I aim to review and understand how causal inference can be helpful in making reinforcement learning better. I think causality can make RL more sample efficient, make it interpretable and broaden its range of applications.
Papers at the Intersection of Reinforcement Learning and Causal Inference
- Causal Discovery with Reinforcement Lerning, Under Review, ICLR, 2020.
- Woulda, Coulda, Shoulda: Counterfactually-Guided Policy Search, Under Review, ICLR, 2019.
- Causal Reasoning from Meta-Reinforcement Learning, Under Review, ICLR, 2019.
- Discovering Latent Causes in Reinforcement Learning, Gershman et al., Behavioral Sciences, 2015.
- Reinforcement Learning and Cauasl Models, Sam Gershman, 2016.
- Representation Balancing MDPs for Off-Policy Policy Evaluation, Liu et. al., NeurIPS, 2018.
- Learning Plannable Representation with Causal InfoGAN, Kurutach et al., PAL, 2018.
- High-Confidence Policy Improvement, Thomas et al., ICML, 2015.
Relevant RL Papers
- Learning Model Based Planning from Scratch, Pascanu et al., arxiv, 2017.
- An Introduction to Deep Reinforcement Learning, Francois-Lavet et al., arxiv, 2018.
- Combined Reinforcement Learning via Abstract Representations, Francois-Lavet et al., arxiv, 2018.
- TreeQN and ATreeC: Differentiable Tree-Structured Models for Deep Reinforcement Learning, Farquhar et al., arxiv, 2018.
- Hindsight Experience Replay, Andrychowicz et al., NIPS, 2017.
- Universal Value Function Approximator, Schaul et al., ICML, 2015.
- Continuous Control with Deep Reinforcement Learning, Lillicrap et al., ICLR, 2016.