AdaBoost

This is a C++ implementation of AdaBoost.

Build

Requirement

- CMake (http://www.cmake.org/)

1. cmake .

2. make

Data format

The format of training and test data file is the same as SVM-Light(http://svmlight.joachims.org/) and libsvm(http://www.csie.ntu.edu.tw/~cjlin/libsvm). That is:

<label> <feature>:<value> <feature>:<value> ... <feature>:<value>  
   .  
   .  
   .  

<label> = {+1, -1}  
<feature>: feature index (integer value starting from 1)  
<value>: feature value (double)'

There are sample data sets at http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/.

Usage

Training

>./abtrain [options] training_set_file [model_file] options: -t: type of boosting (0:discrete, 1:real, 2:gentle) [default:2] -r: the number of rounds [default:100] -v: verbose'

Prediction

>./abpredict [options] test_set_file model_file options: -o: output score file -v: verbose'