2+ years of hard work on using GNNs to improve combinatorial optimisation solvers. Our learned models are faster and better than the best open source MIP solver SCIP on many benchmarks, and also improved the best known solutions for some hard problems, beating CPLEX / Gurobi! https://t.co/NQM6DQtGDx
— Yujia Li (@liyuajia) Jul 26, 2021
from Twitter https://twitter.com/liyuajia
July 26, 2021 at 08:51AM
via IFTTT
No comments:
Post a Comment