Unsupervised learning in second-order neural networks for motion analysis
This paper demonstrates how unsupervised learning based on Hebb-like mechanisms is sufficient for training second-order neural networks to perform different types of motion analysis. The paper studies the convergence properties of the network in several conditions, including different levels of nois...
Saved in:
Main Authors: | , |
---|---|
Format: | Article |
Published: |
Elsevier
2011
|
Subjects: | |
Online Access: | http://eprints.um.edu.my/5670/ https://doi.org/10.1016/j.neucom.2010.09.023 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
id |
my.um.eprints.5670 |
---|---|
record_format |
eprints |
spelling |
my.um.eprints.56702019-11-15T03:44:56Z http://eprints.um.edu.my/5670/ Unsupervised learning in second-order neural networks for motion analysis Maul, Tomas Baba, Mohd Sapiyan QA75 Electronic computers. Computer science T Technology (General) This paper demonstrates how unsupervised learning based on Hebb-like mechanisms is sufficient for training second-order neural networks to perform different types of motion analysis. The paper studies the convergence properties of the network in several conditions, including different levels of noise and motion coherence and different network configurations. We demonstrate the effectiveness of a novel variability dependent learning mechanism, which allows the network to learn under conditions of large feature similarity thresholds, which is crucial for noise robustness. The paper demonstrates the particular relevance of second-order neural networks and therefore correlation based approaches as contributing mechanisms for directional selectivity in the retina. Elsevier 2011 Article PeerReviewed Maul, Tomas and Baba, Mohd Sapiyan (2011) Unsupervised learning in second-order neural networks for motion analysis. Neurocomputing, 74 (6). pp. 884-895. ISSN 0925-2312 https://doi.org/10.1016/j.neucom.2010.09.023 doi:10.1016/j.neucom.2010.09.023 |
institution |
Universiti Malaya |
building |
UM Library |
collection |
Institutional Repository |
continent |
Asia |
country |
Malaysia |
content_provider |
Universiti Malaya |
content_source |
UM Research Repository |
url_provider |
http://eprints.um.edu.my/ |
topic |
QA75 Electronic computers. Computer science T Technology (General) |
spellingShingle |
QA75 Electronic computers. Computer science T Technology (General) Maul, Tomas Baba, Mohd Sapiyan Unsupervised learning in second-order neural networks for motion analysis |
description |
This paper demonstrates how unsupervised learning based on Hebb-like mechanisms is sufficient for training second-order neural networks to perform different types of motion analysis. The paper studies the convergence properties of the network in several conditions, including different levels of noise and motion coherence and different network configurations. We demonstrate the effectiveness of a novel variability dependent learning mechanism, which allows the network to learn under conditions of large feature similarity thresholds, which is crucial for noise robustness. The paper demonstrates the particular relevance of second-order neural networks and therefore correlation based approaches as contributing mechanisms for directional selectivity in the retina. |
format |
Article |
author |
Maul, Tomas Baba, Mohd Sapiyan |
author_facet |
Maul, Tomas Baba, Mohd Sapiyan |
author_sort |
Maul, Tomas |
title |
Unsupervised learning in second-order neural networks for motion analysis |
title_short |
Unsupervised learning in second-order neural networks for motion analysis |
title_full |
Unsupervised learning in second-order neural networks for motion analysis |
title_fullStr |
Unsupervised learning in second-order neural networks for motion analysis |
title_full_unstemmed |
Unsupervised learning in second-order neural networks for motion analysis |
title_sort |
unsupervised learning in second-order neural networks for motion analysis |
publisher |
Elsevier |
publishDate |
2011 |
url |
http://eprints.um.edu.my/5670/ https://doi.org/10.1016/j.neucom.2010.09.023 |
_version_ |
1651867325520936960 |
score |
13.1944895 |