ORBi: Maes Francis - Embedding Monte Carlo search of features in tree-based ensemble methods

Reference : Embedding Monte Carlo search of features in tree-based ensemble methods


	Document type :	Scientific congresses and symposiums : Paper published in a book
	Discipline(s) :	Engineering, computing & technology : Computer science
	To cite this reference:	http://hdl.handle.net/2268/130657

Title :	Embedding Monte Carlo search of features in tree-based ensemble methods
Language :	English
Author, co-author :	Maes, Francis [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
	Geurts, Pierre [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
	Wehenkel, Louis [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
Publication date :	Sep-2012
Main document title :	Machine Learning and Knowledge Discovery in Data Bases
Editor :	Flach, Peter
	De Bie, Tijl
	Cristianini, Nello
Publisher :	Springer
Collection and collection volume :	Lecture Notes in Artificial Intelligence
Pages :	191-206
Audience :	International
Event name :	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases
Event date :	from 24-09-2012 to 28-09-2012
Event place (city) :	Bristol
Event country :	United Kingdom
Keywords :	[en] Embedded Feature Generation ; Monte Carlo Search ; Decision Trees ; Random Forests ; Tree Boosting
Abstract :	[en] Feature generation is the problem of automatically constructing good features for a given target learning problem. While most feature generation algorithms belong either to the filter or to the wrapper approach, this paper focuses on embedded feature generation. We propose a general scheme to embed feature generation in a wide range of tree-based learning algorithms, including single decision trees, random forests and tree boosting. It is based on the formalization of feature construction as a sequential decision making problem addressed by a tractable Monte Carlo search algorithm coupled with node splitting. This leads to fast algorithms that are applicable to large-scale problems. We empirically analyze the performances of these tree-based learners combined or not with the feature generation capability on several standard datasets.
Permalink :	http://hdl.handle.net/2268/130657

File(s) associated to this reference

Fulltext file(s):

	File	Commentary	Version	Size	Access
Open access	Maes2012Ecml.pdf		Author preprint	522.13 kB	View/Open

All documents in ORBi are protected by a user license.

Dspace v1.5