This project is hosted at:
Important Update (2012):
- MultiBoost has now moved and found a new life at www.multiboost.org!
MultiBoost is a portable C++ implementation of the multi-class AdaBoost algorithm, that is AdaBoost.MH. AdaBoost is a powerful meta-learning algorithm commonly used in machine learning. It is ready to use and to crunch numbers!
Key features:
- Fast and easy to use
- Fully documented with Doxygen
- Very easy to add weak learners and extend
- Support abstention using two different techniques
- Support regularization
- Support "AdaBoost.MH with real valued predictions" (see BoosTexter: A Boosting-based System for Text Categorization)
- Supports Haar-Like features similar to the ones used by Viola&Jones' paper "Robust Real-time Object Detection"
- XML serialization
- Entirely written in pure C++. No additional libraries required
- Tested on linux and windows (OSX to come: I don't have a machine around! :P)
Currently available weak learners are:
- Single threshold decision stump (default)
- Multi threshold decision stump
- Single thresdhold haarlike-features decision stump
- Multi thresdhold haarlike-features decision stump