Constrained neural network architectures for target recognition

Conference · Wed Jan 01 00:00:00 EST 1992

OSTI ID:5693906

Hush, D R; Clark, Shang-Ying ^[1]; Moya, M M ^[2]

---

1. New Mexico Univ., Albuquerque, NM (United States). Dept. of Electrical and Computer Engineering
2. Sandia National Labs., Albuquerque, NM (United States)

This paper describes several different types of constraints that can be placed on multilayered feedforward neural networks which are used for automatic target recognition (ATR). We show how unconstrained networks are likely to give poor generalization on the ATR problem. We also show how the ATR problem requires a special type of classifier called a one-class classifier. The network constraints come in two forms: architectural constraints and learning constraints. Some of the constraints are used to improve generalization, while others are incorporated so that the network will be forced to perform one-class classification. 14 refs

• Research Organization: Sandia National Labs., Albuquerque, NM (United States)
• Sponsoring Organization: USDOE; USDOE, Washington, DC (United States)
• DOE Contract Number: AC04-76DP00789
• OSTI ID: 5693906
• Report Number(s): SAND-92-0648C; CONF-920471-3; ON: DE92011159
• Resource Relation: Conference: International Society for Photo Optical Engineering (SPIE) conference, Orlando, FL (United States), 20-24 Apr 1992
• Country of Publication: United States
• Language: English

Related Subjects

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEFENSE
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
MILITARY EQUIPMENT
PATTERN RECOGNITION
NEURAL NETWORKS
ARTIFICIAL INTELLIGENCE
IMAGE PROCESSING
IMAGES
TARGETS
VEHICLES
EQUIPMENT
PROCESSING
450000* - Military Technology
Weaponry
& National Defense
990200 - Mathematics & Computers