Negative momentum enables global convergence in convex-concave min-max optimization and accelerated rates in the strongly-convex-strongly-concave setting.
arXiv preprint arXiv:2206.08303 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
SGD is reformulated via a master equation from discrete updates, producing a discrete Fokker-Planck equation that predicts non-stationary variance growth proportional to learning rate in flat Hessian directions.
citing papers explorer
-
Negative Momentum for Convex-Concave Optimization
Negative momentum enables global convergence in convex-concave min-max optimization and accelerated rates in the strongly-convex-strongly-concave setting.
-
Why SGD is not Brownian Motion: A New Perspective on Stochastic Dynamics
SGD is reformulated via a master equation from discrete updates, producing a discrete Fokker-Planck equation that predicts non-stationary variance growth proportional to learning rate in flat Hessian directions.