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