A bi-agent learning system tailored for automated algorithm design in black-box optimization.

A domain-robust mixture-of-experts for MILP solution prediction across domains

MACS: a new multi-agent reinforcement learning method for faster geometry optimization of crystal structures.