TY  - RPRT
TI  - MAGIC: Multi-Step Advantage-Gated Causal Influence for Multi-agent Reinforcement Learning
AU  - Haohan Yu
AU  - Jinmiao Cong
AU  - Shengzhi Wang
AU  - Lu Wang
AU  - Chanjuan Liu
PY  - 2026
UR  - https://arxiv.org/abs/2605.01805
ID  - 2605.01805
ER  -