Papers › Stochastic Zeroth Order Gradient and Hessian Estimators: Variance Reduction and...

Stochastic Zeroth Order Gradient and Hessian Estimators: Variance Reduction and Refined Bias Bounds

29 May 2022arXiv:2205.14737archive 2025-07-28

Yasong Feng, Tianyu Wang

We study stochastic zeroth order gradient and Hessian estimators for real-valued functions in ℝⁿ. We show that, via taking finite difference along random orthogonal directions, the variance of the stochastic finite difference estimators can be significantly reduced. In particular, we design estimators for smooth functions such that, if one uses Θ( k ) random directions sampled from the Stiefel's manifold St (n,k) and finite-difference granularity δ, the variance of the gradient estimator is bounded by 𝒪 ( ( n/k - 1 ) + ( n²/k - n ) δ² + (n² δ⁴)/k ), and the variance of the Hessian estimator is bounded by 𝒪 ( ( n²/k² - 1 ) + ( n⁴/k² - n² ) δ² + (n⁴ δ⁴)/k² ). When k = n, the variances become negligibly small. In addition, we provide improved bias bounds for the estimators. The bias of both gradient and Hessian estimators for smooth function f is of order 𝒪 ( δ² Γ), where δ is the finite-difference granularity, and Γ depends on high order derivatives of f. Our results are evidenced by empirical observations.

PaperPDFCode

Code

wangt1anyu/grad-hess-var-code officialmentioned in paper report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections