mlsanityData to 14 Oct 2026

Deep RL algorithms and representations

1,369 papers8 this weekLow confidence in this grouping: the papers may belong to more than one topic, or the group may changeLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About

Keywords: reinforcement learning, reinforcement, performance, policy, environments

Key facts. Deep RL algorithms and representations: 1,369 papers on arXiv since 2024 (data through 14 Oct 2026). 47 of its 1,369 mapped papers (3%) cite “Bigger, Regularized, Optimistic: scaling for compute and sample-efficient continuous control” (2024, arXiv 2405.16158). 49 new papers 30 Sep–14 Oct.

What’s new

30 Sep – 14 Oct 2026

49 new papers, 30 Sep – 14 Oct; 8 this week.

  1. This week · 8
  2. Selective Growing Q-Networks (SGQN): An Analysis of Adaptive Action Grids in Decoupled Q-Learning also fits: RL theory and sample complexity
  3. Outperforming 1000 Euclidean Layers With Hyperbolic Goal-Conditioned RL also fits: RL with diffusion and flow models
  4. Provable Transformer Context Compression for In-Context Reinforcement Learning also fits: RL theory and sample complexity
  5. Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction also fits: RL with diffusion and flow models
  6. Reinforcement Learning via Advantage Guidance
  7. MIPS: Mitigating Mode Collapse in Reinforcement Learning for Large Outcome Spaces also fits: RL theory and sample complexity
  8. When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability also fits: RL for games and combinatorial optimization
  9. From Policy Evaluation to Off-Policy Control: Provable Emergence of In-Context Q-Learning also fits: RL theory and sample complexity
  10. Earlier · 41
  11. FARL: A Fully Automated Pipeline of RL Interface and Solver Design also fits: RL for robot locomotion and manipulation
  12. PEARL: Process-Experience Aware Reinforcement Learning for Multi-Turn Agents
All 49
  1. Sparse-TD-JEPA: Improving Zero-Shot Reinforcement Learning With Sparse Latent Task Representations
  2. Does Learning Recurrent Networks Matter in Partially Observable RL?
  3. Cross-Behavior Policy Optimization: Unified Learning from Offline Demonstrations and On- and Off-Policy Rollouts via a Shared Regression Target also fits: Interpretable and logic-guided RL
  4. Q-Shaped Options for Hierarchical Reinforcement Learning also fits: Interpretable and logic-guided RL
  5. Can Jev be Your Q or Policy in Reinforcement Learning? also fits: Interpretable and logic-guided RL
  6. World-Model Policy Arbiter for Goal-Conditioned Reinforcement Learning also fits: RL theory and sample complexity
  7. Sample-Efficiency of Kolmogorov-Arnold Networks also fits: RL theory and sample complexity
  8. A Good Self-Teacher Meets the Student Where They Are: Joint On-Policy Learning and Teaching
  9. CERO: Where and When to Allocate Rollouts for RL Post-Training also fits: RL for games and combinatorial optimization
  10. An Informational Curse of Horizon in Goal-Conditioned Policy Learning
  11. Reinforcement Learning with Conformal Action Sets: An Application to Sequential Recommendation also fits: RL theory and sample complexity
  12. Revisiting Temporal Regularization for Smooth Control in Deep Reinforcement Learning also fits: RL with diffusion and flow models
  13. Considering Context: When World Models Need Context Encoders
  14. Constrained Goal-directed Planar Graph Generation with Grammar-based Reinforcement Learning also fits: RL with diffusion and flow models
  15. Adaptive Expert Guidance for Efficient On-Policy Reinforcement Learning also fits: RL for games and combinatorial optimization
  16. Hierarchical Time-aware Bootstrapping for Off-Policy Subgoal Value Learning
  17. On Semi-Markov Suboptimality in Hierarchical Reinforcement Learning also fits: RL theory and sample complexity
  18. Arithmetic Actor Heads and Training Stabilization for Out-of-Distribution Reinforcement Learning also fits: RL theory and sample complexity
  19. Direction-Conditioned Policies for Online Goal-Conditioned Reinforcement Learning also fits: RL for robot locomotion and manipulation
  20. DreamTest: World-Model Surrogates for Search-Based Testing of Deep Reinforcement Learning Agents also fits: Interpretable and logic-guided RL
  21. Credit Where It Matters: Dependency-Aware Policy Optimization for Terminal Agents also fits: Interpretable and logic-guided RL
  22. UniIntervene++: An Adaptive Intervention Agent for Efficient Real-World Reinforcement Learning also fits: RL for robot locomotion and manipulation
  23. Bidirectional Voronoi-biased Exploration Curriculum for Reinforcement Learning also fits: RL for robot locomotion and manipulation
  24. How to Find and Reuse Policies for Continuous Adaptation in Lifelong Reinforcement Learning
  25. Understanding Enrichment in Reinforcement Learning
  26. Prospective Hindsight: Self-Calibrating Reinforcement Learning via Prediction-Reality Gaps
  27. Learning from Evolving Errors: Adaptive Iterative Repair for On-Policy Distillation also fits: Safe and constrained RL
  28. Reward Inflation: A Healthy Stimulus for Reinforcement Learning
  29. When Do Intrinsic Rewards Lead to Exploration?
  30. Finetuning with Sampling: SFT Learns Better Than You Think also fits: RL with diffusion and flow models
  31. Continual Reinforcement Learning with Neuroevolution
  32. Reusing Past Samples in Proximal Policy Optimization: When and How Does It Help? also fits: RL theory and sample complexity
  33. Does Scaling Reinforcement Learning Really Require More Training? also fits: Interpretable and logic-guided RL
  34. Calibration-risk routing for controlled world-model adaptation also fits: RL for robot locomotion and manipulation
  35. Learning Multiple Timescales for Goal-Conditioned Reinforcement Learning
  36. Scalable Multi-Task Inverse Reinforcement Learning
  37. ALER: Adaptive Learnable Experience Rewriting for Reinforcement Learning also fits: Interpretable and logic-guided RL
  38. Do Better Goal Representations Improve Goal-Conditioned Reinforcement Learning?
  39. Belief-Based Maximum Occupancy Principle and Active Inference also fits: RL theory and sample complexity

