AWhat changed
Combined cross-view invariance, per-dimension variance control and covariance decorrelation to stabilize joint-embedding self-supervision without relying on negative-pair sampling.
Scientific scope
Combined cross-view invariance, per-dimension variance control and covariance decorrelation to stabilize joint-embedding self-supervision without relying on negative-pair sampling. Variance regularization and decorrelation ingredients have predecessors. Objective-level prevention of constant embeddings is not a guarantee of global convergence or complete semantic representation. Original2021 preprint and ICLR2022 publication are distinct.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Variance regularization and decorrelation ingredients have predecessors. Objective-level prevention of constant embeddings is not a guarantee of global convergence or complete semantic representation. Original2021 preprint and ICLR2022 publication are distinct.
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.
- VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning
- Bardes et al.: VICReg explicit variance and covariance regularization — author/institution implementation
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