This is a research project evolving around the machine learning algorithm BOOMER – An algorithm for learning ensembles of gradient boosted multi-output rules that integrates with the popular scikit-learn machine learning framework. It is aimed at multi-output problems, including multi-label classification and multi-output regression.
The BOOMER algorithm is build upon a modular framework for implementing rule learning algorithms. This enables to implement different kinds of algorithms more easily. One example is the multi-label SeCo algorithm provided by this project. It is based on traditional rule learning techniques and is particularly well-suited for learning interpretable models. Additional algorithms may follow in the future. The same applies to tools and utilities evolving around these algorithms.
Software packages provides by this project
A gradient boosting algorithm for multi-output classification and regression
A separate-and-conquer algorithm for multi-label classification
A command line utility for running machine learning experiments
Other sources of information
