Actor-Mimic: Deep Multitask and Transfer Reinforcement Learning:

[deontologician]

https://arxiv.org/pdf/1511.06342.pdf