ml-rapids: Incremental learning written in C++ exposed in Python
ml-rapids implements incremental learning methods in C++ and exposes them via SWIG in Python. Installation can be achieved simply with pip install ml_rapids. You can test your installation with running Python:
# testing ml-rapids
import ml_rapids
ml_rapids.test()Further documentation is available here:
Implemented incremental learning methods
- Classification
- Majority Class
- Naive Bayes
- Logistic Regression
- Perceptron
- VFDT (Very Fast Decision Trees) aka Hoeffding Trees
- HAT (Hoeffding Adaptive Trees)
- Bagging
- Regression
- /
All the methods implement sklearn incremantal learner interface (includes fit, partial_fit and predict methods).
Future plans
Streaming random forest on top of Hoeffding trees will be implemented.
The library will be exposed via also via npm packages.
Development
Development notes can be read here.
Python deployment notes can be read here.
Acknowledgements
ml-rapids is developed by AILab at Jozef Stefan Institute.
This repository is based strongly on streamDM-cpp.
This research was funded by European Union's Horizon 2020 programme project PerceptiveSentinel (Research and Innovation) grant number 776115, project NAIADES (Innovation Action) grant number 820985.