Adaptive mixtures of local experts

  • Authors:
  • Robert A. Jacobs;Michael I. Jordan;Steven J. Nowlan;Geoffrey E. Hinton

  • Affiliations:
  • Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139 USA;Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139 USA;Department of Computer Science, University of Toronto, Toronto, Canada M5S 1A4;Department of Computer Science, University of Toronto, Toronto, Canada M5S 1A4

  • Venue:
  • Neural Computation
  • Year:
  • 1991

Quantified Score

Hi-index 0.02

Visualization

Abstract

We present a new supervised learning procedure for systems composed of many separate networks, each of which learns to handle a subset of the complete set of training cases. The new procedure can be viewed either as a modular version of a multilayer supervised network, or as an associative version of competitive learning. It therefore provides a new link between these two apparently different approaches. We demonstrate that the learning procedure divides up a vowel discrimination task into appropriate subtasks, each of which can be solved by a very simple expert network.