L1-based compression of random forest models

Abstract

High-dimensional supervised learning problems, e.g. in image exploitation and bioinformatics, are more frequent than ever. Tree-based ensemble methods, such as random forests (Breiman, 2001) and extremely randomized trees (Geurts et al., 2006), are effective variance reduction techniques offering in this context a good trade-off between accuracy, computational complexity, and interpretability.

Extracted Key Phrases

Cite this paper

@inproceedings{Joly2012L1basedCO, title={L1-based compression of random forest models}, author={Arnaud Joly and François Schnitzler and Pierre Geurts and Louis Wehenkel}, booktitle={ESANN}, year={2012} }