> Take a look at MADLib. My suggestion would be extending the MADlib > functions; there's plenty of unimplemented ML algrothims which could be > added to it.
I went through the MADLib library and came up with the following two ideas which I feel could be potential GSoC 2013 projects: 1) MADlib currently has K-means clustering implemented.I would suggest implementing the K-medoids clustering as it has better performance as compared to K-means clustering.We could use k-means clustering code base as the starting point for our implementation. 2) A more complex project would be to implement backpropogation algorithm for much better classification. This would require implementing some parts of neural network algorithms as well. Again, I am willing to mentor either of the two projects if they are taken. Regards, Atri -- Regards, Atri l'apprenant -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
