Landscape aware algorithm selection for feature selection

Simple item page
dc.contributor.advisorEngelbrecht, Andries Petrusen_ZA
dc.contributor.advisorMalan, Katherine Maryen_ZA
dc.contributor.authorMostert, Werneren_ZA
dc.contributor.otherStellenbosch University. Faculty of Science. Dept. of Mathematical Sciences. Computer Science Division.en_ZA
dc.date.accessioned2023-10-23T05:37:56Zen_ZA
dc.date.accessioned2024-01-08T12:01:50Zen_ZA
dc.date.available2023-10-23T05:37:56Zen_ZA
dc.date.available2024-01-08T12:01:50Zen_ZA
dc.date.issued2023-10en_ZA
dc.descriptionThesis (PhD)--Stellenbosch University, 2023.en_ZA
dc.description.abstractENGLISH ABSTRACT: Feature selection is commonly applied as a pre-processing technique for machine learning to reduce the dimensionality of a problem by removing redundant and irrelevant features. Another desirable outcome of feature selection lies in the potential performance improvement of predictive models. The development of new feature selection algorithms are common within the field, however, relatively little research has historically been done to better understand the feature selection problem from a theoretical perspective. Researchers and practitioners in the field often rely on a trial-and-error strategy to decide on which feature selection algorithm to use for a specific instance of a machine learning problem. This thesis contributes towards a better understanding of the complex feature selection problem by investigating the link between feature selection problem characteristics and the performance of feature selection algorithms. A variety of fitness landscape analysis techniques are used to gain insights into the structure of the feature selection fitness landscape. Performance complementarity for feature selection algorithms is empirically shown, emphasising the potential value of automated algorithm selection for feature selection algorithms. Towards the realisation of a landscape aware algorithm selector for feature selection, a novel performance metric for feature selection algorithms is presented. The baseline fitness improvement (BFI) performance metric is unbiased and can be used for comparative analysis across feature selection problem instances. The insights obtained via landscape analysis are used with other meta-features of datasets and the BFI performance measure to develop a new landscape aware algorithm selector for feature selection. The landscape aware algorithm selector provides a human-interpretable predictive model of the best feature selection algorithm for a specific dataset and classification problem.en_ZA
dc.description.abstractAFRIKAANSE OPSOMMING: Kenmerkkeuse word algemeen toegepas as ’n voorverwerkingstegniek vir masjienleer om die dimensionaliteit van ’n probleem te verminder deur oortollige en irrelevante kenmerke te verwyder. Nog ’n gewenste uitkoms van kenmerkkeuse lê in die potensiële prestasieverbetering van voorspellende modelle. Die ontwikkeling van nuwe kenmerkseleksie-algoritmes is algemeen binne die veld, maar relatief min navorsing is egter histories gedoen om die kenmerkseleksieprobleem beter vanuit ’n teoretiese perspektief te verstaan. Navorsers en praktisyns in die veld maak dikwels staat op ’n proef-by-fout-strategie om te besluit watter kenmerkseleksie- algoritme om vir ’n spesifieke geval van ’n probleem te gebruik. Hierdie tesis dra by tot ’n beter begrip van die komplekse kenmerkseleksieprobleem deur die verband tussen kenmerkseleksieprobleemkenmerke en die werkverrigting van kenmerkseleksie-algoritmes te ondersoek. ’n Verskeidenheid fiksheidslandskap- ontledingstegnieke word gebruik om insigte te verkry rondom die struktuur van die kenmerkkeuse-fiksheidlandskap. Prestasiekomplementariteit vir kenmerkkeusealgoritmes word empiries getoon, wat die waarde van outomatiese algoritmeseleksie vir kenmerkkeusealgoritmes beklemtoon. Met die oog op die verwesenliking van ’n landskapbewuste algoritmekieser vir kenmerkkeuse, word ’n nuwe prestasiemmaatstaf vir kenmerkkeusealgoritmes aangebied. Die basislyn fiksheidsverbetering (BFI) prestasie maatstaf is onbevooroordeeld en kan gebruik word vir vergelykende analise oor iv kenmerkkeuseprobleme. Die insigte wat verkry is deur landskapanalise word saam met die BFI-prestasiemetriek gebruik om ’n nuwe landskapbewuste algoritme-kieser vir kenmerkkeuse te ontwikkel. Die landskapbewuste algoritme-kieser verskaf voorheen onbeskikbare leiding aan navorsers en praktisyns vir die keuse van ’n kenmerkseleksie- algoritme om te ontplooi.af_ZA
dc.description.versionMastersen_ZA
dc.format.extentxix, 160 pages : illustrationsen_ZA
dc.identifier.urihttps://scholar.sun.ac.za/handle/10019.1/128812en_ZA
dc.language.isoen_ZAen_ZA
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch Universityen_ZA
dc.rights.holderStellenbosch Universityen_ZA
dc.subject.lcshMachine learningen_ZA
dc.subject.lcshAlgorithmsen_ZA
dc.subject.lcshBaseline fitness improvementen_ZA
dc.subject.lcshFeature selection en_ZA
dc.subject.lcshInstance space analysisen_ZA
dc.subject.lcshInstance space analysisen_ZA
dc.titleLandscape aware algorithm selection for feature selectionen_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
mostert_landscape_2023.pdf
Size:
10.94 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Doctoral Degrees (Computer Science)