Frank-Wolfe with the log-adaptive step-size η_t=(2+log(t+1))/(t+2+log(t+1)) provably matches or beats fixed-ℓ open-loop step-sizes in several convergence regimes.
Distribution-based sketching of single-cell samples
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Adaptive Open-Loop Step-Sizes for Accelerated Convergence Rates of the Frank-Wolfe Algorithm
Frank-Wolfe with the log-adaptive step-size η_t=(2+log(t+1))/(t+2+log(t+1)) provably matches or beats fixed-ℓ open-loop step-sizes in several convergence regimes.