Saturday, and pasta and steak at most once a month, it's a little bit easier. If kids like the dinner, we probably have either pizza or taco. Still we have two equally probable answers (pizza or taco), but the last two (pasta and steak) are less probable.
You see what I mean? Cool, isn't it >:)
(The picture is Bayes rule, where P is the probability and A and B as above. I wrote it down on a sheet of paper and put it on the scanner. Tomorrow I'm gonna write it on the blackboard for the students)
