Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks
Autor(a) principal: | |
---|---|
Data de Publicação: | 2015 |
Outros Autores: | , , |
Tipo de documento: | Artigo de conferência |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://hdl.handle.net/11449/168077 |
Resumo: | Machine learning algorithms have been investigated in several scenarios, one of them is the data classification. The predictive performance of the models induced by these algorithms is usually strongly affected by the values used for their hyper-parameters. Different approaches to define these values have been proposed, like the use of default values and optimization techniques. Although default values can result in models with good predictive performance, different implementations of the same machine learning algorithms use different default values, leading to models with clearly different predictive performance for the same dataset. Optimization techniques have been used to search for hyper-parameter values able to maximize the predictive performance of induced models for a given dataset, but with the drawback of a high computational cost. A compromise is to use an optimization technique to search for values that are suitable for a wide spectrum of datasets. This paper investigates the use of meta-learning to recommend default values for the induction of Support Vector Machine models for a new classification dataset. We compare the default values suggested by the Weka and LibSVM tools with default values optimized by meta-heuristics on a large range of datasets. This study covers only classification task, but we believe that similar ideas could be used in other related tasks. According to the experimental results, meta-models can accurately predict whether tool suggested or optimized default values should be used. |
id |
UNSP_2060781e3d7b487dc83b78e47740aaa4 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/168077 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasksDefault valuesHyper-parameter tuningMeta-learningSupport vector machinesMachine learning algorithms have been investigated in several scenarios, one of them is the data classification. The predictive performance of the models induced by these algorithms is usually strongly affected by the values used for their hyper-parameters. Different approaches to define these values have been proposed, like the use of default values and optimization techniques. Although default values can result in models with good predictive performance, different implementations of the same machine learning algorithms use different default values, leading to models with clearly different predictive performance for the same dataset. Optimization techniques have been used to search for hyper-parameter values able to maximize the predictive performance of induced models for a given dataset, but with the drawback of a high computational cost. A compromise is to use an optimization technique to search for values that are suitable for a wide spectrum of datasets. This paper investigates the use of meta-learning to recommend default values for the induction of Support Vector Machine models for a new classification dataset. We compare the default values suggested by the Weka and LibSVM tools with default values optimized by meta-heuristics on a large range of datasets. This study covers only classification task, but we believe that similar ideas could be used in other related tasks. According to the experimental results, meta-models can accurately predict whether tool suggested or optimized default values should be used.Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)Universidade de São Paulo (USP)Universidade Estadual Paulista (UNESP)Eindhoven University of Technology (TU/e)Universidade Estadual Paulista (UNESP)FAPESP: #2012/23114-9Universidade de São Paulo (USP)Universidade Estadual Paulista (Unesp)Eindhoven University of Technology (TU/e)Mantovani, Rafael G.Rossi, André L. D. [UNESP]Vanschoren, JoaquinCarvalho, Andre C. P. L. F.2018-12-11T16:39:38Z2018-12-11T16:39:38Z2015-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/conferenceObject80-92CEUR Workshop Proceedings, v. 1455, p. 80-92.1613-0073http://hdl.handle.net/11449/1680772-s2.0-84944212287Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengCEUR Workshop Proceedings0,167info:eu-repo/semantics/openAccess2021-10-23T16:58:09Zoai:repositorio.unesp.br:11449/168077Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462021-10-23T16:58:09Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
title |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
spellingShingle |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks Mantovani, Rafael G. Default values Hyper-parameter tuning Meta-learning Support vector machines |
title_short |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
title_full |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
title_fullStr |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
title_full_unstemmed |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
title_sort |
Meta-learning recommendation of default hyper-parameter values for SVMs in classifications tasks |
author |
Mantovani, Rafael G. |
author_facet |
Mantovani, Rafael G. Rossi, André L. D. [UNESP] Vanschoren, Joaquin Carvalho, Andre C. P. L. F. |
author_role |
author |
author2 |
Rossi, André L. D. [UNESP] Vanschoren, Joaquin Carvalho, Andre C. P. L. F. |
author2_role |
author author author |
dc.contributor.none.fl_str_mv |
Universidade de São Paulo (USP) Universidade Estadual Paulista (Unesp) Eindhoven University of Technology (TU/e) |
dc.contributor.author.fl_str_mv |
Mantovani, Rafael G. Rossi, André L. D. [UNESP] Vanschoren, Joaquin Carvalho, Andre C. P. L. F. |
dc.subject.por.fl_str_mv |
Default values Hyper-parameter tuning Meta-learning Support vector machines |
topic |
Default values Hyper-parameter tuning Meta-learning Support vector machines |
description |
Machine learning algorithms have been investigated in several scenarios, one of them is the data classification. The predictive performance of the models induced by these algorithms is usually strongly affected by the values used for their hyper-parameters. Different approaches to define these values have been proposed, like the use of default values and optimization techniques. Although default values can result in models with good predictive performance, different implementations of the same machine learning algorithms use different default values, leading to models with clearly different predictive performance for the same dataset. Optimization techniques have been used to search for hyper-parameter values able to maximize the predictive performance of induced models for a given dataset, but with the drawback of a high computational cost. A compromise is to use an optimization technique to search for values that are suitable for a wide spectrum of datasets. This paper investigates the use of meta-learning to recommend default values for the induction of Support Vector Machine models for a new classification dataset. We compare the default values suggested by the Weka and LibSVM tools with default values optimized by meta-heuristics on a large range of datasets. This study covers only classification task, but we believe that similar ideas could be used in other related tasks. According to the experimental results, meta-models can accurately predict whether tool suggested or optimized default values should be used. |
publishDate |
2015 |
dc.date.none.fl_str_mv |
2015-01-01 2018-12-11T16:39:38Z 2018-12-11T16:39:38Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/conferenceObject |
format |
conferenceObject |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
CEUR Workshop Proceedings, v. 1455, p. 80-92. 1613-0073 http://hdl.handle.net/11449/168077 2-s2.0-84944212287 |
identifier_str_mv |
CEUR Workshop Proceedings, v. 1455, p. 80-92. 1613-0073 2-s2.0-84944212287 |
url |
http://hdl.handle.net/11449/168077 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
CEUR Workshop Proceedings 0,167 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
80-92 |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1799965330011848704 |