Representation Learning for RL Reading List

Curated by Mouhssine Rifaki | Stanford Electrical Engineering | Last updated August 2026

Good representations make reinforcement learning data-efficient and generalize beyond the training task. This collection traces the path from auxiliary tasks and contrastive objectives to latent world models that decouple seeing from acting.

Representation Learning for RL: 10 key papers

1. Representation Learning: A Review and New Perspectives
   Bengio et al. arXiv 2012.
2. Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
   Locatello et al. arXiv 2018.
3. Isolating Sources of Disentanglement in Variational Autoencoders
   Chen et al. arXiv 2018.
4. Learning deep representations by mutual information estimation and maximization
   Hjelm et al. arXiv 2018.
5. On Mutual Information Maximization for Representation Learning
   Tschannen et al. arXiv 2019.
6. Decoupling Representation Learning from Reinforcement Learning
   Stooke et al. arXiv 2020.
7. Towards a Definition of Disentangled Representations
   Higgins et al. arXiv 2018.
8. Self-supervised Learning: Generative or Contrastive
   Liu et al. arXiv 2020.
9. Bootstrap Latent-Predictive Representations for Multitask Reinforcement Learning
   Guo et al. arXiv 2020.
10. Learning Invariant Representations for Reinforcement Learning without Reconstruction
    Zhang et al. arXiv 2020.

Related reading lists

• Information Theory & RL
• Contrastive & Self-Supervised Learning
• Block MDPs & Latent State Discovery
• Data Augmentation for Reinforcement Learning
• Intrinsic Motivation & Empowerment
• Goal-Conditioned Reinforcement Learning
• Multi-Task & Transfer Reinforcement Learning
• Variational Inference & ELBO