The proposed hybrid approach involves a close integration of high-performance technologies - knowledge-based techniques, neural nets, and parallel hardware are blended together with conventional algorithmic techniques in a well-integrated system architecture. This combination is particularly significant for automatic target recognition (ATR) because the overall target recognition problem involves both pattern recognition and reasoning. The design allows various system functions to utilize the technology that provides the best overall performance. Several innovations are used to achieve these goals. Among these is the use of neural nets at a 'core' level inside the knowledge-based system. The significance of these innovations is that they will, if successful, produce a powerful architecture for building ATR and other high-performance intelligent systems. The emphasis is exclusively on techniques that can be implemetned in real time on modest size, embedded computers, and is specifically oriented towards the need of military applications.
Hybrid approach to automatic target recognition utilizing artificial intelligence and neural nets
Eine hybride Lösung für die automatische Zielerkennung unter Verwendung künstlicher Intelligenz und neuronaler Netze
1989
8 Seiten, 2 Bilder, 7 Quellen
Conference paper
English
Automatic target recognition fuses sensors and artificial intelligence
Tema Archive | 1984
|Neural networks for automatic target recognition
Tema Archive | 1990
|Emerald Group Publishing | 2017
|Gas Metal Arc Penetration Welding Development Utilizing Neural Nets
British Library Conference Proceedings | 1997
|