Bootstrapping with Noise: An Effective Regularization Technique

Yuval Raviv*, Nathan Intrator

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


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.

Original languageEnglish
Pages (from-to)355-372
Number of pages18
JournalConnection Science
Issue number3-4
StatePublished - Dec 1996


  • Clinical data analysis
  • Combining estimators
  • Noise injection
  • Pattern classification
  • Two-spiral problem


Dive into the research topics of 'Bootstrapping with Noise: An Effective Regularization Technique'. Together they form a unique fingerprint.

Cite this