ART-Based Model Set for Pattern Recognition: FasArt Family

  1. Sainz Palmero, G. I. 4
  2. Dimitriadis, Y. A. 1
  3. Cano Izquierdo, J. M. 2
  4. Gómez Sánchez, E. 1
  5. Parrado Hernández, E. 3
  1. 1 Department of Signal Theory, Communications and Telematics Engineering, School of Telecommunications Engineering, University of Valladolid, Camino del Cementerio s/n, 47011 Valladolid, Spain
  2. 2 Department of Systems Engineering and Control, School of Industrial Engineering, Polytechnical University of Cartagena, Campus de Muralla del Mar s/n, 30203 Cartagena, Spain
  3. 3 Department of Communication Technologies, Universidad Carlos III, Avenida de la Universidad 30, 28930 Leganés, Madrid, Spain
  4. 4 Department of Systems Engineering and Control, School of Industrial Engineering, University of Valladolid, Paseo del Cauce s/n, 47011 Valladolid, Spain
Neuro-Fuzzy Pattern Recognition

ISSN: 1793-0839

Year of publication: 2000

Pages: 145-175

Type: Book chapter

DOI: 10.1142/9789812792204_0007 GOOGLE SCHOLAR lock_openOpen access editor


This chapter introduces a new family of neuro-fuzzy systems suitable for pattern recognition. These architectures are based on the Adaptive Resonance Theory (ART) but introducing formalisms from the Fuzzy Sets theory, in order to solve some theoretical leaks present in Fuzzy ART based models. As a main result, a duality between neural network and fuzzy system can be seen in the proposed FasArt family architectures. The FasArt model is the kernel model of this family. The rest of the models have been developed in order to cope with particular features of typical pattern recognition problems: RFasArt is a recurrent version of FasArt that works with pattern structured sequences and it has been applied to the document recognition problem; FasArt with the combination of the STORE memory module has been applied to the problem of on-line handwriting recognition, which concerns sequences of subpatterns. Distributed FasArt deals with the problem of category proliferation present in all ART systems, such as FasArt family. The proposed architectures are applied to several important pattern recognition problems, which are also described.