Problem solving with insufficient resources

  • Authors:

  • Pei Wang

  • Affiliations:

  • Department of Computer and Information Sciences, Temple University, 1805 N. Broad Street, Philadelphia, Pennsylvania

  • Venue:
  • International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new approach, "controlled concurrency," is introduced for inference control in an adaptive reasoning system working with insufficient knowledge and resources. With this method, a problem-solving process is constructed from atomic steps in run time, according to the system's past experience and the current context. The system carries out many such processes in parallel by distributing its resources among them, and dynamically adjusting the distribution according to feedback. A data structure, "bag," is designed to support this dynamic time-space allocation, and is a kind of probabilistic priority queue. This approach provides a flexible, efficient, and adaptive control mechanism for real-time systems working with uncertain knowledge. To analyze problem solving in such a system, the traditional computability theory and computational complexity theory become inappropriate, because the system no longer follows problem-specific algorithms in problem solving.