TY - GEN
T1 - Visual shape recognition neural network using BESOM model
AU - Hasegawa, Hiroaki
AU - Hagiwara, Masafumi
PY - 2010/11/8
Y1 - 2010/11/8
N2 - In this paper, we propose a neural network recognizing visual shapes based on the BidirEctional SOM (BESOM) model. The proposed network has 4 features. First, the network is based on the BESOM model, which is a computational model of the cerebral cortex. Second, the Gabor filter, a model of a simple cell in the primary visual area, is used to calculate input features. Third, the network structure mimics the ventral visual pathway of the brain, which is said to recognize visual shapes. Finally, this is the first application of the BESOM model which is large-scale and multi-layer as far as we know. We conducted an experiment to assess the network and confirmed that it can recognize alphabets.
AB - In this paper, we propose a neural network recognizing visual shapes based on the BidirEctional SOM (BESOM) model. The proposed network has 4 features. First, the network is based on the BESOM model, which is a computational model of the cerebral cortex. Second, the Gabor filter, a model of a simple cell in the primary visual area, is used to calculate input features. Third, the network structure mimics the ventral visual pathway of the brain, which is said to recognize visual shapes. Finally, this is the first application of the BESOM model which is large-scale and multi-layer as far as we know. We conducted an experiment to assess the network and confirmed that it can recognize alphabets.
UR - http://www.scopus.com/inward/record.url?scp=78049391550&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78049391550&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-15825-4_11
DO - 10.1007/978-3-642-15825-4_11
M3 - Conference contribution
AN - SCOPUS:78049391550
SN - 3642158242
SN - 9783642158247
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 102
EP - 105
BT - Artificial Neural Networks, ICANN 2010 - 20th International Conference, Proceedings
T2 - 20th International Conference on Artificial Neural Networks, ICANN 2010
Y2 - 15 September 2010 through 18 September 2010
ER -