haku: @keyword model selection / yhteensä: 6
viite: 2 / 6
Tekijä: | Adhikari, Prem Raj |
Työn nimi: | Mixture modelling of multiresolution 0-1 data |
Julkaisutyyppi: | Diplomityö |
Julkaisuvuosi: | 2010 |
Sivut: | xi + 80 s. + liitt. Kieli: eng |
Koulu/Laitos/Osasto: | Tietotekniikan laitos |
Oppiaine: | Informaatiotekniikka (T-61) |
Valvoja: | Kaski, Samuel |
Ohjaaja: | Hollmén, Jaakko |
OEVS: | Sähköinen arkistokappale on luettavissa Aalto Thesis Databasen kautta.
Ohje Digitaalisten opinnäytteiden lukeminen Aalto-yliopiston Harald Herlin -oppimiskeskuksen suljetussa verkossaOppimiskeskuksen suljetussa verkossa voi lukea sellaisia digitaalisia ja digitoituja opinnäytteitä, joille ei ole saatu julkaisulupaa avoimessa verkossa. Oppimiskeskuksen yhteystiedot ja aukioloajat: https://learningcentre.aalto.fi/fi/harald-herlin-oppimiskeskus/ Opinnäytteitä voi lukea Oppimiskeskuksen asiakaskoneilla, joita löytyy kaikista kerroksista.
Kirjautuminen asiakaskoneille
Opinnäytteen avaaminen
Opinnäytteen lukeminen
Opinnäytteen tulostus
|
Sijainti: | P1 Ark Aalto 6858 | Arkisto |
Avainsanat: | mixture models multiresolution data 0-1 data model selection cross-validation chromosomal aberration upsampling downsampling cancer genetics |
Tiivistelmä (eng): | Biological systems are complex and measurements in biology are made with high throughput and high resolution techniques often resulting in data in multiple resolutions. Furthermore, ISCN [1] has defined five different resolutions of the chromosome band. Currently, available standard algorithms can only handle data in one resolution at a time. Hence, transformation of the data to the same resolution is inevitable before the data can be fed to the algorithm. Furthermore comparing the results of an algorithm on data in different resolutions can produce interesting results which aids in determining suitable resolution of data. In addition, experiments in different, resolutions can be helpful in determining the appropriate resolution for computational methods. In this thesis, one method for up sampling and three different methods of down sampling 0-1 data are proposed, implemented and experiments are performed on different resolutions. Suitability of the proposed methods is validated and the results are compared across different resolutions. The proposed methods produce plausible results showing that the significant patterns in the data are retained in the transformed resolution. Thereafter, the mixture models are trained on the data original data and the results are analyzed. However, machine learning methods such as mixture models require high amounts of data to produce plausible results. Therefore, the major aim of the data transformation procedure was the integration of databases. Hence, two different datasets available in two different resolutions were integrated after transforming them to a single resolution and mixture models were trained on them. Trained models can be used to classify cancers and cluster the data. The results on integrated data showed significant improvements compared with the data in the original resolution. |
ED: | 2011-01-18 |
INSSI tietueen numero: 41488
+ lisää koriin
INSSI