Self-Attentional Credit Assignment for Transfer in Reinforcement Learning
IJCAI 2020, pp. 2655-2661, 2020.
We investigated the role credit assignment could play in transfer learning and came up with SECRET, a novel transfer learning method that takes advantage of the relational properties of self-attention and transfers credit assignment instead of policy weights
The ability to transfer knowledge to novel environments and tasks is a sensible desiderata for general learning agents. Despite the apparent promises, transfer in RL is still an open and little exploited research area. In this paper, we take a brand-new perspective about transfer: we suggest that the ability to assign credit unveils struc...More
PPT (Upload PPT)