Awesome Model-Based Reinforcement Learning

This is a collection of research papers for model-based reinforcement learning (mbrl). And the repository will be continuously updated to track the frontier of model-based rl.

Welcome to follow and star!

[2024.10.27] New: We update the NeurIPS 2024 paper list of model-based rl!

[2024.05.20] We update the ICML 2024 paper list of model-based rl.

[2023.11.29] We update the ICLR 2024 paper list of model-based rl.

[2023.09.29] We update the NeurIPS 2023 paper list of model-based rl.

[2023.06.15] We update the ICML 2023 paper list of model-based rl.

[2023.02.05] We update the ICLR 2023 paper list of model-based rl.

[2022.11.03] We update the NeurIPS 2022 paper list of model-based rl.

[2022.07.06] We update the ICML 2022 paper list of model-based rl.

[2022.02.13] We update the ICLR 2022 paper list of model-based rl.

[2021.12.28] We release the awesome model-based rl.

Table of Contents

• Awesome Model-Based Reinforcement Learning
  • Table of Contents
  • A Taxonomy of Model-Based RL Algorithms
  • Papers
    • Classic Model-Based RL Papers
    • NeurIPS 2024🔥
    • ICML 2024
    • ICLR 2024
    • NeurIPS 2023
    • ICML 2023
    • ICLR 2023
    • NeurIPS 2022
    • ICML 2022
    • ICLR 2022
    • NeurIPS 2021
    • ICLR 2021
    • ICML 2021
    • Other
  • Tutorial
  • Codebase
  • Contributing
  • License

A Taxonomy of Model-Based RL Algorithms

We’ll start this section with a disclaimer: it’s really quite hard to draw an accurate, all-encompassing taxonomy of algorithms in the Model-Based RL space, because the modularity of algorithms is not well-represented by a tree structure. So we will publish a series of related blogs to explain more Model-Based RL algorithms.

A non-exhaustive, but useful taxonomy of algorithms in modern Model-Based RL.

We simply divide Model-Based RL into two categories: Learn the Model and Given the Model.

• Learn the Model mainly focuses on how to build the environment model.

• Given the Model cares about how to utilize the learned model.

And we give some examples as shown in the figure above. There are links to algorithms in taxonomy.

[1] World Models: Ha and Schmidhuber, 2018
[2] I2A (Imagination-Augmented Agents): Weber et al, 2017
[3] MBMF (Model-Based RL with Model-Free Fine-Tuning): Nagabandi et al, 2017
[4] MBVE (Model-Based Value Expansion): Feinberg et al, 2018
[5] ExIt (Expert Iteration): Anthony et al, 2017
[6] AlphaZero: Silver et al, 2017
[7] POPLIN (Model-Based Policy Planning): Wang et al, 2019
[8] M2AC (Masked Model-based Actor-Critic): Pan et al, 2020

Papers

format:
- [title](paper link) [links]
  - author1, author2, and author3
  - Key: key problems and insights
  - OpenReview: optional
  - ExpEnv: experiment environments

Classic Model-Based RL Papers

Toggle

• Dyna, an integrated architecture for learning, planning, and reacting

  • Richard S. Sutton. ACM 1991
  • Key: dyna architecture
  • ExpEnv: None

• PILCO: A Model-Based and Data-Efficient Approach to Policy Search

  • Marc Peter Deisenroth, Carl Edward Rasmussen. ICML 2011
  • Key: probabilistic dynamics model
  • ExpEnv: cart-pole system, robotic unicycle

• Learning Complex Neural Network Policies with Trajectory Optimization

  • Sergey Levine, Vladlen Koltun. ICML 2014
  • Key: guided policy search
  • ExpEnv: mujoco

• Learning Continuous Control Policies by Stochastic Value Gradients

  • Nicolas Heess, Greg Wayne, David Silver, Timothy Lillicrap, Yuval Tassa, Tom Erez. NIPS 2015
  • Key: backpropagation through paths, gradient on real trajectory
  • ExpEnv: mujoco

• Value Prediction Network

  • Junhyuk Oh, Satinder Singh, Honglak Lee. NIPS 2017
  • Key: value-prediction model
  • ExpEnv: collect domain, atari

• Sample-Efficient Reinforcement Learning with Stochastic Ensemble Value Expansion

  • Jacob Buckman, Danijar Hafner, George Tucker, Eugene Brevdo, Honglak Lee. NIPS 2018
  • Key: ensemble model and Qnet, value expansion
  • ExpEnv: mujoco, roboschool

• Recurrent World Models Facilitate Policy Evolution

  • David Ha, Jürgen Schmidhuber. NIPS 2018
  • Key: vae(representation), rnn(predictive model)
  • ExpEnv: car racing, vizdoom

• Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models

  • Kurtland Chua, Roberto Calandra, Rowan McAllister, Sergey Levine. NIPS 2018
  • Key: probabilistic ensembles with trajectory sampling
  • ExpEnv: cartpole, mujoco

