STACEY is a new ℓ_p steepest descent optimizer with primal-dual interpolation; its convergence theory covers only the unaccelerated base algorithm, and its empirical gains rely on grid-searched hyperparameters.
J., Bartlett, P., and Ravikumar, P
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Stacey: Promoting Stochastic Steepest Descent via Accelerated $\ell_p$-Smooth Nonconvex Optimization
STACEY is a new ℓ_p steepest descent optimizer with primal-dual interpolation; its convergence theory covers only the unaccelerated base algorithm, and its empirical gains rely on grid-searched hyperparameters.