I'm finally sitting down and looking at how to implement Neural Networks (LeNet) in Python (much of the motivation coming from Anatoly's comment). I've been very slowly following along with Learning from Data from CalTech (since October?!), but honestly that dragged on so long without getting to useful code that I found it harder and harder to actually sit down and watch. So today I sat down to look for practical examples and running code, rather than university courses.

Theano does seem to pop up all over the place, and it seems like the place to start with real hacking, but the tutorials on deeplearning.net left me in the "how does this actually work at the code level" funk. Then this afternoon I stumbled across a reference to Hacker's Guide to Neural Networks which did a nice job of building up that missing bits. So, at the end of the day, I'm thinking "okay, now I can imagine sitting down to play with this stuff"... but that has happened toward the end of the day.