Connections Between Adaptive Control and Optimization in Machine Learning

被引：0

作者：

Gaudio, Joseph E. ^{[1
]}

Gibson, Travis E. ^{[2
]}

Annaswamy, Anuradha M. ^{[1
]}

Bolender, Michael A. ^{[3
]}

Lavretsky, Eugene ^{[4
]}

机构：

[1] MIT, Dept Mech Engn, Cambridge, MA 02139 USA

[2] Harvard Med Sch, Dept Pathol, Boston, MA 02115 USA

[3] Air Force Res Lab, Wright Patterson AFB, OH 45433 USA

[4] Boeing Co, Huntington Beach, CA 92647 USA

来源：

2019 IEEE 58TH CONFERENCE ON DECISION AND CONTROL (CDC) | 2019年

关键词：

PARAMETER CONVERGENCE; NEURAL-NETWORKS; ADAPTATION; SYSTEMS;

D O I：

10.1109/cdc40024.2019.9029197

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

This paper demonstrates many immediate connections between adaptive control and optimization methods commonly employed in machine learning. Starting from common output error formulations, similarities in update law modifications are examined. Concepts in stability, performance, and learning, common to both fields are then discussed. Building on the similarities in update laws and common concepts, new intersections and opportunities for improved algorithm analysis are provided. In particular, a specific problem related to higher order learning is solved through insights obtained from these intersections.

引用

页码：4563 / 4568

页数：6

共 85 条

[1]

Anderson B. D., 1982, AUTOMATICA, V18, P1

[2] EXPONENTIAL CONVERGENCE OF ADAPTIVE IDENTIFICATION AND CONTROL ALGORITHMS [J].