Artificial neural networks (ANNs) serve as powerful computational tools in a diversity of applications including: classification, pattern recognition, function approximation, and the modeling of biological neural networks. Equipped with procedures for learning from examples, ANNs can solve problems for which no algorithmic solution is known. In this monograph, the authors introduce a novel FRB, referred to as the Fuzzy All-permutations Rule-Base (FARB). They showthat inferring the FARB, using standard tools from fuzzy logic theory, yieldsan inputoutput relationship that is mathematically equivalent to that of an ANN. Conversely, every standard ANN has an equivalent FARB. They provide the explicit bidirectional transformation between the ANN and the corresponding FARB. Presents the state of the art in knowledge-based neurocomputing Presents a new connection between artificial neural networks (ANNs) and a special fuzzy rule-base - the all permutations fuzzy rule-base (FARB) INDICE: Introduction.- The FARB.- The FARB-ANN Equivalence.- Rule Simplification.- Knowledge Extraction Using the FARB.- Knowledge-Based Design of ANNs.- Conclusions and Future Research.- A Proofs.- B Details of the LED Recognition Network.
- ISBN: 978-3-540-88076-9
- Editorial: Springer
- Encuadernacion: Cartoné
- Páginas: 145
- Fecha Publicación: 01/01/2009
- Nº Volúmenes: 1
- Idioma: Inglés