Stochastic Adaptive Optimization With Dithers

• 成果类型: Article
• 署名作者: Xie, Siyu; Liang, Shu; Wang, Le Yi; Yin, George; Chen, Wen
• 署名单位: Wayne State University; Tongji University; Tongji University; University of Connecticut; Wayne State University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2021.3050438
• 发表日期: 2022
• 页码: 189-202
• 关键词: Optimization; CONVERGENCE; Signal processing algorithms; Stochastic processes; Parameter Estimation; Noise measurement; estimation; Adaptive optimization algorithm; convexity; identification; optimization; stochastic dither

摘要

Optimization methods are essential and have been used extensively in a broad spectrum of applications. Most existing literature on optimization algorithms does not consider systems that involve unknown system parameters. This article studies a class of stochastic adaptive optimization problems in which identification of unknown parameters and search for the optimal solutions must be performed simultaneously. Due to a fundamental conflict between parameter identifiability and optimality in such problems, we introduce a method of adding stochastic dither signals into the system, which provide sufficient excitation for estimating the unknown parameters, leading to convergent adaptive optimization algorithms. Joint identification and optimization algorithms are developed and their simultaneous convergence properties of parameter estimation and optimization variable updates are proved. Under both noise-free and noisy observations, the corresponding convergence rates are established. The main results of this article reveal certain fundamental relationships and tradeoff among updating step sizes, dither magnitudes, parameter estimation errors, optimization accuracy, and convergence rates. Simulation case studies are used to illustrate the adaptive optimization algorithms and their main properties.