Paper 4 MultiObjective Flashcards
1
Q
Pareto optimality
A
Not pareto dominated by anything
2
Q
Pareto dominating
A
Better at at least something and not worst at the rest
3
Q
Naive approaches for MOHPO
A
Scalarization, transform into single objective (e.g. weighted sum)
Random/grid search
4
Q
Evolutionary approach
A
Selection using Pareto dominance
5
Q
MO-BO
A
Multi objective bayesian optimization
Use scalarization or use multiple surrogates
6
Q
ParEGO
A
Scalarization based MO-BO
First needs weights for all objectives
7
Q
EHI
A
Surrogate model per objective
