Izboljšave Gaussovih mešanih modelov za klasifikacijo na področju obratovalne trdnosti
doktorsko delo
Abstract
Doktorsko delo predstavlja rezultate ocenjevanja Gaussovega mešenega modela za potrebe klasifikacije v obratovalni trdnosti. Predstavljene so REBMIX & EM strategije za ocenjevanje parametrov Gaussovega mešanega modela. Ocenjevanje parametrov temelji na kombinaciji algoritma REBMIX in EM. REBMIX algoritem na podlagi znane empirične gostote porazdelitve verjetnosti za določen vzorec opazovanj oceni aproksimativne začetne parametre Gaussovega mešanega modela, ki jih nato izboljšamo s pomočjo EM algoritma. Za oceno empirične gostote porazdelitve verjetnosti, ki jo potrebujemo v algoritmu REBMIX, smo predstavili več možnih rešitev. To je prineslo tri različne REBMIX & EM strategije, in sicer Izčrpna strategija REBMIX & EM, Najboljša strategija REBMIX & EM ter Enotna strategija REBMIX & EM. Za oceno optimalnega histograma, ki ga rabimo pri ocenjevanju empirične gostote porazdelitve verjetnosti, smo prikrojili lasten optimizacijski algoritem. Ta temelji na algoritmu koordinatnega spusta. Za cenilko uspešnosti histograma je bilo izbrano Knuthovo pravilo. Za potrebe klasifikacije smo naredili dodaten tristopenjski postopek filtriranja nepomembnih lastnosti. Hkrati smo nakazali način kako lahko izboljšamo uspešnost klasifikacije, če vplivamo na parameter glajenja v histogramskem predprocesiranju REBMIX algoritma.
Keywords
obratovalna trdnost;Gaussov mešan model;REBMIX algoritem;EM algoritem;histogrami;klasifikacija;zaznavanje napak;segmentacija slik;disertacije;
Data
| Language: | Slovenian |
|---|---|
| Year of publishing: | 2021 |
| Typology: | 2.08 - Doctoral Dissertation |
| Organization: | UL FS - Faculty of Mechanical Engineering |
| Publisher: | [B. Panić] |
| UDC: | 620.178.3:519.218.7:311.13(043.3) |
| COBISS: |
76948227
|
| Views: | 676 |
| Downloads: | 66 |
| Average score: | 0 (0 votes) |
| Metadata: |
|
Other data
| Secondary language: | English |
|---|---|
| Secondary title: | Improvements of Gaussian mixture models for classification of operational strength problems |
| Secondary abstract: | The thesis presents the results of the evaluation of the Gaussian mixture model for the purpose of classification in operational strength. REBMIX & EM strategies for estimating the parameters of a Gaussian mixture model are presented. The parameter estimation is based on a combination of the REBMIX and EM algorithms. The REBMIX algorithm estimates the approximate initial parameters of the Gaussian mixture model based on the known empirical probability density distribution for a given sample of observations, which are then improved using the EM algorithm. To estimate the empirical probability density required for the REBMIX algorithm, we presented several possible solutions. This resulted in three different REBMIX and EM strategies, namely Exhaustive REBMIX & EM strategy, Best REBMIX & EM strategy, and Single REBMIX & EM. The histogram is used to estimate the empirical probability density. To estimate the optimal histogram, we have developed our own optimization algorithm based on the coordinate descent algorithm. Knuth's rule was chosen to evaluate the performance of the histogram. For classification, we have developed an additional three-step procedure to philtre out irrelevant features. At the same time, we have shown a way to improve the classification performance by influencing the smoothing parameter in the histogram preprocessing of the REBMIX algorithm. |
| Secondary keywords: | Gaussian mixture model;REBMIX algorithm;EM algorithm;histogram;classification;fault detection;image segmentation;dissertations;Trdnost;Disertacije; |
| Type (COBISS): | Doctoral dissertation |
| Study programme: | 0 |
| Embargo end date (OpenAIRE): | 1970-01-01 |
| Thesis comment: | Univ. v Ljubljani, Fak. za strojništvo |
| Pages: | XXVIII, 156 str. |
| ID: | 13447112 |
Recommended works: