Statistical and Numerical Techniques of Classification of Acoustic Sources: Generalized Phi-Divergence Applications in Acoustic Emission

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21340%2F11%3A00187900" target="_blank" >RIV/68407700:21340/11:00187900 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Statistical and Numerical Techniques of Classification of Acoustic Sources: Generalized Phi-Divergence Applications in Acoustic Emission

Popis výsledku v původním jazyce

We focus on the classification of acoustic emission signals by means of distribution mixtures (DM). The acoustic signals are separated by suitable parameters obtained directly from the signals and from the normed frequency spectra. The phi-divergence distance measures are employed as the additional signal spectrum attribute. We deal with a simple method of construction of phi-divergences and we introduce several modifications such as generalized LeCam, Hellinger, and Breigman divergences. We are concerned with the efficient set of classification parameters while testing the quality of classification. We combine both the main approaches described above, i.e. the generalized phi-divergences and the distribution mixture method. The advantage of the combined method is that it is able to assess the number of clusters of the signals and simultaneously it is robust in the sense that it ignores sparse outliers that would distort either the standard statistical estimates or classical non-statis

Název v anglickém jazyce

Statistical and Numerical Techniques of Classification of Acoustic Sources: Generalized Phi-Divergence Applications in Acoustic Emission

Popis výsledku anglicky

We focus on the classification of acoustic emission signals by means of distribution mixtures (DM). The acoustic signals are separated by suitable parameters obtained directly from the signals and from the normed frequency spectra. The phi-divergence distance measures are employed as the additional signal spectrum attribute. We deal with a simple method of construction of phi-divergences and we introduce several modifications such as generalized LeCam, Hellinger, and Breigman divergences. We are concerned with the efficient set of classification parameters while testing the quality of classification. We combine both the main approaches described above, i.e. the generalized phi-divergences and the distribution mixture method. The advantage of the combined method is that it is able to assess the number of clusters of the signals and simultaneously it is robust in the sense that it ignores sparse outliers that would distort either the standard statistical estimates or classical non-statis

Druh

O - Ostatní výsledky

CEP obor

BB - Aplikovaná statistika, operační výzkum

OECD FORD obor

—

Projekt

—

Návaznosti

Z - Vyzkumny zamer (s odkazem do CEZ)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2011

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů