Overview
Explore the intersection of machine learning and optimization in this 18-minute talk from The Julia Programming Language. Learn how to accelerate solutions for complex optimization problems using machine-learning approximations. Discover how substituting challenging components with ML approximations or replacing optimization steps entirely can lead to significant performance improvements in real-world scenarios. Understand the process of identifying and exploiting similarities in parametric problems to enhance efficiency for decision-makers facing similar structures across multiple problem instances. Dive into the capabilities of L2O.jl (Learning to Optimize), a Julia package that connects the JuMP ecosystem with Julia ML packages, enabling automated proxy learning and efficient optimization of real-world applications. Examine the seamless integration of the JuMP ecosystem, particularly the Parametric Optimization Interface (POI), with packages like Dualization.jl and DiffOpt.jl to create a cohesive parametric framework for effective problem formulation modifications and efficient ML model learning.
Syllabus
Bridging ML and Optimization with JuMP
Taught by
The Julia Programming Language