Keywords : backpropagation


Personal Identification with Iris Patterns

Mazin R. Khalil; Mahmood S. Majeed; Raid R. Omar

AL-Rafidain Journal of Computer Sciences and Mathematics, 2009, Volume 6, Issue 1, Pages 13-26
DOI: 10.33899/csmj.2009.163762

This research is aimed to design an iris recognition system. There are two main steps to verify the goal. First: applying image processing techniques on the picture of an eye for data acquisition. Second: applying neural networks techniques for identification.
The image processing techniques display the steps for getting a very clear iris image necessary for extracting data from the acquisition of eye image. This picture contains all the eye (iris, pupil and lashes). So, the localization of the iris is very important. The new picture should be enhanced to bring out the pattern. The enhanced picture is segmented into 100 parts, then a standard Deviation (STD) can easily be computed for every part. These values will be used in the neural network for the identification.
For neural network techniques, Backprobagation neural network was used for comparisons. The weights and output values will be stored in a text file to be used later in identification. The Backprobagation network succeeded in identification and attained to (False Acceptance Rate = 10% - False Rejection Rate = 10%).
 

Integration Method with Backpropagation

Nidhal H. AL-Assady; Jamal S. Majeed; Shahbaa I. Khaleel

AL-Rafidain Journal of Computer Sciences and Mathematics, 2005, Volume 2, Issue 1, Pages 49-68
DOI: 10.33899/csmj.2005.164073

In this research, a new method is discovered (combined method) to accelerate the backpropagation network by using the expected values of source units for updating weights, we mean the expected value of unit by the sum of the output of the unit and its error term multiplied by the factor Beta to accelerate the algorithm and also adjust the value of learning coefficient continuously if the value of energy function E decreases the learning rate is increased by a factor , if the value of the energy function E increases , the value of the learning rate is decreased by a factor . To obtain the optimal weight with minimum iteration and minimum time, we applied a new method on many applications to prove the result of this method (pattern compression, encoding and recognition on Arabic, English digits and alphabetic.
 

Improvement the Back-propagation Technique

Nidhal H. AL-Assady; Baydaa I. Khaleel; Shahbaa I. Khaleel

AL-Rafidain Journal of Computer Sciences and Mathematics, 2004, Volume 1, Issue 2, Pages 127-151
DOI: 10.33899/csmj.2004.164115

Error backpropagation neural network (EBP)  used training algorithm for feedforward artificial neural networks (FFANNs). The main problem with the EBP algorithm that it is very slow and the converge to the optimal solution is not guaranteed. This problem leads to search for improvements to speed up this algorithm. In this research we use several methods to speed up the EBP algorithm. A many layer neural network was designed for building pattern compression system, encoding and recognition. We also used many methods to speed up this algorithm (EBP) and comparison between them.