Hi Shankar I am also following the PGM class and I would like to stress out that the way they implement all the factor operations feels to me to be by no means efficient, way too much random memory indexing. However the class seems very insightful, maybe after it ends we will be illuminated as to how to design some PGM algorithms to fit inside scikit-learn.
In this case there seems to be a clear tradeoff between generality and efficiency. Maybe you could instead focus some attention on the HMM module inside scikit-learn. Also, there was some interest in wrapping CRFSuite [1] [2] by Jake. I will need to use CRFs for my research in a short time, so I might be putting some effort into this as well. Currently the effort has hit some design issues, specifically the problem that CRFsuite does input/output via text files in the filesystem. Good luck on the class and looking forward to your contributions. Best, Vlad On Apr 6, 2012, at 19:25 , Gael Varoquaux wrote: > Hey Shankar, > > I respect your decision: it is better for everyone to have less > applications, but well-though out. What you are learning as you go could > help setting up a serious application for next year, hopefully. > > Thanks for keeping us updated. > > Gael > > On Fri, Apr 06, 2012 at 09:41:40PM +0800, Shankar Satish wrote: >> Hello everyone, > >> I was supposed to prepare a proposal for bayesian networks in sklearn. >> However as i researched the details further, i realized out that doing a >> python implementation will be harder than i thought, primarily due to the >> need of many customized data structures. > >> I have also been following the stanford PGM course >> (www.*pgm*-*class*.org/<http://www.pgm-class.org/>). >> They use Matlab for the assignments, and Matlab provides many useful >> operations on sets that are necessary for operations on bayes nets. I think >> i will have a much clearer idea of how to go about about implementing a >> python version at the end of the course. > >> So for these reasons, combined with the fact that i don't want to risk >> putting in a proposal that i might not be able to complete within the >> summer deadline, i have decided to drop the idea of a GSoC proposal for >> bayes nets. I will instead finish the course first and code the sklearn >> version at liesure. > >> I would like to thank you all for your comments and feedback, and i would >> especially like to thank Andy for offering to mentor me :). > >> regards >> shankar. > > > -- > Gael Varoquaux > Researcher, INRIA Parietal > Laboratoire de Neuro-Imagerie Assistee par Ordinateur > NeuroSpin/CEA Saclay , Bat 145, 91191 Gif-sur-Yvette France > Phone: ++ 33-1-69-08-79-68 > http://gael-varoquaux.info > > ------------------------------------------------------------------------------ > For Developers, A Lot Can Happen In A Second. > Boundary is the first to Know...and Tell You. > Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! > http://p.sf.net/sfu/Boundary-d2dvs2 > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------------------------------------------------------------------ For Developers, A Lot Can Happen In A Second. Boundary is the first to Know...and Tell You. Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! http://p.sf.net/sfu/Boundary-d2dvs2 _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
