Publications

sort by:  author date

Comparison of ART-2 and SOFM based neural network verifiers

MAUTNER, P., MATOUŠEK, V., MARŠÁLEK, T., ROHLÍK, O. Comparison of ART-2 and SOFM based neural network verifiers. In Artificial neural networks: data preparation techniques and application development. Setúbal : INSTICC PRESS, 2004. s. 43-50. ISBN: 972-8865-14-7
Abstract PDF BibTeX

The Carpenter-Grosberg ART-2 and Kohonen Self-organizing Feature Map (SOFM) have been developed for the clustering of input vectors and have been commonly used as unsupervised learned clasifiers. In this paper we describe the use of these neuralnetwork models for signature verification. The architecture of the verifiers and achieved results are discussed here and ideas for future research are also suggested.

Morphological filtration for time series

VÁVRA, F., NOVÝ, P., MAŠKOVÁ, H., VOJTÍŠKOVÁ, M., NETRVALOVÁ, A. Morphological filtration for time series. In Aplimat 2004. Bratislava: Slovak University of Technology, 2004. s. 983-989. ISBN: 80-227-1995-1
Abstract PDF BibTeX

In presented article, there will be introduced one possible generalization of using the morphological methods for time series smoothing and filtration. The base of morphological apparatus is the min-max algebra. Presented methodology was developed for the trend analysis in time series of foreign currencies.

Signature verification using self-organizing feature map

MAUTNER, P., MATOUŠEK, V., ROHLÍK, O., KEMPF, J. Signature verification using self-organizing feature map. In CIRAS 2003. Singapore: National University of Singapore, 2003. s. 1-4.
Abstract PDF BibTeX

In this paper we describe the use of the SOFM neural network model for signature verification. The biometric data of all signatures were acquired by a special digital data acquisition pen and fast wavelet transformation was used forfeature extraction. The part of authentic signature data was used for training the SOFM signature verifer. The architecture of the verifer and achieved results are discussed here and ideas for future research are also suggested.

Novel communication concepts for municipal information services

EKŠTEIN, K., HITZENBERGER, L., KLEČKOVÁ, J., KRUTIŠOVÁ, J., KUBIŠTA, J., MATOUŠEK, V., MOUČEK, R., TAUŠER, K. Novel communication concepts for municipal information services. In SoftCOM 2003. Split: Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture University of Split, 2003. s. 705-709. ISBN: 953-6114-64-X
Abstract PDF BibTeX

In this paper we report on some problems arising during the building of a bilingual dialogue system, concerning especially robust speech recognition method, linguistic analysis of incomplete utterances and their sementic interpretation, dialogue modelling and dialogue control, system response generation, as well as a unique knowledge based technique of automated generation of a corpus of training sentences.

Hybrid analytic/ANN-based acoustic-phonetic decoding

EKŠTEIN, K., MATOUŠEK, V., MOUČEK, R. Hybrid analytic/ANN-based acoustic-phonetic decoding. In Elektronische Sprachsignalverarbeitung. Dresden: Universitätsverlag & Buchhandel, 2003. s. 14-21. ISBN: 3-935712-83-9
Abstract PDF BibTeX

The presented article introduces an alternative acoustic-phonetic decoding scheme based on hybrid configuration of multilayer perceptron (MLP) neural network and knowledge-based analytic speech signal survey module.

Our partners