A higher-order approach to meta-learning

Hilan Bensusan; Christophe Giraud-Carrier; Claire Kennedy

A higher-order approach to meta-learning

Hilan Bensusan
, Christophe Giraud-Carrier
, Claire Kennedy

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

Abstract

Meta-learning, as applied to model selection, consists of inducing mappings from tasks to learners. Traditionally, tasks are characterised by the values of pre-computed meta-attributes, such as statistical and information-theoretic measures, induced decision trees' characteristics and/or landmarkers' performances. In this position paper, we propose to (meta-)learn directly from induced decision trees, rather than rely on an \em ad hoc set of pre-computed characteristics. Such meta-learning is possible within the framework of the typed higher-order inductive learning framework we have developed.

Translated title of the contribution	A higher-order approach to meta-learning
Original language	English
Title of host publication	Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination
Publisher	ECML'2000
Pages	109 - 117
Number of pages	8
Publication status	Published - 2000

Bibliographical note

Other page information: 109-117
Conference Proceedings/Title of Journal: Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination
Other identifier: 1000471

Access to Document

http://www.cs.bris.ac.uk/Publications/pub_master.jsp?id=1000471

Handle.net

Persistent link

Cite this

@inproceedings{698414ad7ec6442485158ee1304a4306,

title = "A higher-order approach to meta-learning",

abstract = "Meta-learning, as applied to model selection, consists of inducing mappings from tasks to learners. Traditionally, tasks are characterised by the values of pre-computed meta-attributes, such as statistical and information-theoretic measures, induced decision trees' characteristics and/or landmarkers' performances. In this position paper, we propose to (meta-)learn directly from induced decision trees, rather than rely on an \textbackslash{}em ad hoc set of pre-computed characteristics. Such meta-learning is possible within the framework of the typed higher-order inductive learning framework we have developed. ",

author = "Hilan Bensusan and Christophe Giraud-Carrier and Claire Kennedy",

note = "Other page information: 109-117 Conference Proceedings/Title of Journal: Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination Other identifier: 1000471",

year = "2000",

language = "English",

pages = "109 -- 117",

booktitle = "Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination",

publisher = "ECML'2000",

}

A higher-order approach to meta-learning. / Bensusan, Hilan; Giraud-Carrier, Christophe; Kennedy, Claire.
Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination. ECML'2000, 2000. p. 109 - 117.

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - A higher-order approach to meta-learning

AU - Bensusan, Hilan

AU - Giraud-Carrier, Christophe

AU - Kennedy, Claire

N1 - Other page information: 109-117 Conference Proceedings/Title of Journal: Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination Other identifier: 1000471

PY - 2000

Y1 - 2000

N2 - Meta-learning, as applied to model selection, consists of inducing mappings from tasks to learners. Traditionally, tasks are characterised by the values of pre-computed meta-attributes, such as statistical and information-theoretic measures, induced decision trees' characteristics and/or landmarkers' performances. In this position paper, we propose to (meta-)learn directly from induced decision trees, rather than rely on an \em ad hoc set of pre-computed characteristics. Such meta-learning is possible within the framework of the typed higher-order inductive learning framework we have developed.

AB - Meta-learning, as applied to model selection, consists of inducing mappings from tasks to learners. Traditionally, tasks are characterised by the values of pre-computed meta-attributes, such as statistical and information-theoretic measures, induced decision trees' characteristics and/or landmarkers' performances. In this position paper, we propose to (meta-)learn directly from induced decision trees, rather than rely on an \em ad hoc set of pre-computed characteristics. Such meta-learning is possible within the framework of the typed higher-order inductive learning framework we have developed.

M3 - Conference Contribution (Conference Proceeding)

SP - 109

EP - 117

BT - Proceedings of the ECML'2000 workshop on Meta-Learning: Building Automatic Advice Strategies for Model Selection and Method Combination

PB - ECML'2000

ER -

A higher-order approach to meta-learning

Abstract

Bibliographical note

Access to Document

Handle.net

Fingerprint

Cite this