Project Title:

A Hybrid, Neural-Network and Expert-System Approach to Remote Sensing

07.07-3474
911596
A Hybrid, Neural-Network and Expert-System Approach to Remote Sensing
Charles River Analytics, Inc.
55 Wheeler Street
Cambridge
MA
02138
James M.
Mazzu
617-491-3474
MSFC
NAS8-39310
131

07.07-3474
911596

Abstract:

A Hybrid, Neural-Network and Expert-System Approach to Remote Sensing
The integration of artificial neural networks (ANNs) and knowledge-based expert systems
is an ideal step in the development of intelligent systems. In general, the two methods
complement each other such that ANNs provide "soft" constraints, while expert systems
allow "hard" constrains. Specifically, ANNs can perform nonlinear functions, pattern
recognition, fault-tolerance, and parallel processing; while expert systems involve
language processing, formal logic, and rule interpretation. The complementary strengths
of neural-networks and knowledge-based expert systems will be used to create a hybrid
remote sensing system that can outperform either method alone. An advanced remote
sensing system within a hybrid neural-network/expert system environment will be developed
to be incorporated in the Geostationary Earth Observatory. The system will take advantage
of both ANNs and expert systems to handle global and local event isolation and identification,
measurement validation, instrumentation control, and information storage. The hybrid
remote sensing system will be implemented within the hybrid NueX shell, providing
a visual object-oriented development environment for real-time remote sensing applications.
The commercial application will be a tool for a hybrid remote sensing system applications
by aerospace prime and subcontractors, government, and university research laboratories.
neural networks, expert systems, remote sensing, adaptive, hybrid