ADF

Adaptive Decision Forest (ADF) is an incremental machine learning framework for handling steaming data that arrive batches over the time. ADF is capable to classify new records even if they are associated with previously unseen classes. ADF also is capable of identifying and handling concept drift; it, however, does not forget previously gained knowledge. Moreover, ADF is capable of handling big data if the data can be divided into batches.

How to use ADF into MOA framework

Please find the "ADF Manual.pdf" file (given in the repository) which illustrates step-by-step instructions to run ADF code into the MOA framework.

Reference

Rahman, M. G., and Islam, M. Z. (2022): Adaptive Decision Forest: An Incremental Machine Learning Framework, Pattern Recognition, pg. 108345, vol. 122, ISSN 0031-3203. DOI: https://doi.org/10.1016/j.patcog.2021.108345.

BibTeX

@article{RAHMAN2022108345,
title = {Adaptive Decision Forest: An incremental machine learning framework},
journal = {Pattern Recognition},
volume = {122},
pages = {108345},
year = {2022},
issn = {0031-3203},
doi = {https://doi.org/10.1016/j.patcog.2021.108345},
url = {https://www.sciencedirect.com/science/article/pii/S0031320321005252},
author = {Md Geaur Rahman and Md Zahidul Islam},
keywords = {Incremental learning, Decision forest algorithm, Concept drift, Big data, Online learning}
}

@author Gea Rahman https://csusap.csu.edu.au/~grahman/