- Article
The distributed policy gradient (PG) method has become a mainstream technique for finding optimal policies in multi-agent reinforcement learning. However, global convergence guarantees, which are the most basic measures of theoretical convergence per...