An Alternative Model for Mixtures of Experts
Abstract
We propose an alternative model for mixtures of experts which uses a different parametric form for the gating network. The modified model is trained by the EM algorithm. In comparison with earlier models-trained by either EM or gradient ascent-there is no need to select a learning stepsize. We report simulation experiments which show that the new architecture yields faster convergence. We also apply the new model to two problem domains: piecewise nonlinear function approximation and the combination of multiple previously trained classifiers.
Cite
Text
Xu et al. "An Alternative Model for Mixtures of Experts." Neural Information Processing Systems, 1994.Markdown
[Xu et al. "An Alternative Model for Mixtures of Experts." Neural Information Processing Systems, 1994.](https://mlanthology.org/neurips/1994/xu1994neurips-alternative/)BibTeX
@inproceedings{xu1994neurips-alternative,
title = {{An Alternative Model for Mixtures of Experts}},
author = {Xu, Lei and Jordan, Michael I. and Hinton, Geoffrey E.},
booktitle = {Neural Information Processing Systems},
year = {1994},
pages = {633-640},
url = {https://mlanthology.org/neurips/1994/xu1994neurips-alternative/}
}