Beyond homemade artificial data sets

Núria MacIà; Albert Orriols-Puig; Ester Bernadó-Mansilla

doi:10.1007/978-3-642-02319-4_73

Beyond homemade artificial data sets

Núria MacIà^*
, Albert Orriols-Puig
, Ester Bernadó-Mansilla

^*Corresponding author for this work

Research output: Book chapter › Conference contribution › peer-review

7 Citations (Scopus)

Abstract

One of the most important challenges in supervised learning is how to evaluate the quality of the models evolved by different machine learning techniques. Up to now, we have relied on measures obtained by running the methods on a wide test bed composed of real-world problems. Nevertheless, the unknown inherent characteristics of these problems and the bias of learners may lead to inconclusive results. This paper discusses the need to work under a controlled scenario and bets on artificial data set generation. A list of ingredients and some ideas about how to guide such generation are provided, and promising results of an evolutionary multi-objective approach which incorporates the use of data complexity estimates are presented.

Original language	English
Title of host publication	Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings
Pages	605-612
Number of pages	8
DOIs	https://doi.org/10.1007/978-3-642-02319-4_73
Publication status	Published - 2009
Event	4th International Conference on Hybrid Artificial Intelligence Systems, HAIS 2009 - Salamanca, Spain Duration: 10 Jun 2009 → 12 Jun 2009

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	5572 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	4th International Conference on Hybrid Artificial Intelligence Systems, HAIS 2009
Country/Territory	Spain
City	Salamanca
Period	10/06/09 → 12/06/09

Keywords

Artificial data sets
Data complexity
Machine learning

Access to Document

10.1007/978-3-642-02319-4_73

Cite this

MacIà, N., Orriols-Puig, A., & Bernadó-Mansilla, E. (2009). Beyond homemade artificial data sets. In Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings (pp. 605-612). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5572 LNAI). https://doi.org/10.1007/978-3-642-02319-4_73

@inproceedings{afe220493f064de1b0fb94f113bc45d6,

title = "Beyond homemade artificial data sets",

abstract = "One of the most important challenges in supervised learning is how to evaluate the quality of the models evolved by different machine learning techniques. Up to now, we have relied on measures obtained by running the methods on a wide test bed composed of real-world problems. Nevertheless, the unknown inherent characteristics of these problems and the bias of learners may lead to inconclusive results. This paper discusses the need to work under a controlled scenario and bets on artificial data set generation. A list of ingredients and some ideas about how to guide such generation are provided, and promising results of an evolutionary multi-objective approach which incorporates the use of data complexity estimates are presented.",

keywords = "Artificial data sets, Data complexity, Machine learning",

author = "N{\'u}ria MacI{\`a} and Albert Orriols-Puig and Ester Bernad{\'o}-Mansilla",

year = "2009",

doi = "10.1007/978-3-642-02319-4\_73",

language = "English",

isbn = "3642023185",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "605--612",

booktitle = "Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings",

note = "4th International Conference on Hybrid Artificial Intelligence Systems, HAIS 2009 ; Conference date: 10-06-2009 Through 12-06-2009",

}

MacIà, N, Orriols-Puig, A & Bernadó-Mansilla, E 2009, Beyond homemade artificial data sets. in Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5572 LNAI, pp. 605-612, 4th International Conference on Hybrid Artificial Intelligence Systems, HAIS 2009, Salamanca, Spain, 10/06/09. https://doi.org/10.1007/978-3-642-02319-4_73

Beyond homemade artificial data sets. / MacIà, Núria; Orriols-Puig, Albert; Bernadó-Mansilla, Ester.
Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings. 2009. p. 605-612 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5572 LNAI).

Research output: Book chapter › Conference contribution › peer-review

TY - GEN

T1 - Beyond homemade artificial data sets

AU - MacIà, Núria

AU - Orriols-Puig, Albert

AU - Bernadó-Mansilla, Ester

PY - 2009

Y1 - 2009

N2 - One of the most important challenges in supervised learning is how to evaluate the quality of the models evolved by different machine learning techniques. Up to now, we have relied on measures obtained by running the methods on a wide test bed composed of real-world problems. Nevertheless, the unknown inherent characteristics of these problems and the bias of learners may lead to inconclusive results. This paper discusses the need to work under a controlled scenario and bets on artificial data set generation. A list of ingredients and some ideas about how to guide such generation are provided, and promising results of an evolutionary multi-objective approach which incorporates the use of data complexity estimates are presented.

AB - One of the most important challenges in supervised learning is how to evaluate the quality of the models evolved by different machine learning techniques. Up to now, we have relied on measures obtained by running the methods on a wide test bed composed of real-world problems. Nevertheless, the unknown inherent characteristics of these problems and the bias of learners may lead to inconclusive results. This paper discusses the need to work under a controlled scenario and bets on artificial data set generation. A list of ingredients and some ideas about how to guide such generation are provided, and promising results of an evolutionary multi-objective approach which incorporates the use of data complexity estimates are presented.

KW - Artificial data sets

KW - Data complexity

KW - Machine learning

UR - https://www.scopus.com/pages/publications/70350632465

U2 - 10.1007/978-3-642-02319-4_73

DO - 10.1007/978-3-642-02319-4_73

M3 - Conference contribution

AN - SCOPUS:70350632465

SN - 3642023185

SN - 9783642023187

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 605

EP - 612

BT - Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings

T2 - 4th International Conference on Hybrid Artificial Intelligence Systems, HAIS 2009

Y2 - 10 June 2009 through 12 June 2009

ER -

MacIà N, Orriols-Puig A, Bernadó-Mansilla E. Beyond homemade artificial data sets. In Hybrid Artificial Intelligence Systems - 4th International Conference, HAIS 2009, Proceedings. 2009. p. 605-612. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-02319-4_73

Beyond homemade artificial data sets

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this