This project implements a surrogate-assisted evolutionary algorithm, called -DEA-DP, for expensive multi-obejctive optimziation. This algorithm maintains two deep neural networks as surrogates, one for Pareto dominance prediction and another for
-dominance prediction.
Through a two-stage preselection strategy, the two classification-based surrogates interact with a multi-objective evolutionary optimization process in order to select promising solutions for function evaluation.
This project requires
- Python (>= 3.6)
- NumPy (>= 1.13.3)
- Pytorch (>= 1.4.0)
- DEAP (>= 1.3.1)
- pymop (>= 0.2.4)
- optproblems (>= 1.3)
- matplotlib (>= 3.1.3)
An example is provided under the folder examples/ which demonstrates how to run the algorithm on a specific multi-objective optimization problem.
For questions and feedback, please contact [email protected]