I was curious how the classifier works.

(i) is it some neural network stuff? or support vector machines?
(ii) how many different classes can it handle? more than 2?


The current version handles only 2 classes, and uses decision stumps with gentle boosting. I did not write the algorithm, so I can’t go into detail. I’ve since rewritten classifier GUI in Python using a multi-class classifier written by one of our comp biologists here. We’re beginning to do platform testing on it now, so it won’t be too long before it’s released to the world.

Now that CellProfiler Analyst 2.0 has been released (which contains the Classifier function), you can check its manual for details about the algorithms, and see also our relevant paper (Jones, et al., PNAS 2009):

Also, good news: the current version of CellProfiler Analyst 2.0 now supports sorting cells into as many classes as you like.