• When to Trust Your Model: Model-Based Policy Optimization

  • Michael Janner, Justin Fu, Marvin Zhang, Sergey Levine. NeurIPS 2019
  • Key: ensemble model, sac, k-branched rollout
  • ExpEnv: mujoco

• Algorithmic Framework for Model-based Deep Reinforcement Learning with Theoretical Guarantees

  • Yuping Luo, Huazhe Xu, Yuanzhi Li, Yuandong Tian, Trevor Darrell, Tengyu Ma. ICLR 2019
  • Key: Discrepancy Bounds Design, ME-TRPO with multi-step, Entropy regularization
  • ExpEnv: mujoco

• Model-Ensemble Trust-Region Policy Optimization

  • Thanard Kurutach, Ignasi Clavera, Yan Duan, Aviv Tamar, Pieter Abbeel. ICLR 2018
  • Key: ensemble model, TRPO
  • ExpEnv: mujoco

• Dream to Control: Learning Behaviors by Latent Imagination

  • Danijar Hafner, Timothy Lillicrap, Jimmy Ba, Mohammad Norouzi. ICLR 2019
  • Key: DreamerV1, latent space imagination
  • ExpEnv: deepmind control suite, atari, deepmind lab

• Exploring Model-based Planning with Policy Networks

  • Tingwu Wang, Jimmy Ba. ICLR 2020
  • Key: model-based policy planning in action space and parameter space
  • ExpEnv: mujoco

• Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model

  • Julian Schrittwieser, Ioannis Antonoglou, Thomas Hubert, Karen Simonyan, Laurent Sifre, Simon Schmitt, Arthur Guez, Edward Lockhart, Demis Hassabis, Thore Graepel, Timothy Lillicrap, David Silver. Nature 2020
  • Key: MCTS, value equivalence
  • ExpEnv: chess, shogi, go, atari

NeurIPS 2024

Toggle

• The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning

  • Moritz Schneider, Robert Krug, Narunas Vaskevicius, Luigi Palmieri, Joschka Boedecker
  • Key: reinforcement learning, rl, model-based reinforcement learning, representation learning, pvr, visual representations
  • ExpEnv: DMC, ManiSkill2, Miniworld

• Multi-Agent Domain Calibration with a Handful of Offline Data

  • Tao Jiang, Lei Yuan, Lihe Li, Cong Guan, Zongzhang Zhang, Yang Yu
  • Key: Multi-agent reinforcement learning, domain transfer
  • ExpEnv: D4RL

• WorldCoder, a Model-Based LLM Agent: Building World Models by Writing Code and Interacting with the Environment

  • Hao Tang, Darren Key, Kevin Ellis
  • Key: learn world models as code, LLM
  • ExpEnv: sokoban, minigrid, alfworld

• The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning

  • Anya Sims, Cong Lu, Jakob Foerster, Yee Whye Teh
  • Key: edge-of-reach problem, reach-aware value learning
  • ExpEnv: d4rl, v-r4rl

• Deterministic Uncertainty Propagation for Improved Model-Based Offline Reinforcement Learning

  • Abdullah Akgül, Manuel Haussmann, Melih Kandemir
  • Key: The paper argues that uncertainty-based reward penalization introduces excessive conservatism, potentially resulting in suboptimal policies through underestimation.
  • ExpEnv: d4rl

• BECAUSE: Bilinear Causal Representation for Generalizable Offline Model-based Reinforcement Learning

  • Haohong Lin, Wenhao Ding, Jian Chen, Laixi Shi, Jiacheng Zhu, Bo Li, DING ZHAO
  • Key: objective mismatch problem, capture causal representation for both states and actions
  • ExpEnv: list, unlock, crash

• Model-Based Transfer Learning for Contextual Reinforcement Learning

  • Jung-Hoon Cho, Vindula Jayawardana, Sirui Li, Cathy Wu
  • Key: bayesian optimization, contextual rl
  • ExpEnv: gaussian process, traffic signal, eco-driving, advisory autonomy, control tasks

• Rethinking Model-based, Policy-based, and Value-based Reinforcement Learning via the Lens of Representation Complexity

  • Guhao Feng, Han Zhong
  • Key: rl representation complexity
  • ExpEnv: mujoco

ICML 2024

Toggle

• HarmonyDream: Task Harmonization Inside World Models

  • Haoyu Ma, Jialong Wu, Ningya Feng, Chenjun Xiao, Dong Li, Jianye Hao, Jianmin Wang, Mingsheng Long
  • Key: observation modeling and reward modeling analysis in world models
  • ExpEnv: meta-world, rlbench, deepmind control suite, atari 100k

• 3D-VLA: A 3D Vision-Language-Action Generative World Model

  • Haoyu Zhen, Xiaowen Qiu, Peihao Chen, Jincheng Yang, Xin Yan, Yilun Du, Yining Hong, Chuang Gan
  • Key: unify 3D perception, reasoning, and action with a generative world model; create a large-scale 3D embodied instruction tuning dataset
  • ExpEnv: rlbench, calvin

