Inspiration
Inspiration can usually be found in odd places.
When I get home, I need to get back into my electronics box and start playing. I need to figure out some way to keep my predatorial kitten from eating the capacitors, though.
« January 2005 | Main | March 2005 »
Inspiration can usually be found in odd places.
When I get home, I need to get back into my electronics box and start playing. I need to figure out some way to keep my predatorial kitten from eating the capacitors, though.
"Now what?"
but really, he left behind more inspiration than anyone really knows what to do with.
When I was an undergraduate at USC, I almost failed Probability. I recall homework assignments that simply stumped me: I knew neither how to solve the problems, nor even how to interpret tutorials that claimed to explain similar problems. It affected me financially: my apartment security deposit was hit for repairs to the wall, after I flung the Probability book one night in frustration.
I am convinced that the rest of my life will be spent re-learning (slowly) the things I failed to learn the first time when they were taught to me in high school and college. Math, science, history, philosophy, literature... In small ways I've revisited each of these with fresh interest. Why couldn't I have just gotten it them first time around?
So, too it is with probability. Specifically, Bayesian nets. Bayesian nets are a fascinating melding of computer science and mathematics. They are (in my limited understanding thusfar) directed acyclic graphs, where nodes represent random variables, and edges represent conditional probabilities. Inferring joint probabilities from them makes tons of sense. Updating them is harder. Designing them from scratch is a bitch.
But I'm learning. I've found plenty of sites on Bayesian nets. See http://del.icio.us/ehgradman/bayes for a collection of helpful resources.