scholarly journals 1P223 The computational model of mutual interaction between noise and signal in the visual system(Chemoreception, neuron and sensory system, neural network, and brain computing,Oral Presentations)

2007 ◽  
Vol 47 (supplement) ◽  
pp. S79
Author(s):  
Sohei Wakisaka ◽  
Hiroyuki Ohta
2002 ◽  
Vol 44-46 ◽  
pp. 881-887 ◽  
Author(s):  
T. Koshizen ◽  
Koji Akatsuka ◽  
Hiroshi Tsujino

1993 ◽  
Vol 04 (01) ◽  
pp. 43-54 ◽  
Author(s):  
CHRISTOPHER HIAN-ANN TING

In the mammalian visual system, magnocellular pathway and parvocellular pathway cooperatively process visual information in parallel. The magnocellular pathway is more global and less particular about the details while the parvocellular pathway recognizes objects based on the local features. In many aspects, Neocognitron may be regarded as the artificial analogue of the parvocellular pathway. It is interesting then to model the magnocellular pathway. In order to achieve "rotation invariance" for Neocognitron, we propose a neural network model after the magnocellular pathway and expand its roles to include surmising the orientation of the input pattern prior to recognition. With the incorporation of the magnocellular pathway, a basic shift in the original paradigm has taken place. A pattern is now said to be recognized when and only when one of the winners of the magnocellular pathway is validified by the parvocellular pathway. We have implemented the magnocellular pathway coupled with Neocognitron parallel on transputers; our simulation programme is now able to recognize numerals in arbitrary orientation.


Sign in / Sign up

Export Citation Format

Share Document