Heuristic-Supervised-DRL: A Unified Optimization Framework With Convergence Analysis
Many real-world problems involve hierarchical multi-objective optimization over coupled sequential decisions. A common strategy is to combine heuristic planning with deep reinforcement learning (DRL). While existing hybrid methods often show strong empirical performance, the coupled closed-loop learning dynamics betwee...