Profile photo of ivan

by

Journal fightclub Tuesday May 20

May 15, 2014 in Labwide Announcements

Here is the paper I will be presenting: KODAMA

The paper describes what is essentially a statistical pipeline that includes iterative cross-validation of the results. The authors argue that this produces very accurate (high confidence) classifications. I suppose that one argument against this is that cross-validation may give you a "false sense of security" - since you are resampling the same dataset multiple times.

The project website has a tutorial and a few example datasets if you would want to play with it: kodama-project.com

Please comment to let me know if you want me to just present the paper (more theoretical explanation), or prepare a little run-through tutorial (more practical explanation).

 

1 response to Journal fightclub Tuesday May 20

  1. ivan said on May 20, 2014

    Just an update.
    The time is today from 2 to 4 PM. We do not have to stay there for the complete 2 hours, but we can.
    The room is G750.
    I will be going into the methods more so than the results (if at all). The paper is quite dense, and I would like you guys to understand the reasoning behind selecting the solution methods.

Leave a reply

Your email address will not be published. Required fields are marked *

Skip to toolbar