Share Email Print
cover

Proceedings Paper

A Biological Plausible Neural Network Model for Processing Spatial Knowledge
Author(s): Oleg G. Jakubowicz
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

A multi-layer neural network model is presented which is able to distinguish distorted multi-feature input patterns in noisy environments. The model closely follows principles of the biological visual system and has several incarnationswhich have ha.d successful implementations in both the field of visual pattern recognition and situational analysis. The situational analysis version treats each of his input features as a different object and encodes the spatial relationships between the objects in various trained grouped patterns. Besides recognition the situational analysis model points out (i.e. completes) missing objects in incomplete testing inputs. The model is a simulation of the visual retino-cortex system and employs localized receptive fields, lateral inhibition in both learning and recognition modes, multiple resolution and self-organizing topographical mappings. Localized inputs and lateral inhibition incurs special problems in learning and these are described.

Paper Details

Date Published:
PDF: 8 pages
Proc. SPIE 1192, Intelligent Robots and Computer Vision VIII: Algorithms and Techniques, ; doi: 10.1117/12.969766
Show Author Affiliations
Oleg G. Jakubowicz, State University of New York at Buffalo (United States)


Published in SPIE Proceedings Vol. 1192:
Intelligent Robots and Computer Vision VIII: Algorithms and Techniques
David P. Casasent, Editor(s)

© SPIE. Terms of Use
Back to Top