Towards a framework for designing full model selection and optimization systems

Simple item page
dc.contributor.authorSun, Quan
dc.contributor.authorPfahringer, Bernhard
dc.contributor.authorMayo, Michael
dc.coverage.spatialConference held at Nanjing, Chinaen_NZ
dc.date.accessioned2013-07-09T04:26:18Z
dc.date.available2013-07-09T04:26:18Z
dc.date.issued2013
dc.description.abstractPeople from a variety of industrial domains are beginning to realise that appropriate use of machine learning techniques for their data mining projects could bring great benefits. End-users now have to face the new problem of how to choose a combination of data processing tools and algorithms for a given dataset. This problem is usually termed the Full Model Selection (FMS) problem. Extended from our previous work [10], in this paper, we introduce a framework for designing FMS algorithms. Under this framework, we propose a novel algorithm combining both genetic algorithms (GA) and particle swarm optimization (PSO) named GPS (which stands for GA-PSO-FMS), in which a GA is used for searching the optimal structure for a data mining solution, and PSO is used for searching optimal parameters for a particular structure instance. Given a classification dataset, GPS outputs a FMS solution as a directed acyclic graph consisting of diverse data mining operators that are available to the problem. Experimental results demonstrate the benefit of the algorithm. We also present, with detailed analysis, two model-tree-based variants for speeding up the GPS algorithm.en_NZ
dc.format.mimetypeapplication/pdf
dc.identifier.citationSun, Q., Prahringer, B. & Mayo, M. (2013). In Z.-H. Zhou, F. Roli, and J. Kittler (Eds.), Proceedings of the 11th International Workshop on Multiple Classifier Systems (MCS'13), Nanjing, China, LNCS 7872 (pp. 259-270). Berlin Heidelberg: Springer.en_NZ
dc.identifier.doi10.1007/978-3-642-38067-9_23en_NZ
dc.identifier.urihttps://hdl.handle.net/10289/7762
dc.language.isoenen_NZ
dc.publisherSpringeren_NZ
dc.relation.isPartOfProc 11th International Workshop on Multiple Classifier Systemsen_NZ
dc.relation.urihttp://link.springer.com/chapter/10.1007%2F978-3-642-38067-9_23en_NZ
dc.rightsThis is the author's accepted version of a paper published by Springer in the series Lecture Notes in Computer Science (LNCS). The original publication is available at www.springerlink.com.en_NZ
dc.sourceMCS 2013en_NZ
dc.subjectcomputer scienceen_NZ
dc.subjectdata miningen_NZ
dc.subjectMachine learningen_NZ
dc.titleTowards a framework for designing full model selection and optimization systemsen_NZ
dc.typeConference Contributionen_NZ
pubs.begin-page259en_NZ
pubs.elements-id22963
pubs.end-page270en_NZ
pubs.finish-date2013-05-17en_NZ
pubs.place-of-publicationGermanyen_NZ
pubs.start-date2013-05-15en_NZ
pubs.volumeLNCS 7872en_NZ
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
mcs13_working_paper 2.pdf
Size:
563.38 KB
Format:
Adobe Portable Document Format
Thumbnail Image
Name: mcs13_working_paper 2.pdf
Size: 563.38 KB
Kind: Adobe PDF
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Name: license.txt
Size: 1.71 KB
Kind: License
Collections
Computing and Mathematical Sciences Papers