> I'm definitely awaiting for traveling salesman solver or similar, guided by NN, solving things faster and reaching optimality more frequently that optimized h

> I'm definitely awaiting for traveling salesman solver or similar, guided by NN, solving things faster and reaching optimality more frequently that optimized heuristic algos. There was a competition for exactly this at Neurips 2021 https://www.ecole.ai/2021/ml4co-competition/ https://www.ecole.ai/2021/ml4co-competition/ Not sure how much they improved over handcrafted heuristics, but the summary paper may give some insights https://arxiv.org/abs/2203.02433 https://arxiv.org/abs/2203.02433