This package contains MATLAB-code for the AL-RF classification method, which is described in the publication cited below. This iterative active learning algorithm uses random forests in combination with a training utility value criterion based on second order distributions. It can be applied for convenient annotation of secondary ion mass spectrometry (SIMS) images. In each learning iteration, it then determines that pixel of the MS image that is deemed most useful for improving the current classifier and updates the classifier.
The corresponding publication can be found here:
Michael Hanselmann, Jens Röder, Ullrich Köthe, Bernhard Y. Renard, Ron M.A. Heeren, Fred A. Hamprecht: Active Learning for Convenient Annotation and Classification of Secondary Ion Mass Spectrometry Images, Analytical Chemistry 2012.
DOI: http://pubs.acs.org/doi/abs/10.1021/ac3023313
Techreport: hanselmann_12_active.pdf