DocumentCode :
1818646
Title :
Encoding and recall of spatiotemporal images with Oscillatron-1
Author :
Atkins, Mark A.
Author_Institution :
Accurate Automation Corp., Chattanooga, TN, USA
Volume :
1
fYear :
1992
fDate :
7-11 Jun 1992
Firstpage :
459
Abstract :
As a result of investigating a neural network architecture based on neuron synchronization theory, a concept known as Oscillatron-1 has been developed using a binary Hopfield net for its content addressable memory. A sequence of images representing a moving straight line was input to the network, which was able to encode instantly the sequence exactly as perceived. The moving images could then be recalled from memory by inputting one or more clues, whether the clues were visual or temporal. This architecture is potentially capable of teleologically manipulating arbitrary images and abstractions that presumably constitute the data structures of conscious thought
Keywords :
Hopfield neural nets; content-addressable storage; data structures; encoding; image processing; Oscillatron-1; abstractions; binary Hopfield net; content addressable memory; data structures; neural network architecture; neuron synchronization theory; recall; spatiotemporal images; teleologically manipulating arbitrary images; Associative memory; Automation; Clocks; Computer vision; Detectors; Hebbian theory; Image coding; Neurons; Spatiotemporal phenomena; Synchronization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks, 1992. IJCNN., International Joint Conference on
Conference_Location :
Baltimore, MD
Print_ISBN :
0-7803-0559-0
Type :
conf
DOI :
10.1109/IJCNN.1992.287170
Filename :
287170
Link To Document :
بازگشت