Learning Combinatorial Node Labeling Algorithms
This addresses combinatorial optimization on graphs for applications in scheduling or network design, but it is incremental as it builds on existing graph attention networks and reinforcement learning methods.
The paper tackles graph optimization problems like graph coloring by proposing a neural architecture trained with reinforcement learning, achieving better solutions than classical greedy heuristics and outperforming state-of-the-art baselines in seconds for large graphs.
We present a novel neural architecture to solve graph optimization problems where the solution consists of arbitrary node labels, allowing us to solve hard problems like graph coloring. We train our model using reinforcement learning, specifically policy gradients, which gives us both a greedy and a probabilistic policy. Our architecture builds on a graph attention network and uses several inductive biases to improve solution quality. Our learned deterministic heuristics for graph coloring give better solutions than classical degree-based greedy heuristics and only take seconds to apply to graphs with tens of thousands of vertices. Moreover, our probabilistic policies outperform all greedy state-of-the-art coloring baselines and a machine learning baseline. Finally, we show that our approach also generalizes to other problems by evaluating it on minimum vertex cover and outperforming two greedy heuristics.