Skip to main content

Module 23: Surrogate-Assisted & Expensive Optimization - When Each Evaluation Costs a Fortune

Some fitness evaluations take hours (CFD simulations), cost money (physical experiments), or are ethically limited (drug trials). Surrogate-assisted optimization builds cheap approximations and uses them to guide the search intelligently.

Learning Objectives​

  1. Understand when surrogate models are needed
  2. Build a GP (Gaussian Process) surrogate for a benchmark function
  3. Implement model-based optimization with an infill criterion (EI, LCB)
  4. Understand multi-fidelity optimization
  5. Know the connection to Bayesian optimization

Concept Explanation​

Coming soon.

Code Examples​

Coming soon.

Exercises​

Coming soon.

Milestone Checklist​

  • Built a surrogate-assisted EA
  • Understand expected improvement (EI)
  • Can choose between direct EA and surrogate-assisted
  • Know when Bayesian optimization vs surrogate EA is preferred

Was this page helpful?