Ant Colony Sampling with GFlowNets for Combinatorial Optimization
This addresses combinatorial optimization problems, but appears incremental as it hybridizes existing methods.
The paper tackles combinatorial optimization by combining GFlowNets and Ant Colony Optimization to generate near-optimal solutions, showing promising results across seven problems.
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.