AWhat changed
Applied user-adjacent DP to language-model training using independently sampled users, clipped user updates, bounded-sensitivity averaging estimators, Gaussian noise and moments accounting.
Scientific scope
Applied user-adjacent DP to language-model training using independently sampled users, clipped user updates, bounded-sensitivity averaging estimators, Gaussian noise and moments accounting. Guarantee concerns adding/removing an entire user's data under specified sampling/estimator/clipping/accounting assumptions, not cryptographic concealment from a trusted curator. Accuracy preservation was demonstrated at substantial computation cost on a sufficiently large dataset, not free privacy or all datasets.2017 v1 title Language Models differs from later Recurrent Language Models.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Guarantee concerns adding/removing an entire user's data under specified sampling/estimator/clipping/accounting assumptions, not cryptographic concealment from a trusted curator. Accuracy preservation was demonstrated at substantial computation cost on a sufficiently large dataset, not free privacy or all datasets.2017 v1 title Language Models differs from later Recurrent Language Models.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
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