• CompeteAI: Understanding the Competition Behaviors in Large Language Model-based Agents

  • Qinlin Zhao, Jindong Wang, Yixuan Zhang, Yiqiao Jin, Kaijie Zhu, Hao Chen, Xing Xie
  • Key: propose a competitive framework for LLM-based agents; build a simulated competitive environment
  • ExpEnv: a virtual town with only restaurants and customers

• Model-based Reinforcement Learning for Parameterized Action Spaces

  • Renhao Zhang, Haotian Fu, Yilin Miao, George Konidaris
  • Key: discrete-continuous hybrid action space, dynamics model with parameterized actions, MPC with parameterized actions
  • ExpEnv: platform, goal, hard goal, catch point, hard move

• Learning Latent Dynamic Robust Representations for World Models

  • Ruixiang Sun, Hongyu Zang, Xin Li, Riashat Islam
  • Key: modified Dreamer architecture, hybrid-recurrent state space model
  • ExpEnv: deepmind control suite, distracted deepmind control suite, mani-skill2

• AD3: Implicit Action is the Key for World Models to Distinguish the Diverse Visual Distractors

  • Yucen Wang, Shenghua Wan, Le Gan, Shuai Feng, De-Chuan Zhan
  • Key: implicit action generator, action-conditioned separated world models
  • ExpEnv: deepmind control suite

• Hieros: Hierarchical Imagination on Structured State Space Sequence World Models

  • Paul Mattes, Rainer Schlosser, Ralf Herbrich
  • Key: state-space models, multilayered hierarchical imagination, S5 based world model
  • ExpEnv: atari 100k

• Improving Token-Based World Models with Parallel Observation Prediction

  • Lior Cohen, Kaixin Wang, Bingyi Kang, Shie Mannor
  • Key: pixel-based mbrl, token-based world models, retentive environment model
  • ExpEnv: atari 100k

• Do Transformer World Models Give Better Policy Gradients?

  • Michel Ma, Tianwei Ni, Clement Gehring, Pierluca D'Oro, Pierre-Luc Bacon
  • Key: actions world model
  • ExpEnv: double-pendulum, Myriad

• Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming

  • Hany Hamed, Subin Kim, Dongyeong Kim, Jaesik Yoon, Sungjin Ahn
  • Key: during strategeic dreaming, train three policies -- highway policy, explorer policy and achiever policy, and then achieve downstream tasks
  • ExpEnv: 2D Navigation, 3D-Maze Navigation, RoboKitchen

• Towards Robust Model-Based Reinforcement Learning Against Adversarial Corruption

  • Chenlu Ye, Jiafan He, Quanquan Gu, Tong Zhang
  • Key: theoretical analysis of adversarial corruption for model-based rl, encompassing both online and offline settings
  • ExpEnv: None

• Model-based Reinforcement Learning for Confounded POMDPs

  • Mao Hong, Zhengling Qi, Yanxun Xu
  • Key: model-based RL, POMDP
  • ExpEnv: None

ICLR 2024

Toggle

• Policy Rehearsing: Training Generalizable Policies for Reinforcement Learning

  • Chengxing Jia, Chenxiao Gao, Hao Yin, Fuxiang Zhang, Xiong-Hui Chen, Tian Xu, Lei Yuan, Zongzhang Zhang, Zhi-Hua Zhou, Yang Yu
  • Key: Reinforcement Learning, Model-based Reinforcement Learning, Offline Reinforcement Learning
  • OpenReview: 8, 8, 8, 6
  • ExpEnv: d4rl

• Efficient Dynamics Modeling in Interactive Environments with Koopman Theory

  • Arnab Kumar Mondal, Siba Smarak Panigrahi, Sai Rajeswar, Kaleem Siddiqi, Siamak Ravanbakhsh
  • Key: Koopman Theory, Reinforcement Learning, Dynamical System, Planning, Longe range dynamics prediction models, Efficient forward dynamics
  • OpenReview: 8, 6, 5, 3
  • ExpEnv: mujoco

• Combining Spatial and Temporal Abstraction in Planning for Better Generalization

  • Mingde Zhao, Safa Alver, Harm van Seijen, Romain Laroche, Doina Precup, Yoshua Bengio
  • Key: Reinforcement Learning, Planning, Neural Networks, Temporal Difference Learning, Generalization, Deep Reinforcement Learning
  • OpenReview: 6, 6, 6, 5
  • ExpEnv: MiniGrid-BabyAI framework

• Mastering Memory Tasks with World Models

  • Mohammad Reza Samsami, Artem Zholus, Janarthanan Rajendran, Sarath Chandar
  • Key: recall to imagine module, based on DreamerV3
  • OpenReview: 10, 8, 6
  • ExpEnv: bsuite, popgym, atari, deepmind control suite, memory maze

