Domain Adversarial Reinforcement Learning
arXiv:2102.07097
Abstract
We consider the problem of generalization in reinforcement learning where visual aspects of the observations might differ, e.g. when there are different backgrounds or change in contrast, brightness, etc. We assume that our agent has access to only a few of the MDPs from the MDP distribution during training. The performance of the agent is then reported on new unknown test domains drawn from the distribution (e.g. unseen backgrounds). For this "zero-shot RL" task, we enforce invariance of the learned representations to visual domains via a domain adversarial optimization process. We empirically show that this approach allows achieving a significant generalization improvement to new unseen domains.
References in corpus (17)
- Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
- The Kinetics Human Action Video Dataset
- Unsupervised Domain Adaptation by Backpropagation
- DeepMind Control Suite
- Domain-Adversarial Neural Networks
- Reinforcement Learning with Augmented Data
- A Study on Overfitting in Deep Reinforcement Learning
- Actor-Mimic: Deep Multitask and Transfer Reinforcement Learning
- Leveraging Procedural Generation to Benchmark Reinforcement Learning
- Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels
- Learning Invariant Representations for Reinforcement Learning without Reconstruction
- Self-Supervised Policy Adaptation during Deployment
- Provably efficient RL with Rich Observations via Latent State Decoding
- Natural Environment Benchmarks for Reinforcement Learning
- Deep Reinforcement and InfoMax Learning
- Robust Visual Domain Randomization for Reinforcement Learning
- Mutual Information Maximization for Robust Plannable Representations