Srivastava and colleagues: dropout for neural-network regularization

What changed

Specified dropout training and test-time approximation, reporting benchmark generalization gains.

Scientific scope

The record is the exact 2014 JMLR five-author article. Earlier unpublished use, DropConnect, batch normalization, Bayesian interpretations and claims that dropout always improves generalization remain separate.

Evidence

The linked sources identify this contribution and its scope. WRS has not independently reconstructed the scientific claims. The record is the exact 2014 JMLR five-author article. Earlier unpublished use, DropConnect, batch normalization, Bayesian interpretations and claims that dropout always improves generalization remain separate.

Contributors

• Nitish Srivastava · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.
• Geoffrey Hinton · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.
• Alex Krizhevsky · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.
• Ilya Sutskever · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.
• Ruslan Salakhutdinov · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.

• JMLR record for Dropout
• JMLR full paper for Dropout

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction