One solution is the deep deterministic policy gradient (DDPG) algorithm, which combines the actor-critic approach with insights from the success of DQN. This is discussed in the following papers:
- D. Silver, G. Lever, N. Heess, T. Degris, D. Wierstra and M. Riedmiller. Deterministic policy gradient algorithms. In ICML, 2014.
- T. P. Lillicrap, J. J. Hunt, A. Pritzel, N. Heess, T. Erez, Y. Tassa, D. Silver and D. Wierstra. Continuous control with deep reinforcement learning. In ICLR, 2016.
The reason why DDPG is introduced first is that it is quite similar to DQN, so you can understand the mechanism behind it much more easily after finishing the previous chapter. Recall that DQN is able to train the Q-network ...