Towards graph neural networks for provably solving convex optimization problems

Chendi Qian, Christopher Morris

Recently, message-passing graph neural networks (MPNNs) have shown potential for solving combinatorial and continuous optimization problems due to their ability to capture variable…