We're going to switch gears for a couple of sessions. We'll focus for a couple of sessions on software structures available for gpu and how those might be used to program standard algorithms in machine learning (gradient descent, binary decision tree, association rules, regularization path methods, etc.). Several architectures are available for machine learning algo (gpu, map-reduce on Hadoop, Spark, etc.) Our goal here is to see how the various available architectures compare on some standard algorithms. I'll post some links to software packages.
There's an attendance limit of 10. About 50% of those who RSVP "yes" won't show up. If you're less tenth on the waitlist you'll get in.