Parallel algorithms and architectures for very fast AI search
A wide range of problems in natural and artificial intelligence, computer vision, computer graphics, database engineering, operations research, symbolic logic, robot manipulation and hardware design automation are special cases of Consistent Labeling Problems (CLP). CLP has long been viewed as an efficient computational model based on a unit constraint relation containing 2N-tuples of units and labels which specifies which N-tuples of labels are compatible with which N-tuples of units. Due to high computation cost and design complexity, most currently best-known algorithms and computer architectures have usually proven infeasible for solving the consistent labeling problems. Efficiency in CLP computation during the last decade has only been improved a few times. This research presents several parallel algorithms and computer architectures for solving CLP within a parallel processing framework. For problems of practical interest, 4 to 10 orders of magnitude of efficiency improvement can be easily reached. Several simple wafer scale computer architectures are given which implement these parallel algorithms at a surprisingly low cost.
- Research Organization:
- Utah Univ., Salt Lake City, UT (USA)
- OSTI ID:
- 5923908
- Resource Relation:
- Other Information: Thesis (Ph. D.)
- Country of Publication:
- United States
- Language:
- English
Similar Records
Parallel algorithms and architectures for the manipulator inertia matrix
Parallel algorithms and VLSI architectures for robotics and assembly scheduling
Related Subjects
ARTIFICIAL INTELLIGENCE
ALGORITHMS
COMPUTER ARCHITECTURE
COMPUTER GRAPHICS
DATA BASE MANAGEMENT
DESIGN
PARALLEL PROCESSING
RESEARCH PROGRAMS
ROBOTS
VISION
MANAGEMENT
MATHEMATICAL LOGIC
PROGRAMMING
990200* - Mathematics & Computers