mlsanityData to 14 Oct 2026

Action-Conditional Video Prediction using Deep Networks in Atari Games

arXiv 1507.08750 ↗First posted 31 Jul 2015cs.LG cs.AI cs.CVcited by 172

Not on the map. Why?

Cited by 39 in World models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Motivated by vision-based reinforcement learning (RL) problems, in particular Atari games from the recent benchmark Aracade Learning Environment (ALE), we consider spatio-temporal prediction problems where future (image-)frames are dependent on control variables or actions as well as previous frames. While not composed of natural scenes, frames in Atari games are high-dimensional in size, can involve tens of objects with one or more objects being controlled by the actions directly and many other objects being influenced indirectly, can involve entry and departure of objects, and can involve deep partial observability. We propose and evaluate two deep neural network architectures that consist of encoding, action-conditional transformation, and decoding layers based on convolutional neural networks and recurrent neural networks. Experimental results show that the proposed architectures are able to generate visually-realistic frames that are also useful for control over approximately 100-step action-conditional futures in some games. To the best of our knowledge, this paper is the first to make and evaluate long-term predictions on high-dimensional video conditioned by control inputs.

Key facts. Cited by 172 papers in the map (data through 14 Oct 2026). First posted on arXiv 31 Jul 2015.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2018: 2; 2019: 2; 2020: 3; 2021: 52; 2022: 23; 2023: 27; 2024: 27; 2025: 19; 2026: 17.

172 papers.

Sort:

By area:

  1. 2020Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problemscited by 1,920
  2. 2024Genie: Generative Interactive Environmentscited by 747
  3. 2018An Algorithmic Perspective on Imitation Learningcited by 459
  4. 2024Vista: A Generalizable Driving World Model with High Fidelity and Versatile Controllabilitycited by 347
  5. 2023Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Drivingcited by 319
  6. 2020Language Conditioned Imitation Learning over Unstructured Datacited by 209
  7. 2021Towards Causal Representation Learningcited by 209
  8. 2024MuirBench: A Comprehensive Benchmark for Robust Multi-image Understandingcited by 188
  9. 2021PredRNN: A Recurrent Neural Network for Spatiotemporal Predictive Learningcited by 182
  10. 2025Ctrl-World: A Controllable Generative World Model for Robot Manipulationcited by 167
  11. 2023Video Language Planningcited by 162
  12. 2024iVideoGPT: Interactive VideoGPTs are Scalable World Modelscited by 146
38 more
  1. 2025mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAscited by 142
  2. 2024Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web Navigationcited by 126
  3. 2021DriveGAN: Towards a Controllable High-Quality Neural Simulationcited by 125
  4. 2021Pathdreamer: A World Model for Indoor Navigationcited by 118
  5. 2022EgoTaskQA: Understanding Human Tasks in Egocentric Videoscited by 115
  6. 2022Reinforcement Learning with Action-Free Pre-Training from Videoscited by 108
  7. 2021FitVid: Overfitting in Pixel-Level Video Predictioncited by 87
  8. 2023DREAMWALKER: Mental Planning for Continuous Vision-Language Navigationcited by 85
  9. 2026PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulationcited by 82
  10. 2024World Models for Autonomous Driving: An Initial Surveycited by 80
  11. 2021Greedy Hierarchical Variational Autoencoders for Large-Scale Video Predictioncited by 75
  12. 2020Safe Imitation Learning via Fast Bayesian Reward Inference from Preferencescited by 73
  13. 2022BYOL-Explore: Exploration by Bootstrapped Predictioncited by 73
  14. 2021Playable Video Generationcited by 64
  15. 2025Long-Context State-Space Video World Modelscited by 64
  16. 2021Predicting Video with VQVAEcited by 61
  17. 2025Embodied AI Agents: Modeling the Worldcited by 61
  18. 2021Exploration in Deep Reinforcement Learning: From Single-Agent to Multiagent Domaincited by 58
  19. 2023Self-Supervised Multimodal Learning: A Surveycited by 58
  20. 2025Back to the Features: DINO as a Foundation for Video World Modelscited by 58
  21. 2021MotionRNN: A Flexible Model for Video Prediction with Spacetime-Varying Motionscited by 54
  22. 2021Muesli: Combining Improvements in Policy Optimizationcited by 53
  23. 2022Iso-Dream: Isolating and Leveraging Noncontrollable Visual Dynamics in World Modelscited by 53
  24. 2026World Action Models: The Next Frontier in Embodied AIcited by 50
  25. 2021Vector Quantized Models for Planningcited by 48
  26. 2023Pre-training Contextualized World Models with In-the-wild Videos for Reinforcement Learningcited by 45
  27. 2021Human-Level Reinforcement Learning through Theory-Based Modeling, Exploration, and Planningcited by 43
  28. 2022Evaluating Long-Term Memory in 3D Mazescited by 43
  29. 2024Dynamic 3D Gaussian Tracking for Graph-Based Neural Dynamics Modelingcited by 43
  30. 2021Clockwork Variational Autoencoderscited by 42
  31. 2023Paxion: Patching Action Knowledge in Video-Language Foundation Modelscited by 42
  32. 2025Towards General-Purpose Model-Free Reinforcement Learningcited by 41
  33. 2024Autoregressive Models in Vision: A Surveycited by 39
  34. 2021Meta-Learning Dynamics Forecasting Using Task Inferencecited by 38
  35. 2026Temporal Straightening for Latent Planningcited by 38
  36. 2021EarthNet2021: A large-scale dataset and challenge for Earth surface forecasting as a guided video prediction taskcited by 34
  37. 2021Learning Knowledge Graph-based World Models of Textual Environmentscited by 31
  38. 2022Joint Embedding Predictive Architectures Focus on Slow Featurescited by 30

50 most cited of 172 All 172 (JSON)