Online Statistical Inference for Stochastic Optimization via Kiefer-Wolfowitz Methods

Abstract

This paper investigates the problem of online statistical inference of model parameters in stochastic optimization problems via the Kiefer-Wolfowitz algorithm with random search directions. We first present the asymptotic distribution for the Polyak-Ruppert-averaging type Kiefer-Wolfowitz (AKW) estimators, whose asymptotic covariance matrices depend on the distribution of search directions and the function-value query complexity. The distributional result reflects the trade-off between statistical efficiency and function query complexity. We further analyze the choice of random search directions to minimize certain summary statistics of the asymptotic covariance matrix. Based on the asymptotic distribution, we conduct online statistical inference by providing two construction procedures of valid confidence intervals.

References in corpus (8)

• Query Complexity of Derivative-Free Optimization
• A companion for the Kiefer--Wolfowitz--Blum stochastic approximation algorithm
• Stochastic Zeroth-order Optimization in High Dimensions
• HiGrad: Uncertainty Quantification for Online Learning and Stochastic Approximation
• Fast and Robust Online Inference with Stochastic Gradient Descent via Random Scaling
• Statistical Inference for the Population Landscape via Moment Adjusted Stochastic Gradients
• A generalization of regularized dual averaging and its dynamics
• On Scalable Inference with Stochastic Gradient Descent

Cited by in corpus (1)

• Fast and Robust Online Inference with Stochastic Gradient Descent via Random Scaling