Bootstrapping with Noise: An Effective Regularization Technique
作者:
YUVAL RAVIV,
NATHAN INTRATOR,
期刊:
Connection Science
(Taylor Available online 1996)
卷期:
Volume 8,
issue 3-4
页码: 355-372
ISSN:0954-0091
年代: 1996
DOI:10.1080/095400996116811
出版商: Taylor & Francis Group
关键词: Noise Injection;Combining Estimators;Pattern Classification;Two-spiral Problem;Clinical Data Analysis
数据来源: Taylor
摘要:
Bootstrap samples with noise are shown to be an effective smoothness and capacity control technique for training feedforward networks and for other statistical methods such as generalized additive models. It is shown that noisy bootstrap performs best in conjunction with weight-decay regularization and ensemble averaging. The two-spiral problem, a highly non-linear, noise-free data, is used to demonstrate these findings. The combination of noisy bootstrap and ensemble averaging is also shown useful for generalized additive modelling, and is also demonstrated on the well-known Cleveland heart data.
点击下载:
PDF (648KB)
返 回