I'd say we have a pretty standard command line interface for working with all the various options now, too. Not as sexy as all the other cool stuff, but it sure makes it easier to run/use. I don't know if we are perfect there yet, but it's a lot better and a lot more consistent across algorithms.
On Oct 19, 2010, at 1:50 PM, Ted Dunning wrote: > - New sequential training framework for logistic regression > - New vector encoding framework for high speed vectorization without a > pre-built dictionary > - Additional elements of supervised model evaluation framework > - Promoted several pieces of old Colt framework to tested status (QR > decomposition, in particular) > > > On Tue, Oct 19, 2010 at 7:30 AM, Jeff Eastman > <[email protected]>wrote: > >> On 10/19/10 7:00 AM, Sean Owen wrote: >> >> I've even lost track of what the big-ticket changes have been since 0.3. I'm >> compiling 7-8 bullet points for the release notes, as I am going through the >> release process now. >> >> Would anyone please volunteer some bullet points? I don't want to miss >> anything and want to describe it correctly. I'll do my best to fill in what >> seems missing. >> >> >> For clustering, here's a few: >> >> - Model refactoring and CLI changes to improve integration and >> consistency >> - New ClusterEvaluator and CDbwClusterEvaluator offer new ways to >> evaluate clustering effectiveness >> - New Spectral Clustering and MinHash Clustering from GSoC (still >> experimental) >> - New VectorModelClassifier allows any set of clusters to be used for >> classification >> >> -------------------------- Grant Ingersoll http://www.lucidimagination.com
