Multi-resolution Self-organizing Neural Networks for Pattern Recognition
Author | : Penny Pei Chen |
Publisher | : |
Total Pages | : |
Release | : 2001 |
ISBN-10 | : OCLC:47779142 |
ISBN-13 | : |
Rating | : 4/5 (42 Downloads) |
Book excerpt: The MRF-ART neural network employs fast competitive learning and efficient parallel matching to solve complex data classification problems. The architecture of the MRF-ART not only preserves the ART-type neural network's characteristics but also extends its capability to represent input patterns in a hierarchical fashion. To achieve this, the MRF-ART network uses multiple output layers arranged in a cascaded manner which is completely different from a conventional fuzzy ART network with only one output layer. Moreover, the parallel matching process makes the MRF-ART network suitable for hardware implementation.