Optimization and Learning via Stochastic Gradient Search

URL study guide

https://studiegids.vu.nl/en/courses/2025-2026/E_EORM_OLSGS

Course Objective

In this course, the student will learn how to apply gradient-based recursive algorithms for optimization and learning. After successful participation in this course, the student will be able to conduct a gradient-based stochastic optimization solution to real life problems.

Course Content

In presence of uncertainty, gradients typically fail to be available in analytical form and optimization has to resort to simulation-based algorithms. Gradient proxies are the main ingredient in such simulation-based optimization methods. The focus of this course is on (pseudo-) gradient estimators and their application in stochastic simulation-based optimization and learning algorithms. The recursive nature of the algorithms discussed in the course make them also applicable to data streaming, and data-streaming/online optimization will addressed in the course as well. This is a course on advanced simulation techniques. The methodological part of the course focuses on Discrete Event Simulation Techniques, Stochastic Models, and the theory of gradient based optimization.

Teaching Methods

Combined lectures and tutorials

Method of Assessment

Final exam – Individual assessment Individual assignment
- Individual assessment

Literature

Handout of monograph “Gradient based Stochastic Optimization”, B. Heidergott and F. Vásquez-Abad, 2023.

Entry Requirements

Analysis, basic probability theory, basic programming