• Privileged Sensing Scaffolds Reinforcement Learning

  • Edward S. Hu, James Springer, Oleh Rybkin, Dinesh Jayaraman
  • Key: privileged information, based on DreamerV3
  • OpenReview: 10, 8, 8, 8
  • ExpEnv: gymnasium robotics

• TD-MPC2: Scalable, Robust World Models for Continuous Control

  • Nicklas Hansen, Hao Su, Xiaolong Wang
  • Key: implicit world model, model predictive control, generalist td-mpc2
  • OpenReview: 8, 8, 8, 8
  • ExpEnv: deepmind control suite, Meta-World, maniskill2, myosuite

• Robust Model Based Reinforcement Learning Using L1 Adaptive Control

  • Minjun Sung, Sambhu Harimanas Karumanchi, Aditya Gahlawat, Naira Hovakimyan
  • Key: L1 Adaptive Control
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: mujoco

• Learning Hierarchical World Models with Adaptive Temporal Abstractions from Discrete Latent Dynamics

  • Christian Gumbsch, Noor Sajid, Georg Martius, Martin V. Butz
  • Key: Context-specific Recurrent State Space Model, hierarchical world model
  • OpenReview: 8, 6, 6
  • ExpEnv: MiniHack, VisualPinPad, MultiWorld

• Learning Unsupervised World Models for Autonomous Driving via Discrete Diffusion

  • Lunjun Zhang, Yuwen Xiong, Ze Yang, Sergio Casas, Rui Hu, Raquel Urtasun
  • Key: discrete diffusion; world model; autonomous driving
  • OpenReview: 10, 8, 6, 6, 6
  • ExpEnv: NuScenes, KITTI Odometry, Argoverse2 Lidar

• COPlanner: Plan to Roll Out Conservatively but to Explore Optimistically for Model-Based RL

  • Xiyao Wang, Ruijie Zheng, Yanchao Sun, Ruonan Jia, Wichayaporn Wongkamjan, Huazhe Xu, Furong Huang
  • Key: conservative model rollouts, optimistic environment exploration
  • OpenReview: 6, 6, 6
  • ExpEnv: mujoco, deepmind control suite

• Efficient Multi-agent Reinforcement Learning by Planning

  • Qihan Liu, Jianing Ye, Xiaoteng Ma, Jun Yang, Bin Liang, Chongjie Zhang
  • Key: mcts, optimistic search lambda, advantage-weighted policy optimization
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: smac

• Differentiable Trajectory Optimization as a Policy Class for Reinforcement and Imitation Learning

  • Weikang Wan, Yufei Wang, Zackory Erickson, David Held
  • Key: differentiable trajectory optimization
  • OpenReview: 10, 8, 8, 5
  • ExpEnv: deepmind control suite, robomimic, maniskill

• DMBP: Diffusion model based predictor for robust offline reinforcement learning against state observation perturbations

  • Zhihe YANG, Yunjian Xu
  • Key: conditional diffusion, offline RL
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: d4rl

• MAMBA: an Effective World Model Approach for Meta-Reinforcement Learning

  • Zohar Rimon, Tom Jurgenson, Orr Krupnik, Gilad Adler, Aviv Tamar
  • Key: context-based meta-RL, based on dreamer
  • OpenReview: 6, 6, 6, 6
  • ExpEnv: Point Robot Navigation, Escape Room, Reacher Sparse

• Reward-Consistent Dynamics Models are Strongly Generalizable for Offline Reinforcement Learning

  • Fan-Ming Luo, Tian Xu, Xingchen Cao, Yang Yu
  • Key: reward learning, offline RL
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: d4rl, NeoRL

• DreamSmooth: Improving Model-based Reinforcement Learning via Reward Smoothing

  • Vint Lee, Pieter Abbeel, Youngwoon Lee
  • Key: learn to predict a temporally-smoothed reward rather than the exact reward at each timestep
  • OpenReview: 6, 6, 6, 5
  • ExpEnv: robodesk, hand, earthmoving

• Informed POMDP: Leveraging Additional Information in Model-Based RL

  • Gaspard Lambrechts, Adrien Bolland, Damien Ernst
  • Key: informed world model, based on DreamerV3
  • OpenReview: 6, 6, 6, 5
  • ExpEnv: varying mountain hike, deepmind control suite, pop gym, flickering atari and flickering control

NeurIPS 2023

Toggle

• Large Language Models as Commonsense Knowledge for Large-Scale Task Planning

  • Zirui Zhao, Wee Sun Lee, David Hsu
  • Key: LLM-MCTS
  • ExpEnv: VirtualHome

• Describe, Explain, Plan and Select: Interactive Planning with LLMs Enables Open-World Multi-Task Agents

  • Zihao Wang, Shaofei Cai, Guanzhou Chen, Anji Liu, Xiaojian (Shawn) Ma, Yitao Liang
  • Key: interactive planning approach based on LLM
  • ExpEnv: minecraft

• Facing Off World Model Backbones: RNNs, Transformers, and S4

  • Fei Deng, Junyeong Park, Sungjin Ahn
  • Key: world model backbones
  • ExpEnv: MiniGrid, memory maze

• Pre-training Contextualized World Models with In-the-wild Videos for Reinforcement Learning

  • Jialong Wu, Haoyu Ma, Chaoyi Deng, Mingsheng Long
  • Key: Contextualized World Models
  • ExpEnv: CARLA, deepmind control suite

• Conformal Prediction for Uncertainty-Aware Planning with Diffusion Dynamics Model

  • Jiankai Sun, Yiqi Jiang, Jianing Qiu, Parth Nobel, Mykel J Kochenderfer, Mac Schwager
  • Key: Diffusion Dynamics Model
  • ExpEnv: d4rl, Maze2D

• LightZero: A Unified Benchmark for Monte Carlo Tree Search in General Sequential Decision Scenarios

  • Yazhe Niu, Yuan Pu, Zhenjie Yang, Xueyan Li, Tong Zhou, Jiyuan Ren, Shuai Hu, Hongsheng Li, Yu Liu
  • Key: MCTS-style benchmark
  • ExpEnv: board games, atari, mujoco, gobigger

• Diffusion Model is an Effective Planner and Data Synthesizer for Multi-Task Reinforcement Learning

  • Haoran He, Chenjia Bai, Kang Xu, Zhuoran Yang, Weinan Zhang, Dong Wang, Bin Zhao, Xuelong Li
  • Key: GPT-based diffusion model for planning and data synthesizing
  • ExpEnv: Meta-World, Maze2D

• MoVie: Visual Model-Based Policy Adaptation for View Generalization

  • Sizhe Yang, Yanjie Ze, Huazhe Xu
  • Key: view generalization, spatial adaptive encoder
  • ExpEnv: deepmind control suite, adroit, xArm

• Model-Based Reparameterization Policy Gradient Methods: Theory and Practical Algorithms

  • Shenao Zhang, Boyi Liu, Zhaoran Wang, Tuo Zhao
  • Key: model-based reparameterization policy gradient method, smoothness regularization
  • ExpEnv: mujoco

• Leveraging Pre-trained Large Language Models to Construct and Utilize World Models for Model-based Task Planning

  • Lin Guan, Karthik Valmeekam, Sarath Sreedharan, Subbarao Kambhampati
  • Key: construct an explicit world (domain) model in planning domain definition language
  • ExpEnv: household-robot domain, tyreworld and logistics

• RePo: Resilient Model-Based Reinforcement Learning by Regularizing Posterior Predictability

  • Chuning Zhu, Max Simchowitz, Siri Gadipudi, Abhishek Gupta
  • Key: representation resilience for visual RL
  • ExpEnv: deepmind control suite, maniskill

• Model-Based Control with Sparse Neural Dynamics

  • Ziang Liu, Jeff He, Genggeng Zhou, Tobia Marcucci, Fei-Fei Li, Jiajun Wu, Yunzhu Li
  • Key: network sparsification, mixed-integer formulation of ReLU neural dynamics
  • ExpEnv: gym, cartpole, reacher

• Optimal Exploration for Model-Based RL in Nonlinear Systems

  • Andrew Wagenmaker, Guanya Shi, Kevin Jamieson
  • Key: optimal sample complexity for nonlinear dynamical systems
  • ExpEnv: affine dynamics system

• State2Explanation: Concept-Based Explanations to Benefit Agent Learning and User Understanding

  • Devleena Das, Sonia Chernova, Been Kim
  • Key: a joint embedding model between state-action pairs and concept-based explanations
  • ExpEnv: connect4, lunar lander

• Efficient Exploration in Continuous-time Model-based Reinforcement Learning

  • Lenart Treven, Jonas Hübotter, Bhavya, Florian Dorfler, Andreas Krause
  • Key: nonlinear ordinary differential equations, regret bound, measurement selection strategies
  • ExpEnv: system’s tasks

• Action Inference by Maximising Evidence: Zero-Shot Imitation from Observation with World Models

  • Xingyuan Zhang, Philip Becker-Ehmck, Patrick van der Smagt, Maximilian Karl
  • Key: pretrained world models, imitation learning from observation only
  • ExpEnv: deepmind control suite

• STORM: Efficient Stochastic Transformer based World Models for Reinforcement Learning

  • Weipu Zhang, Gang Wang, Jian Sun, Yetian Yuan, Gao Huang
  • Key: categorical-VAE, transformer structure, DreamerV3
  • ExpEnv: atari

ICML 2023

Toggle

• Mastering the Unsupervised Reinforcement Learning Benchmark from Pixels

  • Sai Rajeswar Mudumba, Pietro Mazzaglia, Tim Verbelen, Alexandre Piche, Bart Dhoedt, Aaron Courville, Alexandre Lacoste
  • Key: unsupervised pretrain, task-aware finetune, dyna-mpc
  • ExpEnv: URLB benchmark, RWRL suite

