| Titles |
|
English :
|
The udder milk problem : A new application for neural network
|
|
Arabic :
|
تطبيق جديد للشبكات العصبية الاصطناعية
|
|
| Abstract |
In this paper a new application area for Artificial Neural Network(ANN) has been presented. Udder and teat geometrical shape and milk machining characteristics are used to classify the ability of producing milk poor, normal and good )of Friesian cattle using Multi-Layer perceptron (MLP). A Generalized feed forward Network has done the job with 100% classification rate. The Network learning rate was good. Moreover, weights of the network and output variability with the input are used as a tool for feature selection in the measurement domain. Thus a much reduction in parameters normally used are achieved. The selected features are with much importance for selecting animals to repeat features for increase giving milk. Moreover, it shows that the season has a smaller effect on animals that are tagged good producer, in comparison with its effect on the other two classes. Thus measurements for this class can be considered over a shorter period of time rather than a year as it is usually done. Peoples of the application field appreciated these results.
Key Words : Artificial Neural Network, MLP, Momentum back-propagation algorithm, Generalized feed forward network, Udder Milk Problem
|
| Publication year |
2001
|
| Pages |
55-64
|
| Availability location |
معهد بحوث الانتاج الحيوانى- شارع نادى الصيد- الدقى - الجيزة
|
| Availability number |
|
| Organization Name |
Animal Production Research Institute (APRI)
|
| City |
القاهرة
|
| serial title |
Egyptian Information Journal
|
| Author(s) from ARC |
|
| External authors (outside ARC) |
على محمود ابو زيد
جامعة 6 اكتوبر
سامى هنداوى
جامعة 6 اكتوبر
|
| Agris Categories |
Animal husbandry
|
| Proposed Agrovoc |
Udder Milk Problem;Artificial Neural Network;MLP;Multi-Layer perceptron;Network learning rate;
|
| Publication Type |
Journal
|