Talk
DARLA: Improving Zero-Shot Transfer in Reinforcement Learning
Irina Higgins · Arka Pal · Andrei Rusu · Loic Matthey · Christopher Burgess · Alexander Pritzel · Matthew Botvinick · Charles Blundell · Alexander Lerchner

Mon Aug 7th 04:42 -- 05:00 PM @ C4.5

Domain adaptation is an important open problem in deep reinforcement learning (RL). In many scenarios of interest data is hard to obtain, so agents may learn a source policy in a setting where data is readily available, with the hope that it generalises well to the target domain. We propose a new multi-stage RL agent, DARLA (DisentAngled Representation Learning Agent), which learns to see before learning to act. DARLA’s vision is based on learning a disentangled representation of the observed environment. Once DARLA can see, it is able to acquire source policies that are robust to many domain shifts - even with no access to the target domain. DARLA significantly outperforms conventional baselines in zero-shot domain adaptation scenarios, an effect that holds across a variety of RL environments (Jaco arm, DeepMind Lab) and base RL algorithms (DQN, A3C and EC).

Author Information

Irina Higgins (DeepMind)
Arka Pal (DeepMind)
Andrei Rusu (DeepMind)
Loic Matthey (DeepMind)
Christopher Burgess (DeepMind)
Alexander Pritzel (Deepmind)
Matthew Botvinick (DeepMind)
Charles Blundell (DeepMind)
Alexander Lerchner (DeepMind)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors