AWhat changed
Used Hessian-gradient products at initialization to score weight removals by their first-order effect on gradient norm, then trained the selected sparse subnetwork.
Scientific scope
Used Hessian-gradient products at initialization to score weight removals by their first-order effect on gradient norm, then trained the selected sparse subnetwork. This is a data-dependent pruning heuristic. Neural-tangent-kernel discussion motivates trainability but constant-kernel/infinite-width conditions do not prove finite-network accuracy or globally optimal masks. It differs from SNIP's immediate-loss sensitivity; all reported training comparisons remain benchmark-specific.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. This is a data-dependent pruning heuristic. Neural-tangent-kernel discussion motivates trainability but constant-kernel/infinite-width conditions do not prove finite-network accuracy or globally optimal masks. It differs from SNIP's immediate-loss sensitivity; all reported training comparisons remain benchmark-specific.
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