Methods For Efficient Deep Reinforcement Learning