[CS241] cs241 mailing list

[Matt Lease]

1. Here are current cs241 listserv subscribers (only 9)-- please see who
in your group isn't signed up and bug them to do it. :)

http://list.cs.brown.edu/mailman/listinfo/CS241

    * backman at cs.brown.edu
    * brendan_shean at brown.edu
    * christian.pfrang at gmx.de
    * js at cs.brown.edu
    * kburchet at cs.brown.edu
    * kl at cs.brown.edu
    * pat at cs.brown.edu
    * pyadolla at cs.brown.edu
    * ryan at cs.brown.edu

2. Monday I said you can get flat clusters out of a hierarchical
clustering algorithm by taking a slice through the final hierarchical
tree, but I'm not sure this was clear to everyone. An easier way maybe
to think about it is simply as early termination of bottom-up
hierarchical clustering -- instead of continuing to merge until you have
1 super cluster encompassing all the data, you just run until you have k
clusters.  Similarly, instead of running top-down clustering until each
cluster has only one point, you can similarly terminate it early when
you have k clusters.  Relative to the complete hierarchical tree, each
of these output a particular slice through it. Other alternative slices
are also possible: any way you can prune the complete tree such that
there are only k leaves (clusters).

3. I'm happy to loan out (the Brown lib copy of) Cover & Thomas if
anyone would like to borrow it.

matt