Google Developers Day US – Theorizing from Data
So before I got on a plane to fly from NM to RH, I prepared myself for boredom. One of the things I did was encode some of the Google Developer Day videos on Youtube for my Sony PSP. I think it was one of three times I’ve actually used my PSP!
Anyway, most of the videos were pretty fluffy and didn’t hold my interest, but the talk by Peter Norvig about statistical analysis was pretty darn interesting. Funny sidenote: when I went to Google for an interview, Peter Norvig was the special speaker dude. He had a pretty cool Hawaiian shirt on then too, as I recall.
Anyway, the talk brings up some pretty interesting things, like how if you feed enough statistics to a computer, the actual algorithm matters less and less. I’ve been interested in AI and machine learning, but I never really took any formal classes.
The stuff in the talk has sort of stuck in my head now. So I’m tending to see a lot of problems as being solvable by statistical analysis/classification. Like that Spock Challenge thing I blogged about earlier. Anyway, I’ve got an idea for a wacky application of Naive Bayesian Classification, but I won’t mention it yet (in case it’s an actual good idea, or in case I decide to bail after I don’t want to figure out the probability math).