Contributions To Multi-Agent Reinforcement Learning