Paper

Ant Colony Sampling with GFlowNets for Combinatorial Optimization

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 \emph{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.

arXiv (Cornell University)Published 2024-03-11Paper linkPDF

Authors: Kim, Minsu · Choi, Sanghyeok · Kim, Hyeonah · Son, Jiwoo · Park, Jinkyoo · Bengio, Yoshua

Topics

Relevant entities

People

Related coverage

Linked coverage will appear here.

Related events

Linked events will appear here.

Related discussions

Related discussion nodes will appear here.