The student will learn the concepts of parallel optimization assisted by statistical learning. He/She will be introduced to the methods of optimization, including meta-heuristics and Bayesian optimization, their coupling to machine learning and their parallelization. Engineering simulation will be considered as a field of application and experimentation.

Program

o   General introduction

o   Statistical learning models (reminder on statistics, Metamodels - Gaussian Processes, Neural Networks)

o   Statistical learning–assisted Optimization

  • Optimization methods (Gradient descent, meta-heuristics)
  • Bayesian optimization (acquisition functions, constraints handling, some basic examples, parallelization)
  • Neural Networks-assisted Meta-heuristics (evolution control mechanisms, constraints handling, some basic examples, parallelization)

o   Applications of ML-assisted optimization to some problems in various fields including  design engineering, image processing, etc.

Acquired skills

The student will learn to design and implement optimization methods assisted by statistical learning. He/She will also learn how to apply, test and validate them in the context of simulation engineering. He/She will also discover the existing statistical learning tools and their linkage with optimization software platforms.

Exam and grades

An exam mark (Exam/ 20) and a practical project mark (PP / 20), applying the formula : MARK = (Exam + PP) / 2.