Thursday, October 25, 2012

InteliFest 2012 Wednesday


New Day here in San Diego - beautiful and cool.   First up is Carlos Seranno-Morales showing how to create rules from Big Data when your SMEs don't have all the answers.  This is heresy for most BRMS guys but it has a bit of truth; sometimes they really don't have all the answers and you have to do quite a bit of data analysis to dig it out, especially in the world of fraud analysis where the SMEs can help guide but the statistics hold the ultimate guide to the answers.

Keynote Speaker is Dr. Stephen Grossberg, Godfather of Neural Networks; formerly at MIT and now at Boston University.  Quite a trip!  He explained a lot on neural connections, visual recognition and ART, Adaptive Resonance Technology.  He ended up showing a lot of what Dr. Gail Carpenter is doing on the ARTMAP front where she did in one day what some SMEs had taken over a year to analyze on some highly complex data mapping and analyzing.

After lunch Gary Riley gave a talk on how he implemented another commercial rulebase (it as ILOG JRules we found out later...) and some of the problems that he experienced with the project.  Most of the problems that he had were mostly with people and planning and PPPPPP.  One thing of note was that they went to using XML for their object representation which led to problems with extensibility.  This is much easier to accomplish in POJO (Plain Old Java Objects) than in XML but XML does extend to using with other software.  It was a swap-out and the Java developers lost out on this one.

Days End we all went to supper at another seafood / steak place again, Saska's, and, of course, I had to get more fresh seafood.  Can't wait until tomorrow.

No comments: