Neural Network Arhitecture Definition in Image Recognition Tasks
2007
Maksims Alekseičevs, Aleksandrs Glazs

In this article the results of the research of committee methods in recognition neural networks are presented, applying to human faces recognition. Problem is specificity consists of similar images recognition (in particular, faces of twins). It is impossible to achieve training sample’s 100% recognition. Therefore committee decision with weighted voting algorithm for classification is suggested.


Keywords
seju atpazīšana, neironu tīkls, komiteju lēmums, vienlīdzīgās balsošanas algoritms, svērtais balsošanas algoritms

Alekseičevs, M., Glazs, A. Neural Network Arhitecture Definition in Image Recognition Tasks. Technologies of Computer Control. Vol.32, 2007, pp.22-28. ISSN 1407-7493.

Publication language
Latvian (lv)
The Scientific Library of the Riga Technical University.
E-mail: uzzinas@rtu.lv; Phone: +371 28399196