Training multilayered perceptrons for pattern recognition: a comparative study of four training algorithms


Pham D., Sagiroglu Ş.

INTERNATIONAL JOURNAL OF MACHINE TOOLS & MANUFACTURE, cilt.41, sa.3, ss.419-430, 2001 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 41 Sayı: 3
  • Basım Tarihi: 2001
  • Doi Numarası: 10.1016/s0890-6955(00)00073-0
  • Dergi Adı: INTERNATIONAL JOURNAL OF MACHINE TOOLS & MANUFACTURE
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.419-430
  • Anahtar Kelimeler: multilayered perceptrons, training algorithms, control chart pattern recognition, wood inspection, IMAGES
  • Gazi Üniversitesi Adresli: Hayır

Özet

This paper presents an overview of four algorithms used for training multilayered perceptron (MLP) neural networks and the results of applying those algorithms to teach different MLPs to recognise control chart patterns and classify wood veneer defects. The algorithms studied are Backpropagation (BP), Quick-prop (QP), Delta-Bar-Delta (DBD) and Extended-Delta-Bar-Delta (EDBD). The results show that, overall, BP was the best algorithm for the two applications tested. (C) 2001 Elsevier Science Ltd. All rights reserved.