Ant Colony Sampling with GFlowNets for Combinatorial Optimization
Abstract
We present the Generative Flow Ant Colony Sampler (GFACS), a novel meta-heuristic method that hierarchically combines amortized inference and parallel stochastic search. Our method first leverages Generative Flow Networks (GFlowNets) to amortize a multi-modal prior distribution over combinatorial solution space that encompasses both high-reward and diversified solutions. This prior is iteratively updated via parallel stochastic search in the spirit of Ant Colony Optimization (ACO), leading to the posterior distribution that generates near-optimal solutions. Extensive experiments across seven combinatorial optimization problems demonstrate GFACS’s promising performances.
Cite
Text
Kim et al. "Ant Colony Sampling with GFlowNets for Combinatorial Optimization." Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, 2025.Markdown
[Kim et al. "Ant Colony Sampling with GFlowNets for Combinatorial Optimization." Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, 2025.](https://mlanthology.org/aistats/2025/kim2025aistats-ant/)BibTeX
@inproceedings{kim2025aistats-ant,
title = {{Ant Colony Sampling with GFlowNets for Combinatorial Optimization}},
author = {Kim, Minsu and Choi, Sanghyeok and Kim, Hyeonah and Son, Jiwoo and Park, Jinkyoo and Bengio, Yoshua},
booktitle = {Proceedings of The 28th International Conference on Artificial Intelligence and Statistics},
year = {2025},
pages = {469-477},
volume = {258},
url = {https://mlanthology.org/aistats/2025/kim2025aistats-ant/}
}