A general square exponential kernel to handle mixed-categorical variables for Gaussian process

Saves, Paul; Diouane, Youssef; Bartoli, Nathalie; Lefèbvre, Thierry; Morlier, Joseph

doi:10.2514/6.2022-3870

Cited by 4 publications

(3 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Chapter 4 presents the open-source software in which these models are implemented and the development of new hierarchical GP models as in the article [208], corresponding to the points 3 and 6 in the list of contributions. Lastly, Chapter 5 regroups all the optimized application cases tackled during the last three years and correspond to our articles [13,91,182,204,205,207,209] for really high dimension, mixed hierarchical and multi-objective optimization and regroups the points 4, 5, 7, 8 and 9 in the list of contributions. This has been made possible thanks to the several co-workers of this thesis that applied the developed algorithms to industrial test case of high interest, notably for green aircraft developments.…”

Section: Constrained Bayesian Optimization In High Dimensionmentioning

confidence: 99%

High-dimensional multidisciplinary design optimization for aircraft eco-design

SAVES

2024

Preprint

View full text Add to dashboard Cite

Nowadays, there is a significant and growing interest in improving the efficiency of vehicle design processes through the development of tools and techniques in the field of MDO. Specifically, in aerostructure design, aerodynamic and structural variables influence each other and have a joint effect on quantities of interest like weight or fuel consumption and, as such, MDO arises as a powerful tool for automatically making interdisciplinary trade-offs. In the aircraft design context, the process generally involves mixed continuous and categorical design variables. For instance, the size of an aircraft’s structural parts can be described using continuous variables, while discrete variables may include either integer variables, like the number of panels, or categorical variables, like cross-sections or material choices. The objective of this Philosophiae Doctor (Ph.D) thesis is to propose an efficient approach for optimizing a multidisciplinary black-box model when the optimization problem is constrained and involves a large number of mixed integer design variables (typically 100 variables). The targeted optimization approach, called EGO, is based on a sequential enrichment of an adaptive surrogate model and, in this context, GP surrogate models are one of the most widely used in engineering problems to approximate time-consuming high fidelity models. EGO is a heuristic BO method that performs well in terms of solution quality. However, like any other global optimization method, EGO suffers from the curse of dimensionality, meaning that its performance is satisfactory on lower dimensional problems, but deteriorates as the dimensionality of the optimization search space increases. For realistic aircraft design problems, the typical size of the design variables can even exceed 100 and, thus, trying to solve directly the problems using EGO is ruled out. The latter is especially true when the problems involve both continuous and categorical variables increasing even more the size of the search space. In this Ph.D thesis, effective parameterization tools are investigated, including techniques like partial least squares regression, to significantly reduce the number of design variables. Additionally, Bayesian optimization is adapted to handle discrete variables and high-dimensional spaces in order to reduce the number of evaluations when optimizing innovative aircraft concepts such as the “DRAGON” hybrid airplane to reduce their climate impact.

show abstract

Section: Constrained Bayesian Optimization In High Dimensionmentioning

confidence: 99%

High-dimensional multidisciplinary design optimization for aircraft eco-design

SAVES

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…The L-NTADE algorithm was implemented in C++ and run on an OpenMPI-powered cluster of 8 AMD Ryzen 3700 PRO devices, with each core each using Linux 20.04. The EGO implementation from the Surrogate Modeling Toolbox (SMT) [37,38] was used in Python 3.8 to determine the Taylor series parameters and automatically run C++ code, as well as Mann-Whitney tests. Results post-processing was also performed using Python 3.8.…”

Section: Benchmark Functions and Parametersmentioning

confidence: 99%

Surrogate-Assisted Automatic Parameter Adaptation Design for Differential Evolution

Stanovov

Semenkin

2023

Mathematics

View full text Add to dashboard Cite

In this study, parameter adaptation methods for differential evolution are automatically designed using a surrogate approach. In particular, Taylor series are applied to model the searched dependence between the algorithm’s parameters and values, describing the current algorithm state. To find the best-performing adaptation technique, efficient global optimization, a surrogate-assisted optimization technique, is applied. Three parameters are considered: scaling factor, crossover rate and population decrease rate. The learning phase is performed on a set of benchmark problems from the CEC 2017 competition, and the resulting parameter adaptation heuristics are additionally tested on CEC 2022 and SOCO benchmark suites. The results show that the proposed approach is capable of finding efficient adaptation techniques given relatively small computational resources.

show abstract

“…To solve optimization problems, SBArchOpt implements the following (interfaces to) optimization libraries/algorithms: 1. pymoo: SBArchOpt provides a pre-configured version of the NSGA2 evolutionary optimization algorithm; 2. ArchSBO: a custom implementation of a mixed-discrete, multi-objective Surrogate-Based Optimization algorithm, with support for design variable correction, hidden constraints, and restart, using state-of-the-art mixed-discrete, hierarchical Gaussian Process models(Saves et al, 2023); 3. three open-source Bayesian Optimization libraries: BoTorch (Ax)(Balandat et al, 2020),Trieste(Picheny et al, 2023), and HEBO (Cowen-Rivers et al, 2022); 4. two proprietary Bayesian Optimization libraries: SEGOMOE(Bartoli et al, 2019) and SMARTy(Bekemeyer et al, 2022); and 5. a Tree Parzen Estimator (TPE) algorithm with support for hidden constraints.…”

mentioning

confidence: 99%

SBArchOpt: Surrogate-Based Architecture Optimization

Bussemaker

2023

JOSS

View full text Add to dashboard Cite

In engineered systems, the architecture of a system describes how the components of a system work together to fulfill the system functions and meet stakeholder expectations (Crawley et al., 2015). As the architecture is developed in early conceptual design stages, decisions involving architecture often have a large influence on final system performance, for example, in terms of how well the functions are fulfilled, or at what cost and in what timeframe. However, architecture design spaces (i.e., the set of all possible architecture alternatives) can be very large due to the combinatorial nature of architecture decisions, making it infeasible to compare all alternatives to each other. Additionally, for new systems there might not be any prior experience to start from, requiring the use of (typically) computationally-expensive physics-based simulation to estimate system performance.

show abstract

A general square exponential kernel to handle mixed-categorical variables for Gaussian process

Cited by 4 publications

References 27 publications

High-dimensional multidisciplinary design optimization for aircraft eco-design

High-dimensional multidisciplinary design optimization for aircraft eco-design

Surrogate-Assisted Automatic Parameter Adaptation Design for Differential Evolution

SBArchOpt: Surrogate-Based Architecture Optimization

Contact Info

Product

Resources

About