Efficient Global Optimization using Deep Gaussian Processes

Ali Hebbal (1, 2, 3) , Loïc Brevault (1) , Mathieu Balesdent (1) , El-Ghazali Talbi (2, 3) , Nouredine Melab (2, 3)

1 DTIS, ONERA, Université Paris Saclay (COmUE) [Palaiseau]
2 CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189
3 BONUS - Optimisation de grande taille et calcul large échelle

Efficient Global Optimization (EGO) is widely used for the optimization of computationally expensive black-box functions. It uses a surrogate modeling technique based on Gaussian Processes (Kriging). However, due to the use of a stationary covariance, Kriging is not well suited for approximating non stationary functions. This paper explores the integration of Deep Gaussian processes (DGP) in EGO framework to deal with the non-stationary issues and investigates the induced challenges and opportunities. Numerical experimentations are performed on analytical problems to highlight the different aspects of DGP and EGO.

Mots clés: Efficient Global Optimization; Non-stationary Kriging; Deep Gaussian Processes

Domaines: Machine Learning [stat.ML]; Modélisation et simulation

Ali Hebbal, Loïc Brevault, Mathieu Balesdent, El-Ghazali Talbi, Nouredine Melab. Efficient Global Optimization using Deep Gaussian Processes. CEC 2018 - Congress on Evolutionary Computation, Jul 2018, Rio de Janeiro, Brazil. ⟨hal-01919795⟩