DARLA: Improving Zero-Shot Transfer in Reinforcement Learning
This work addresses the challenge of data scarcity in RL by enabling agents to adapt to new domains without additional training, which is incremental as it builds on existing representation learning methods.
The paper tackles the problem of domain adaptation in deep reinforcement learning by proposing DARLA, a multi-stage agent that learns disentangled representations before acting, enabling robust zero-shot transfer to unseen target domains. DARLA significantly outperforms conventional baselines across various RL environments and algorithms, demonstrating improved generalization without access to target data.
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).