/AdaBoost-SVM

AdaBoost and Support Vector Machines for Unbalanced Data Sets.

Primary LanguageC++

Usage:

1. Compile 
	make clean; make

2. Run training mode
	./Zhang-Project4 -c c_value -W weight_file training_file
	
3. Run testing mode
	./Zhang-Project4-predict testing_file ./saved_models output_file

4. Batch run
	./run_dataname.sh 
	
** Note **
	Please modify the initial gamma (named "gamma" in Line 116) in Zhang-Project4.c before compiling. The C value for each dataset was found by grid.py to get the best performance. 
	glass: gamma = 2;    C = 515.561241629
	liver: gamma = 1.5;  C = 147.03338944
	vowel: gamma = 10;   C = 4.0



File included:
	dataname.wgt				Initial weight file
	Zhang-Project4.c			Training code
	Zhang-Project4-predict.c	Testing code
	saved_models				Directory to store component learners. Training code writes all the model files to this directory and testing code read models from this directory.
	run_*.sh					Batch file to run AdaBoost.M1 and AdaBoost.M1 with varying C.