Theory and Algorithms for Memory Based Inference

Abstract

The objectives of this project were to develop new methods for analyzing memory-based techniques in machine learning and general inference problems. The project successfully developed new mathematical results explaining when memory-based methods are useful. Several applications were developed and studied in this project. A system for building transportable agents was developed and several information processing systems have been based on it. Technology has been successfully transitioned to both DoD and commercial entities.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 22, 1997
Accession Number
ADA342323

Entities

People

  • George Cybenko

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Algorithms
  • Computational Science
  • Computer Networks
  • Computer Programs
  • Computers
  • Digital Communications
  • Engineering
  • Information Processing
  • Information Systems
  • Knowledge Management
  • Linear Algebra
  • Machine Learning
  • Networks
  • Neural Networks
  • Object Recognition
  • Recognition
  • Signal Processing

Readers

  • Computational Linguistics
  • Fire Suppression Systems Design.
  • Theoretical Analysis.

Technology Areas

  • AI & ML