Publication details

Number of components and initialization in Gaussian mixture model for pattern recognition

Conference Paper (international conference)

Paclík P., Novovičová Jana


serial: Artificial Neural Nets and Genetic Algorithms. Proceedings, p. 406-409 , Eds: Kůrková J., Neruda R., Kárný M., Steele N. C.

publisher: Springer, (Wien 2001)

action: International Conference on Artificial Neural Nets and Genetic Algorithms /5./, (Prague, CZ, 22.04.2001-25.04.2001)

research: AV0Z1075907

project(s): VS96063, GA MŠk, KSK1075601, GA AV ČR

keywords: pattern recognition, Gaussian mixture model, kernel density estimate

abstract (eng):

The method for complete mixture initialization based on a product kernel estimate of probability density function is proposed for mixture estimation using EM-algorithm. The mixture components are assumed to correspond to local maxima of optimaly smoothed kernel density estimate. The gradient method is used for local extrema finding. As the last step, agglomerative hiearchical clustering methods merges closest components together. A comparison to scale-space approaches is given on examples.

Cosati: 12B, 09K

RIV: BB