• Reparameterized Policy Learning for Multimodal Trajectory Optimization

  • Zhiao Huang, Litian Liang, Zhan Ling, Xuanlin Li, Chuang Gan, Hao Su
  • Key: multimodal policy learning, reparameterized policy gradient
  • ExpEnv: Meta-World, mujoco

• Live in the Moment: Learning Dynamics Model Adapted to Evolving Policy

  • Xiyao Wang, Wichayaporn Wongkamjan, Ruonan Jia, Furong Huang
  • Key: policy-adapted model learning, weight design
  • ExpEnv: mujoco

• Predictable MDP Abstraction for Unsupervised Model-Based RL

  • Seohong Park, Sergey Levine
  • Key: predictable MDP abstraction, tackle model exploitation
  • ExpEnv: mujoco

• Investigating the Role of Model-Based Learning in Exploration and Transfer

  • Jacob C Walker, Eszter Vértes, Yazhe Li, Gabriel Dulac-Arnold, Ankesh Anand, Jessica Hamrick, Theophane Weber
  • Key Insights: (1) Is there an advantage to an agent being model-based during unsupervised exploration and/or fine-tuning? (2) What are the contributions of each component of a model-based agent for downstream task learning? (3) How well does the model-based agent deal with environmental shift between the unsupervised and downstream phases?
  • ExpEnv: Crafter, RoboDesk, Meta-World

• The Virtues of Laziness in Model-based RL: A Unified Objective and Algorithms

  • Anirudh Vemula, Yuda Song, Aarti Singh, J. Bagnell, Sanjiban Choudhury
  • Key: objective mismatch, mbrl framework
  • ExpEnv: Helicopter, WideTree, Linear Dynamical System, Maze, mujoco

• The Benefits of Model-Based Generalization in Reinforcement Learning

  • Kenny Young, Aditya Ramesh, Louis Kirsch, Jürgen Schmidhuber
  • Key: experience replay, when and how learned model generalization
  • ExpEnv: ProcMaze, ButtonGrid, PanFlute

• STEERING: Stein Information Directed Exploration for Model-Based Reinforcement Learning

  • Souradip Chakraborty, Amrit Bedi, Alec Koppel, Mengdi Wang, Furong Huang, Dinesh Manocha
  • Key: information directed sampling, kernelized Stein discrepancy
  • ExpEnv: DeepSea

• Model-based Reinforcement Learning with Scalable Composite Policy Gradient Estimators

  • Paavo Parmas, Takuma Seno, Yuma Aoki
  • Key: extension of Dreamer, total propagation computation graph
  • ExpEnv: deepmind control suite

• Reinforcement Learning with History Dependent Dynamic Contexts

  • Guy Tennenholtz, Nadav Merlis, Lior Shani, Martin Mladenov, Craig Boutilier
  • Key: non-Markov context dynamics, logistic DCMDPs, theoretical analysis, extension of MuZero
  • ExpEnv: MovieLens dataset

• Model-Bellman Inconsistency for Model-based Offline Reinforcement Learning

  • Yihao Sun, Jiaji Zhang, Chengxing Jia, Haoxin Lin, Junyin Ye, Yang Yu
  • Key: pessimistic value estimation, theoretical analysis
  • ExpEnv: d4rl, NeoRL

• Simplified Temporal Consistency Reinforcement Learning

  • Yi Zhao, Wenshuai Zhao, Rinu Boney, Juho Kannala, Joni Pajarinen
  • Key: representation learning, temporal consistency
  • ExpEnv: deepmind control suite

• Curious Replay for Model-based Adaptation

  • Isaac Kauvar, Chris Doyle, Linqi Zhou, Nick Haber
  • Key: extension of DreamerV3, curious replay, count-based replay, adversarial replay
  • ExpEnv: Crafter, deepmind control suite

• On Many-Actions Policy Gradient

  • Michal Nauman, Marek Cygan
  • Key: bias and variance, theoretical analysis
  • ExpEnv: deepmind control suite

• Posterior Sampling for Deep Reinforcement Learning

  • Remo Sasso, Michelangelo Conserva, Paulo Rauber
  • Key: posterior sampling, continual value network
  • ExpEnv: atari

• Model-based Offline Reinforcement Learning with Count-based Conservatism

  • Byeongchan Kim, Min-hwan Oh
  • Key: count estimation, theoretical analysis
  • ExpEnv: d4rl

ICLR 2023

Toggle

• Transformers are Sample-Efficient World Models

  • Vincent Micheli, Eloi Alonso, François Fleuret
  • Key: discrete autoencoder, transformer based world model
  • OpenReview: 8, 8, 8, 8
  • ExpEnv: atari

• Conservative Bayesian Model-Based Value Expansion for Offline Policy Optimization

  • Jihwan Jeong, Xiaoyu Wang, Michael Gimelfarb, Hyunwoo Kim, Baher Abdulhai, Scott Sanner
  • Key: model-based offline, bayesian posterior value estimate
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: d4rl

