Quantum-Inspired Recommendation and Ban-Pick Optimization for Professional MOBA Tournaments
Treas Huang · 2025
Optimizing global Ban-Pick (BP) strategies in Multiplayer Online Battle Arena (MOBA) games is a complex challenge due to the multi-round structure, dynamic game updates, and intricate inter-hero synergies. Existing methods often fail to balance round-specific utility with long-term strategic objectives under evolving game metas. We propose a unified framework integrating quantum-inspired multi-agent reinforcement learning, meta-learning for version adaptation, and graph neural networks to address these challenges. By modeling BP as a hierarchical decision process, our approach dynamically adapts to game updates, enhances synergy modeling, and optimizes multi-round strategies. Real-world experiments with professional MOBA players show significant improvements in performance, establishing the framework's effectiveness in competitive scenarios.