Show simple item record  

dc.contributor.authorSchmidberger, Gabien_NZ
dc.date.accessioned2009-09-08T16:05:42Z
dc.date.available2009-09-11T14:44:49Z
dc.date.issued2009en_NZ
dc.identifier.citationSchmidberger, G. (2009). Tree-based Density Estimation: Algorithms and Applications (Thesis). The University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/3283en
dc.identifier.urihttps://hdl.handle.net/10289/3283
dc.description.abstractData Mining can be seen as an extension to statistics. It comprises the preparation of data and the process of gathering new knowledge from it. The extraction of new knowledge is supported by various machine learning methods. Many of the algorithms are based on probabilistic principles or use density estimations for their computations. Density estimation has been practised in the field of statistics for several centuries. In the simplest case, a histogram estimator, like the simple equalwidth histogram, can be used for this task and has been shown to be a practical tool to represent the distribution of data visually and for computation. Like other nonparametric approaches, it can provide a flexible solution. However, flexibility in existing approaches is generally restricted because the size of the bins is fixed either the width of the bins or the number of values in them. Attempts have been made to generate histograms with a variable bin width and a variable number of values per interval, but the computational approaches in these methods have proven too difficult and too slow even with modern computer technology. In this thesis new flexible histogram estimation methods are developed and tested as part of various machine learning tasks, namely discretization, naive Bayes classification, clustering and multiple-instance learning. Not only are the new density estimation methods applied to machine learning tasks, they also borrow design principles from algorithms that are ubiquitous in artificial intelligence: divide-andconquer methods are a well known way to tackle large problems by dividing them into small subproblems. Decision trees, used for machine learning classification, successfully apply this approach. This thesis presents algorithms that build density estimators using a binary split tree to cut a range of values into subranges of varying length. No class values are required for this splitting process, making it an unsupervised method. The result is a histogram estimator that adapts well even to complex density functions a novel density estimation method with flexible density estimation ability and good computational behaviour. Algorithms are presented for both univariate and multivariate data. The univariate histogram estimator is applied to discretization for density estimation and also used as density estimator inside a naive Bayes classifier. The multivariate histogram, used as the basis for a clustering method, is applied to improve the runtime behaviour of a well-known algorithm for multiple-instance classification. Performance in these applications is evaluated by comparing the new approaches with existing methods.en_NZ
dc.format.mimetypeapplication/pdf
dc.language.isoen
dc.publisherThe University of Waikatoen_NZ
dc.rightsAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.
dc.subjectmachine learningen_NZ
dc.subjectdata miningen_NZ
dc.subjectdensity estimationen_NZ
dc.subjectmultivariate density estimationen_NZ
dc.subjecthistogramsen_NZ
dc.subjectmulti-dimensional histogramsen_NZ
dc.subjectnonparametric density estimationen_NZ
dc.subjectdiscretizationen_NZ
dc.subjectnaive Bayesen_NZ
dc.subjectclusteringen_NZ
dc.subjectmulti-dimensional clusteringen_NZ
dc.subjectsubspace clusteringen_NZ
dc.subjectmultiple-instance learningen_NZ
dc.subjectdata explorationen_NZ
dc.subjectdata visualizationen_NZ
dc.titleTree-based Density Estimation: Algorithms and Applicationsen_NZ
dc.typeThesisen_NZ
thesis.degree.disciplineComputer Scienceen_NZ
thesis.degree.grantorUniversity of Waikatoen_NZ
thesis.degree.levelDoctoral
uow.date.accession2009-09-08T16:05:42Zen_NZ
uow.date.available2009-09-11T14:44:49Zen_NZ
uow.identifier.adthttp://adt.waikato.ac.nz/public/adt-uow20090908.160542en_NZ
pubs.place-of-publicationHamilton, New Zealanden_NZ


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record