Behavior Self-Organization in Multi-Agent Learning.

Abstract

There are four primary results of the first year of the project: It was discovered that clustering algorithms for pre-sorting high-dimensional datasets was not effective in improving subsequent processing by reinforcement learning methods. It was discovered that Bayesian belief networks can be combined with decision nodes and an incremental assessment algorithm to mimic human patterns of data reduction and knowledge representation. The human immunological system was identified as a possible model for a "bidirectional" distributed decision network. Initial work has identified a model-balancing technique, borrowed from linear system theory, that is a strong candidate for a pruning and model reduction method for large modular networks.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 28, 1999
Accession Number
ADA366166

Entities

People

  • Hugh F. Vanlandingham
  • John S. Bay

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Clustering
  • Contracts
  • Control Systems
  • Data Processing
  • Data Reduction
  • Learning
  • Linear Systems
  • Machine Learning
  • Military Research
  • Organizational Structure
  • Reinforcement Learning
  • Robots
  • Self Organizing Systems
  • Signal Processing
  • Virginia

Fields of Study

  • Computer science

Readers

  • Artificial Intelligence
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks