A Neural Network for Arabic Character Recognition. | ||||
| MEJ- Mansoura Engineering Journal | ||||
| Article 2, Volume 18, Issue 1, March 1993, Page 61-71 PDF (469.58 K) | ||||
| Document Type: Research Studies | ||||
| DOI: 10.21608/bfemu.1993.164885 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| Ahmed El-Saeed Tolba Alam El-Din* | ||||
| Lecturer at Electrical Engineering Department., Faculty of Engineering., University of Suez Canal., Port Said., Egypt. | ||||
| Abstract | ||||
| Neural network-based learning is an essential part of any intelligent system and is an inherent property in Artificial Neural Network (ANN) models. Recently, artificial neural network models have begun to emerge as powerful tools for learning. Neural networks are designed and applied to the classification of isolated Arabic characters. We have studied the ability of networks to correctly classify both training and testing examples. Multilayered neural networks were trained to classify the characters using the error backpropagation learning algorithm. Noisy characters could be efficiently recognized. In this paper we evaluate the performance of the backpropagation technique on recognition of Arabic characters. Results indicate a high percentage of correct recognition and fault tolerance capability. The most effective number of neural network layers and the number of units in the bidden layers are conducted through extensive experimental work on isolated Arabic characters, Further research for recognition of connected hand written characters is going on. | ||||
|
Statistics Article View: 54 PDF Download: 113 |
||||
