At the developers meeting on Friday, January 18 2013 we discussed the following topics.
- Various funding opportunities for the OpenMx Project were brought up.
- Setting up an OpenMx Foundation similar to other open source projects is a possibility. The the University of Virginia Patent Foundation could help with the legal overhead for this.
- The MIDDLE project could be repackaged for NSF's EAGER call. This is largely a computer and behavioral science collaboration for cybersecurity.
- NIH may be interested in data security issues and MIDDLE related to HIPPA violations and the non-anonymizable nature of genome data.
- The current plan is for Steve Boker to revise the MIDDLE grant proposal and circulate it for comments.
- EM algorithms were discussed at length. Mike Hunter distributed an R implementation of an EM for exploratory factor analysis so the group could see the relevant issues in a simple example. The IRT case differs in that optimization must happen in the M step. Tim Brick brought up the case for fitting active appearance models to faces in which locally free polynomial coefficients are also estimated during the M step optimization, but are not free in a global sense. This implies a use case for a general multistep optimization, of which EM is a special case.
- Tim Brick has made a rough draft of a status report for Multilevel. He'll be circulating the report soon after some polishing.
- Ryne Estabrook is organizing at least one symposium to submit to the APS conference. He'll be working on the draft(s) of the submission documents and requesting feedback.