Training Optimization for Artificial Neural Networks
Nowadays, with the capacity to model complex problems, the artificial Neural Networks (nn) are very popular in the areas of Pattern Recognition, Data Mining and Machine Learning. Nevertheless, the high computational cost of the learning phase when big data bases are used is their main disadvantage....
Gespeichert in:
Veröffentlicht in: | CIENCIA ergo-sum 2010-11, Vol.17 (3), p.313-317 |
---|---|
Hauptverfasser: | , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Nowadays, with the capacity to model complex problems, the artificial Neural Networks (nn) are very popular in the areas of Pattern Recognition, Data Mining and Machine Learning. Nevertheless, the high computational cost of the learning phase when big data bases are used is their main disadvantage. This work analyzes the advantages of using pre-processing in data sets In order to diminish the computer cost and improve the nn convergence. Specifically the Relative Neighbor Graph (rng), Gabriel’s Graph (gg) and k-ncn methods were evaluated. The experimental results prove the feasibility and the multiple advantages of these methodologies to solve the describedproblems.
Debido a la habilidad para modelarproblemas complejos, actualmente las RedesNeuronales Artificiales (nn) son muy popularesen Reconocimiento de Patrones, Minería deDatos y Aprendizaje Automático. No obstante,el elevado costo computacional asociado a lafase en entrenamiento, cuando grandes bases dedatos son utilizados, es su principal desventaja.Con la intención de disminuir el costocomputacional e incrementar la convergencia dela nn, el presente trabajo analiza la convenienciade realizar pre-procesamiento a los conjuntosde datos. De forma específica, se evalúan losmétodos de grafo de vecindad relativa (rng),grafo de Gabriel (gg) y el método basado enlos vecinos envolventes k-ncn. Los resultadosexperimentales muestran la factibilidad y lasmúltiples ventajas de esas metodologías parasolventar los problemas descritos previamente. |
---|---|
ISSN: | 1405-0269 |