AOSOLogitBoost_.zip_ensemble matlab_machine learning

* Platform The codes are supposed to be cross-platform. Binaries for win32 are already shipped in "./private" directory. To compile the codes yourself, following below instructions: ** Third party library OpenCV2.2 (or later) is needed.(Specifically, the module opencv_core is used for the matrix.) ** Change current directory to "AOSOLogitBoost\mex". Make sure the OpenCV incude and lib paths in settings.m are right in your computer. Run make_AOSOLogitBoost.m. This will generate mex files to directory "AOSOLogitBoost\private". * To run the sample code ** Change current directory to "AOSOLogitBoost". ** Run main_run.m for single run of datasets residing in memory (as Matlab matrix). ** Run main_run_files.m for batch run of datasets files. Results would be printed to command window and saved to ".\rst_uci" * Warning Runs on some datasets (e.g. Convertype) are extremely time consuming, we suggest one use parallel computation when running on all datasets and all parameters (e.g. the Matlab Distributed Computation Environment) * Description main_run.m, main_run_files.m: the main sript file. AOSOLogitBoost.m: class file for AOSOLogitBoost batch_boost_basic.m & batch_AOSOLogitBoost.m: class for batch run of aosoboost on datasets files \dataset: store datasets \mex: source codes for mex files \private: binaries for mex files \rst: saved results