A novel approach to generalisation is presented that is able, under certain circumstances, to guarantee the generalisation to binary-output data for which no targets have been given. The basis of the guarantee is the recognition of a persistent global minimum error solution. An empirical test for whether the guarantee holds is provided which uses a technique called target reversal. The technique employs two neural networks whose convergence using opposing targets signals validity of the guarantee.
Описание
An approach to guaranteeing generalisation in neural networks - ScienceDirect
%0 Journal Article
%1 POLHILL20011035
%A Polhill, J.Gary
%A Weir, Michael K.
%D 2001
%J Neural Networks
%K ExhaustiveSearch Generalization NeuralNetworks TargetReversal myown
%N 8
%P 1035 - 1048
%R https://doi.org/10.1016/S0893-6080(01)00061-2
%T An approach to guaranteeing generalisation in neural networks
%U http://www.sciencedirect.com/science/article/pii/S0893608001000612
%V 14
%X A novel approach to generalisation is presented that is able, under certain circumstances, to guarantee the generalisation to binary-output data for which no targets have been given. The basis of the guarantee is the recognition of a persistent global minimum error solution. An empirical test for whether the guarantee holds is provided which uses a technique called target reversal. The technique employs two neural networks whose convergence using opposing targets signals validity of the guarantee.
@article{POLHILL20011035,
abstract = {A novel approach to generalisation is presented that is able, under certain circumstances, to guarantee the generalisation to binary-output data for which no targets have been given. The basis of the guarantee is the recognition of a persistent global minimum error solution. An empirical test for whether the guarantee holds is provided which uses a technique called target reversal. The technique employs two neural networks whose convergence using opposing targets signals validity of the guarantee.},
added-at = {2019-11-18T11:40:42.000+0100},
author = {Polhill, J.Gary and Weir, Michael K.},
biburl = {https://www.bibsonomy.org/bibtex/204f97f5c6a72a2d66a409efb78a667e5/garypolhill},
description = {An approach to guaranteeing generalisation in neural networks - ScienceDirect},
doi = {https://doi.org/10.1016/S0893-6080(01)00061-2},
interhash = {9b096623d753c910c0e802d4dd2e03e7},
intrahash = {04f97f5c6a72a2d66a409efb78a667e5},
issn = {0893-6080},
journal = {Neural Networks},
keywords = {ExhaustiveSearch Generalization NeuralNetworks TargetReversal myown},
number = 8,
pages = {1035 - 1048},
timestamp = {2019-11-18T11:40:42.000+0100},
title = {An approach to guaranteeing generalisation in neural networks},
url = {http://www.sciencedirect.com/science/article/pii/S0893608001000612},
volume = 14,
year = 2001
}