A framework for multi-backpropagation
Backpropagation algorithm is one of the most popular learning algorithms in the Neural Network. It has been successfully implemented in many applications. However, training Neural Networks involve a large amount of data. Therefore, training the network is time consuming as each training session req...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Universiti Utara Malaysia
2003
|
| Subjects: | |
| Online Access: | https://repo.uum.edu.my/id/eprint/94/1/Fadzilah_Siraj_2.pdf https://repo.uum.edu.my/id/eprint/94/ http://ijms.uum.edu.my |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Backpropagation algorithm is one of the most popular learning algorithms in the Neural Network. It has been successfully implemented in many applications. However, training Neural Networks involve a large amount of data.
Therefore, training the network is time consuming as each training session requires several epochs, which usually takes smeral seconds or even minutes.This paper proposes a multi-backpropagation approach to minimize the complexity
of the network. The approach does not require an alteration of the algorithm. Instead, the large network is split into several smaller networks. An integrating network is then constructed to integrate the output from the smaller networks.
|
|---|