bluelogo.gif (1643 bytes)

home page
people
research
publications
seminars
travel
search
UCL

Evaluation of Adaptive Mixtures of Competing Experts

Steven Nowlan and Geoffrey E. Hinton
Department of Computer Science
University of Toronto

Abstract

We compare the performance of the modular architecture, composed of competing expert networks, suggested by Jacobs, Jordan, Nowlan and Hinton (1991) to the performance of a single back-propagation network on a complex, but low-dimensional, vowel recognition task. Simulations reveal that this system is capable of uncovering interesting decompositions in a complex task.  The type of decomposition is strongly influenced by the nature of the input to the gating network that decides which expert to use for each case.  The modular architecture also exhibits consistently better generalization on many variations of the task.

Download  [Postscript
] [pdf]

Advances in Neural Information Processing Systems 3 (1991). D.S. Touretzky, M.C. Mozer and M.E. Hasselmo. MIT Press.

[home page]  [publications]