MO-AGH

Repository for MSc Thesis: Learning Based Multi-Objective Optimisation of Airport Ground Handling

This code extends on the existing Neural Airport Ground Handling model by incorporating the hypernetwork logic for multi-objective optimisation provided in the PSL-MOCO repository.

How to run

To train the baseline model, its variant using TCH scalarisation, the latent vector model, or its variant using TCH scalarisation run:

• python run_baseline_MO.py --graph_size 20 --baseline "rollout"
• python run_baseline_MO_TCH.py --graph_size 20 --baseline "rollout"
• python run_latent_vector_model.py --graph_size 20 --baseline "rollout"
• python run_latent_vector_model_TCH.py --graph_size 20 --baseline "rollout"

To evaluate trained models:

python eval.py --baseline_or_latent_vector_model "baseline" --datasets "./data/agh/agh20_validation_seed4321.pkl" --model "./data/mo_agh_20"

References

@article{wu2023neural,
title       = {Neural Airport Ground Handling},
author      = {Yaoxin Wu and Jianan Zhou and Yunwen Xia and Xianli Zhang and Zhiguang Cao and Jie Zhang},
journal     = {IEEE Transactions on Intelligent Transportation Systems},
year        = {2023},
doi         = {10.1109/TITS.2023.3253552},
publisher   = {IEEE}
}

@inproceedings{lin2022pareto,
  title={Pareto Set Learning for Neural Multi-Objective Combinatorial Optimization},
  author={Xi Lin, Zhiyuan Yang, Qingfu Zhang},
  booktitle={International Conference on Learning Representations},
  year={2022},
  url={https://openreview.net/forum?id=QuObT9BTWo}
}