An augmented backward-corrected projector-splitting integrator (abc-PSI) trains rank-adaptive low-rank neural networks with one QR decomposition per step and a claimed convergence guarantee to locally optimal weights.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.NA 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
An Augmented Backward-Corrected Projector Splitting Integrator for Dynamical Low-Rank Training
An augmented backward-corrected projector-splitting integrator (abc-PSI) trains rank-adaptive low-rank neural networks with one QR decomposition per step and a claimed convergence guarantee to locally optimal weights.