Gaussian process regression models, a.k.a. Kriging models, are applied to global multi-objective optimization of black-box functions. Multi-objective Expected Improvement and Step-wise Uncertainty Reduction sequential infill criteria are available. A quantification of uncertainty on Pareto fronts is provided using conditional simulations.
|Author||Mickael Binois, Victor Picheny|
|Date of publication||2016-11-11 17:36:23|
|Maintainer||Mickael Binois <email@example.com>|
checkPredict: Prevention of numerical instability for a new observation
CPF: Conditional Pareto Front simulations
crit_EHI: Expected Hypervolume Improvement with m objectives
crit_EMI: Expected Maximin Improvement with m objectives
crit_optimizer: Maximization of multiobjective Expected Improvement criteria
crit_SMS: Analytical expression of the SMS-EGO criterion with m>1...
crit_SUR: Analytical expression of the SUR criterion for two or three...
easyGParetoptim: EGO algorithm for multiobjective optimization
fastfun: Fastfun function
fastfun-class: Class for fast to compute objective.
getDesign: Get design corresponding to an objective target
GPareto: Package GPareto
GParetoptim: Sequential multi-objective Expected Improvement maximization...
integration_design_optim: Function to build integration points (for the SUR criterion)
ParetoSetDensity: Estimation of Pareto set density
plotGPareto: Plot multi-objective optimization results and post-processing
plotParetoEmp: Pareto front visualization
plotParetoGrid: Visualisation of Pareto front and set
plotSymDevFun: Display the Symmetric Deviation Function
plotSymDifRNP: Symmetrical difference of RNP sets
plot_uncertainty: Plot uncertainty
TestFunctions: Test functions of x