search

garlicdevs / Fruit-API

A Universal Deep Reinforcement Learning Framework
http://fruitlab.org/
GNU General Public License v3.0
67 stars 22 forks source link

pareto front #11

Open TyrandeWhisperwind opened 4 years ago

TyrandeWhisperwind commented 4 years ago

how to get the pareto front after training a model, saving it and evaluating?

garlicdevs commented 4 years ago

You need to calculate the hypervolume, I think

garlicdevs commented 4 years ago

Can you check this paper: https://www.researchgate.net/publication/220343783_Empirical_evaluation_methods_for_multiobjective_reinforcement_learning_algorithms But I think 2, 3 objectives are feasible to calculate, more than that, you need to find an approximate front