Course details
Sample Efficient Optimization
MATH 80618A
Business and industry stakeholders rely on high-fidelity and costly to compute models to inform their decision making. This course surveys machine learning (ML) and operations research (OR) methods that enable cost-efficient and/or sample efficient decision making with such models.
This course presents methods that combine domain knowledge from management, science or engineering within classical ML and OR techniques to yield practice-relevant optimization tools. Topics include Gaussian process modeling, Bayesian optimization, dimensionality reduction, variance reduction, continuous and/or discrete simulation-based optimization, physics-informed optimization, surrogate modeling. Concepts are illustrated using practical applications. The course presents established and emerging research through lectures and paper discussions. It includes guest research lecturers from major technology companies and / or top academic institutions worldwide.
Themes covered
Gaussian processes
Bayesian optimization
Dimensionality reduction
Variance reduction
Continuous simulation-based optimization
Discrete simulation-based optimization
Surrogate modeling
Course code
MATH 80618A
Subject
Mathematics
Program
PhD
Location
Côte-des-Neiges
Instruction mode
On-site learning
Credits
3