[Noisebridge-discuss] ML 8/3/2011 @ 7:30pm: deep nets part 1; pybrain and restricted boltzmann machines

[Mike Schachter]

Hola everyone!

Through popular request we've decided to start
self-teaching ourselves the ways of deep networks,
starting tomorrow in the Church room at 7:30pm.

One popular route is to first understand restricted
Boltzmann machines, graphical models that can
learn the statistical distributions of their inputs.

Our path will be more concrete - the first step will
be to get everyone up and running with PyBrain,
a python-based machine learning library that implements
RBMs. I put together some basic instructions for
installing PyBrain here, for linux:

https://www.noisebridge.net/wiki/PyBrain

I'll be installing it on OSX tonight and will update the
instructions accordingly. Tomorrow, we'll go through
the installation, and then review the RBM example
supplied with PyBrain and see how it works with
handwritten digit recognition, or perhaps some fake data.

This will be the first part of an N part meetup on deep
nets. See you tomorrow!

  mike