Foundations

Most cited by papers here

  1. Bigger, Regularized, Optimistic: scaling for compute and sample-efficient continuous control2024cited by 47 here
  2. Mixtures of Experts Unlock Parameter Scaling for Deep RL2024cited by 34 here
  3. Craftax: A Lightning-Fast Benchmark for Open-Ended Reinforcement Learning2024cited by 34 here
  4. Disentangling the Causes of Plasticity Loss in Neural Networks2024cited by 34 here
  5. SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning2024cited by 34 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2024Mixtures of Experts Unlock Parameter Scaling for Deep RLcited by 34 here
  2. 2024In value-based deep reinforcement learning, a pruned network is a good networkcited by 24 here
  3. 2024Stop Regressing: Training Value Functions via Classification for Scalable Deep RLcited by 33 here
  4. 2024Bigger, Regularized, Optimistic: scaling for compute and sample-efficient continuous controlcited by 47 here
  5. 2024Simplifying Deep Temporal Difference Learningcited by 32 here
  6. 2024SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learningcited by 34 here
  7. 20251000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilitiescited by 14 here
17 more it builds on, not in this chain
  1. 2024Bridging State and History Representations: Understanding Self-Predictive RLcited by 20 here
  2. 2024Exploration and Anti-Exploration with Distributional Random Network Distillationcited by 14 here
  3. 2024Foundation Policies with Hilbert Representationscited by 27 here
  4. 2024Craftax: A Lightning-Fast Benchmark for Open-Ended Reinforcement Learningcited by 34 here
  5. 2024Disentangling the Causes of Plasticity Loss in Neural Networkscited by 34 here
  6. 2024Overestimation, Overfitting, and Plasticity in Actor-Critic: the Bitter Lesson of Reinforcement Learningcited by 30 here
  7. 2024EfficientZero V2: Mastering Discrete and Continuous Control with Limited Datacited by 16 here
  8. 2024Reinforcement learningcited by 30 here
  9. 2024Diffusion for World Modeling: Visual Details Matter in Ataricited by 16 here
  10. 2024A Study of Plasticity Loss in On-Policy Deep Reinforcement Learningcited by 13 here
  11. 2024Learning Temporal Distances: Contrastive Successor Features Can Provide a Metric Structure for Decision-Makingcited by 23 here
  12. 2024Normalization and effective learning rates in reinforcement learningcited by 24 here
  13. 2024Gymnasium: A Standard Interface for Reinforcement Learning Environmentscited by 87 here
  14. 2024Deep Reinforcement Learning for Robotics: A Survey of Real-World Successescited by 18 here
  15. 2025Towards General-Purpose Model-Free Reinforcement Learningcited by 22 here
  16. 2025Hyperspherical Normalization for Scalable Deep Reinforcement Learningcited by 17 here
  17. 2025Zero-Shot Whole-Body Humanoid Control via Behavioral Foundation Modelscited by 14 here

