Metamorphosis Networks: An Alternative to Constructive Models

Abstract

Given a set oft raining examples, determining the appropriate num(cid:173) ber of free parameters is a challenging problem. Constructive learning algorithms attempt to solve this problem automatically by adding hidden units, and therefore free parameters, during learn(cid:173) ing. We explore an alternative class of algorithms-called meta(cid:173) morphosis algorithms-in which the number of units is fixed, but the number of free parameters gradually increases during learning. The architecture we investigate is composed of RBF units on a lat(cid:173) tice, which imposes flexible constraints on the parameters of the network. Virtues of this approach include variable subset selec(cid:173) tion, robust parameter selection, multiresolution processing, and interpolation of sparse training data.

Cite

Text

Bonnlander and Mozer. "Metamorphosis Networks: An Alternative to Constructive Models." Neural Information Processing Systems, 1992.

Markdown

[Bonnlander and Mozer. "Metamorphosis Networks: An Alternative to Constructive Models." Neural Information Processing Systems, 1992.](https://mlanthology.org/neurips/1992/bonnlander1992neurips-metamorphosis/)

BibTeX

@inproceedings{bonnlander1992neurips-metamorphosis,
  title     = {{Metamorphosis Networks: An Alternative to Constructive Models}},
  author    = {Bonnlander, Brian V. and Mozer, Michael},
  booktitle = {Neural Information Processing Systems},
  year      = {1992},
  pages     = {131-138},
  url       = {https://mlanthology.org/neurips/1992/bonnlander1992neurips-metamorphosis/}
}