Deep Q-learning from Demonstrations

Todd Hester, Matej Vecerik, Olivier Pietquin +11

Deep reinforcement learning (RL) has achieved several high profile successes in difficult decision-making problems. However, these algorithms typically require a huge amount of dat…