Learn Backend Development Part-Time, Online
AI, Data Science & Cloud Certificates from Google, IBM & Meta
Overview
Build a Learning Habit
Download Class Central's free printable study calendar
Download for Free
Explore essential benchmarking methodologies for optimization algorithms in this 40-minute seminar from GERAD Research Center. Learn standard benchmarking techniques including convergence plots, performance profiles, data profiles, and accuracy profiles used to evaluate algorithm effectiveness. Discover formal extensions of these methods to three specialized contexts: constrained optimization, multi-objective optimization, and surrogate-based optimization. Understand the critical importance of benchmarking in derivative-free optimization where complex simulations require extensive evaluation time, limiting the number of possible evaluations. Gain insights into assessing algorithm quality and potential when working with computationally expensive target problems. Access practical benchmarking codes made available to the optimization community for implementing these evaluation techniques in your own research and development work.
Syllabus
Benchmarking algorithms on large test sets, Charles Audet
Taught by
GERAD Research Center