dc.contributor.author | Cai, Yun | |
dc.date.accessioned | 2014-08-25T18:23:51Z | |
dc.date.available | 2014-08-25T18:23:51Z | |
dc.date.issued | 2014-08-25 | |
dc.identifier.uri | http://hdl.handle.net/10222/54041 | |
dc.description.abstract | Non-Negative Matrix Factorization (NMF) is a very useful tool to reduce dimension of data in machine learning and data mining. But it is an unsupervised learning method. To extract more discriminant information in the training data and improve the performance of classification, we developed a new supervised NMF method. We combine feature matrices from different classes as the feature matrix for the whole data and fit a non-negative Poisson regression to calculate the weight matrix. Our method is tested on the animal dataset and moving picture dataset. The experimental results show that our supervised NMF could greatly enhance the performance of NMF for classification. | en_US |
dc.language.iso | en | en_US |
dc.subject | Microbiome Data | en_US |
dc.subject | NMF | en_US |
dc.title | Supervised Non-negative Matrix Factorization for Analysis of Microbiome Data | en_US |
dc.type | Thesis | en_US |
dc.date.defence | 2014-08-21 | |
dc.contributor.department | Department of Mathematics & Statistics - Statistics Division | en_US |
dc.contributor.degree | Master of Science | en_US |
dc.contributor.external-examiner | n/a | en_US |
dc.contributor.graduate-coordinator | David Hamilton | en_US |
dc.contributor.thesis-reader | Bruce Smith | en_US |
dc.contributor.thesis-reader | Chris Field | en_US |
dc.contributor.thesis-supervisor | Hong Gu and Toby Kenney | en_US |
dc.contributor.ethics-approval | Not Applicable | en_US |
dc.contributor.manuscripts | Not Applicable | en_US |
dc.contributor.copyright-release | Not Applicable | en_US |