Concept非渐近分析
Non-Asymptotic Convergence Analysis
A convergence analysis method that provides explicit error bounds under finite time and finite samples, offering guidance for practical hyperparameter selection. It differs from asymptotic analysis, which only proves that error tends to zero as the number of iterations tends to infinity.
Timeline (last 90 days)
Oct 15
The paper covers two settings, compactly supported targets and power-law targets, under the p-Wasserstein distance framework, gives non-asymptotic convergence bounds, and separates the contributions of early-stopping error, initialization error, score approximation error, and time discretization error.
Unverified50%