• User-Interactive Offline Reinforcement Learning

  • Phillip Swazinna, Steffen Udluft, Thomas Runkler
  • Key: let the user adapt the policy behavior after training is finished
  • OpenReview: 10, 8, 6, 3
  • ExpEnv: 2d-world, industrial benchmark

• CLARE: Conservative Model-Based Reward Learning for Offline Inverse Reinforcement Learning

  • Sheng Yue, Guanbo Wang, Wei Shao, Zhaofeng Zhang, Sen Lin, Ju Ren, Junshan Zhang
  • Key: offline IRL, reward extrapolation error
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: d4rl

• Efficient Offline Policy Optimization with a Learned Model

  • Zichen Liu, Siyi Li, Wee Sun Lee, Shuicheng YAN, Zhongwen Xu
  • Key: offline rl, analysis of MuZero Unplugged, one-step look-ahead policy improvement
  • OpenReview: 8, 6, 5
  • ExpEnv: atari dataset

• Efficient Planning in a Compact Latent Action Space

  • zhengyao jiang, Tianjun Zhang, Michael Janner, Yueying Li, Tim Rocktäschel, Edward Grefenstette, Yuandong Tian
  • Key: planning with VQ-VAE
  • OpenReview: 6, 6, 6, 6
  • ExpEnv: d4rl dataset

• Is Model Ensemble Necessary? Model-based RL via a Single Model with Lipschitz Regularized Value Function

  • Ruijie Zheng, Xiyao Wang, Huazhe Xu, Furong Huang
  • Key: lipschitz regularization
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: mujoco

• MoDem: Accelerating Visual Model-Based Reinforcement Learning with Demonstrations

  • Nicklas Hansen, Yixin Lin, Hao Su, Xiaolong Wang, Vikash Kumar, Aravind Rajeswaran
  • Key: three phases -- policy pretraining, targeted exploration, interactive learning
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: adroit, meta-world, deepmind control suite

• Simplifying Model-based RL: Learning Representations, Latent-space Models, and Policies with One Objective

  • Raj Ghugare, Homanga Bharadhwaj, Benjamin Eysenbach, Sergey Levine, Ruslan Salakhutdinov
  • Key: Aligned Latent Models
  • OpenReview: 8, 6, 6, 6, 6
  • ExpEnv: mujoco

• Diminishing Return of Value Expansion Methods in Model-Based Reinforcement Learning

  • Daniel Palenicek, Michael Lutter, Joao Carvalho, Jan Peters
  • Key: longer horizons yield diminishing returns in terms of sample efficiency
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: brax

• Planning Goals for Exploration

  • Edward S. Hu, Richard Chang, Oleh Rybkin, Dinesh Jayaraman
  • Key: sampling-based planning, set goals for each training episode to directly optimize an intrinsic exploration reward
  • OpenReview: 8, 8, 8, 8, 6
  • ExpEnv: point maze, walker, ant maze, 3-block stack

• Making Better Decision by Directly Planning in Continuous Control

  • Jinhua Zhu, Yue Wang, Lijun Wu, Tao Qin, Wengang Zhou, Tie-Yan Liu, Houqiang Li
  • Key: deep differentiable dynamic programming planner
  • OpenReview: 8, 8, 8, 6
  • ExpEnv: mujoco

• Latent Variable Representation for Reinforcement Learning

  • Tongzheng Ren, Chenjun Xiao, Tianjun Zhang, Na Li, Zhaoran Wang, sujay sanghavi, Dale Schuurmans, Bo Dai
  • Key: variational learning, representation learning
  • OpenReview: 8, 6, 6, 3
  • ExpEnv: mujoco, deepmind control suite

• SpeedyZero: Mastering Atari with Limited Data and Time

  • Yixuan Mei, Jiaxuan Gao, Weirui Ye, Shaohuai Liu, Yang Gao, Yi Wu
  • Key: distributed model-based rl, speed up EfficientZero
  • OpenReview: 6, 6, 5
  • ExpEnv: atari 100k

• Transformer-based World Models Are Happy With 100k Interactions

  • Jan Robine, Marc Höftmann, Tobias Uelwer, Stefan Harmeling
  • Key: autoregressive world model, Transformer-XL, balanced cross-entropy loss, balanced dataset sampling
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: atari 100k

• On the Feasibility of Cross-Task Transfer with Model-Based Reinforcement Learning

  • Yifan Xu, Nicklas Hansen, Zirui Wang, Yung-Chieh Chan, Hao Su, Zhuowen Tu
  • Key: offline multi-task pretraining, online finetuning
  • OpenReview: 6, 6, 6, 6
  • ExpEnv: atari 100k

• Become a Proficient Player with Limited Data through Watching Pure Videos

  • Weirui Ye, Yunsheng Zhang, Pieter Abbeel, Yang Gao
  • Key: unsupervised pre-training, finetune with down-stream tasks
  • OpenReview: 8, 6, 6, 5
  • ExpEnv: atari 100k