The 24 most cited of 816.

Browse

30 of 1,369, ranked 6–35

  1. 6Stop Regressing: Training Value Functions via Classification for Scalable Deep RL202433
  2. 7Simplifying Deep Temporal Difference Learning202432
  3. 8Foundation Policies with Hilbert Representations202427
  4. 9In value-based deep reinforcement learning, a pruned network is a good network202424
  5. 10Normalization and effective learning rates in reinforcement learning202424
  6. 11Learning Temporal Distances: Contrastive Successor Features Can Provide a Metric Structure for Decision-Making202423
  7. 12Towards General-Purpose Model-Free Reinforcement Learning202522
  8. 13Bridging State and History Representations: Understanding Self-Predictive RL202420
  9. 14Hyperspherical Normalization for Scalable Deep Reinforcement Learning202517
  10. 15EfficientZero V2: Mastering Discrete and Continuous Control with Limited Data202416
All 30
  1. 161000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilities202514
  2. 17Zero-Shot Whole-Body Humanoid Control via Behavioral Foundation Models202514
  3. 18Exploration and Anti-Exploration with Distributional Random Network Distillation202414
  4. 19A Study of Plasticity Loss in On-Policy Deep Reinforcement Learning202413
  5. 20MaxInfoRL: Boosting exploration in reinforcement learning through information gain maximization202413
  6. 21Sharing Knowledge in Multi-Task Deep Reinforcement Learning202412
  7. 22On the consistency of hyper-parameter selection in value-based deep reinforcement learning202412
  8. 23Streaming Deep Reinforcement Learning Finally Works202412
  9. 24Mastering Memory Tasks with World Models202411
  10. 25Transformers Can Learn Temporal Difference Methods for In-Context Reinforcement Learning202410
  11. 26No Regrets: Investigating and Improving Regret Approximations for Curriculum Discovery202410
  12. 27Plasticity Loss in Deep Reinforcement Learning: A Survey202410
  13. 28AMAGO-2: Breaking the Multi-Task Barrier in Meta-Reinforcement Learning with Transformers202410
  14. 29Parseval Regularization for Continual Reinforcement Learning202410
  15. 30Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization20259
  16. 31A Survey of In-Context Reinforcement Learning20259
  17. 32Meta-World+: An Improved, Standardized, RL Benchmark20259
  18. 33XQC: Well-conditioned Optimization Accelerates Deep Reinforcement Learning20259
  19. 34Refining Minimax Regret for Unsupervised Environment Design20249
  20. 35Unsupervised Zero-Shot Reinforcement Learning via Functional Reward Encodings20249