Regularized greedy forest (RGF) software in C++
Regularized greedy forest (RGF) is a tree ensemble learning method.
With the code provided here, you can do the following:
- Conduct RGF training for regression tasks and binary classification tasks
- Apply trained models to new data
- Reproduce the experimental results reported in
"Learning nonlinear functions using regularized greedy forest"
[1]
This program is free software issued under the
GNU General Public License V3
.
Get started
To get started:
- Download the RGF code archive below and extract the contents.
- Prepare the executable following the instructions in
README included in the code archive.
- Play with sample data going through the examples in the user guide
[2].
Download
- RGF code archive (Feb14-2012-build; change history)
Click here to download the RGF code archive
(rgf_v1.tar.gz; 1.3MB compressed; 3.6MB decompressed; Feb14-2012-build).
See README
for the instructions for installation.
- [Optional] Data archive for reproducing the experiments
This archive is needed only for reproducing the experimental results
in the report
[1].
Click here
to download the data archive
(data_for_exp.tar.gz; 39MB compressed; 141MB decompressed).
See README (for the data archive)
for the instructions for installation.
See the Appendix of the user guide [2]
for how to run the experiments.
The data files in this archive were derived from parts of the data obtained
from the
UCI repository, the
MNIST database, and
CMU StatLib.
See README for more information on the
original data sources and data conversion.
References
[1]
Rie Johnson and Tong Zhang. Learning nonlinear functions using regularized greedy forest.
arXiv:1109.0887, 2011.
[2]
Regularized Greedy Forest Version1: User Guide
(included in the code archive)