The Generalisation Cost of RAMnets

Abstract

Given unlimited computational resources, it is best to use a crite(cid:173) rion of minimal expected generalisation error to select a model and determine its parameters. However, it may be worthwhile to sac(cid:173) rifice some generalisation performance for higher learning speed. A method for quantifying sub-optimality is set out here, so that this choice can be made intelligently. Furthermore, the method is applicable to a broad class of models, including the ultra-fast memory-based methods such as RAMnets. This brings the added benefit of providing, for the first time, the means to analyse the generalisation properties of such models in a Bayesian framework .

Cite

Text

Rohwer and Morciniec. "The Generalisation Cost of RAMnets." Neural Information Processing Systems, 1996.

Markdown

[Rohwer and Morciniec. "The Generalisation Cost of RAMnets." Neural Information Processing Systems, 1996.](https://mlanthology.org/neurips/1996/rohwer1996neurips-generalisation/)

BibTeX

@inproceedings{rohwer1996neurips-generalisation,
  title     = {{The Generalisation Cost of RAMnets}},
  author    = {Rohwer, Richard and Morciniec, Michal},
  booktitle = {Neural Information Processing Systems},
  year      = {1996},
  pages     = {253-259},
  url       = {https://mlanthology.org/neurips/1996/rohwer1996neurips-generalisation/}
}