Regularly, for the hard scheduling problems (e.g., scheduling, sequencing, network design, etc.) we use heuristic/meta-heuristic/hyper-heuristic algorithms to so-called reveal landscape of an objective function, and finally to find an optimum or a near-optimum solution for that.
Do you have any idea how to use machine learning methods to PREDICT landscape (or something similar to it) of the objective function, in a more timely manner?
Please share your thoughts.