Community Research and Development Information Service - CORDIS

Abstract

Researchers assessed the performance of the Multi Layer Perceptron (MLP) neural network in the automatic classification of defects found in strip steel. Experiments with MLP cluster patterns took two approaches: single layer MLPs were programmed to classify defects according to group, while double layer MLPs focused on identifying single defect classes. Both MLP formations proved equally effective in defect classification, showing that existing neural network techniques are efficient. Preliminary trials indicate that the double layer MLP formation may have greater accuracy potential once cluster patterns have been investigated more closely.

Additional information

Authors: FALESSI R; STOFELLA P, PENSINI M P; REMENTERIA ING S, MARIJUAN G, Centro Sviluppo Materiali SPA (IT); Advanced Computing Systems (GB); LABEIN
Bibliographic Reference: EUR 18410 EN (1998) 61pp.
Availability: Available from OOPEC Sales agents
ISBN: 92-828-4557-5
Record Number: 199910010 / Last updated on: 1999-02-12
Category: PUBLICATION
Original language: en
Available languages: en
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top