• EUCLID: Towards Efficient Unsupervised Reinforcement Learning with Multi-choice Dynamics Model

  • Yifu Yuan, Jianye HAO, Fei Ni, Yao Mu, YAN ZHENG, Yujing Hu, Jinyi Liu, Yingfeng Chen, Changjie Fan
  • Key: jointly pretrain the multi-headed dynamics model and unsupervised exploration policy, finetune to downstream tasks
  • OpenReview: 6, 6, 6, 6
  • ExpEnv: URLB benchmark

• Choreographer: Learning and Adapting Skills in Imagination

  • Pietro Mazzaglia, Tim Verbelen, Bart Dhoedt, Alexandre Lacoste, Sai Rajeswar
  • Key: world model, skill discovery, skill learning, Skill adaptation
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: deepmind control suite, Meta-World

NeurIPS 2022

Toggle

• Bidirectional Learning for Offline Infinite-width Model-based Optimization

  • Can Chen, Yingxue Zhang, Jie Fu, Xue Liu, Mark Coates
  • Key: model-based, offline
  • OpenReview: 7, 6, 5
  • ExpEnv: design-bench

• A Unified Framework for Alternating Offline Model Training and Policy Learning

  • Shentao Yang, Shujian Zhang, Yihao Feng, Mingyuan Zhou
  • Key: model-based, offline, marginal importance weight
  • OpenReview: 7, 6, 6, 5
  • ExpEnv: d4rl dataset

• Model-Based Offline Reinforcement Learning with Pessimism-Modulated Dynamics Belief

  • Kaiyang Guo, Shao Yunfeng, Yanhui Geng
  • Key: model-based, offline
  • OpenReview: 8, 8, 7, 7
  • ExpEnv: d4rl dataset

• Double Check Your State Before Trusting It: Confidence-Aware Bidirectional Offline Model-Based Imagination

  • Jiafei Lyu, Xiu Li, Zongqing Lu
  • Key: double check mechanism, bidirectional modeling, offline RL
  • OpenReview: 7, 6, 6
  • ExpEnv: d4rl dataset

• Model-Based Opponent Modeling

  • XiaoPeng Yu, Jiechuan Jiang, Wanpeng Zhang, Haobin Jiang, Zongqing Lu
  • Key: multi-agent, model-based
  • OpenReview: 7, 6, 4, 3
  • ExpEnv: mpe, google research football

• Mingling Foresight with Imagination: Model-Based Cooperative Multi-Agent Reinforcement Learning

  • Zhiwei Xu, Dapeng Li, Bin Zhang, Yuan Zhan, Yunpeng Bai, Guoliang Fan
  • Key: multi-agent, model-based
  • OpenReview: 6, 5
  • ExpEnv: StarCraft II, Google Research Football, Multi-Agent Discrete MuJoCo

• MoCoDA: Model-based Counterfactual Data Augmentation

  • Silviu Pitis, Elliot Creager, Ajay Mandlekar, Animesh Garg
  • Key: data augmentation framework, offline RL
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: 2D Navigation, Hook-Sweep

• When to Update Your Model: Constrained Model-based Reinforcement Learning

  • Tianying Ji, Yu Luo, Fuchun Sun, Mingxuan Jing, Fengxiang He, Wenbing Huang
  • Key: event-triggered mechanism, constrained model-shift lower-bound optimization
  • OpenReview: 6, 6, 5, 5
  • ExpEnv: mujoco

• Model-based Safe Deep Reinforcement Learning via a Constrained Proximal Policy Optimization Algorithm

  • Ashish Jayant, Shalabh Bhatnagar
  • Key: constrained RL, model-based
  • OpenReview: 7, 6, 5, 5
  • ExpEnv: safety gym

• Learning to Attack Federated Learning: A Model-based Reinforcement Learning Attack Framework

  • Henger Li, Xiaolin Sun, Zizhan Zheng
  • Key: attack & defense, federated learning, model-based
  • OpenReview: 6, 6, 6, 5
  • ExpEnv: MNIST, FashionMNIST, EMNIST, CIFAR-10 and synthetic dataset

• Model-Based Imitation Learning for Urban Driving

  • Anthony Hu, Gianluca Corrado, Nicolas Griffiths, Zachary Murez, Corina Gurau, Hudson Yeo, Alex Kendall, Roberto Cipolla, Jamie Shotton
  • Key: model-based, imitation learning, autonomous driving
  • OpenReview: 7, 6, 6
  • ExpEnv: CARLA

• Data-Driven Model-Based Optimization via Invariant Representation Learning

  • Han Qi, Yi Su, Aviral Kumar, Sergey Levine
  • Key: domain adaptation, invariant objective models, representation learning (no about model-based RL)
  • OpenReview: 7, 6, 6, 5, 5
  • ExpEnv: design-bench

• Model-based Lifelong Reinforcement Learning with Bayesian Exploration

  • Haotian Fu, Shangqun Yu, Michael Littman, George Konidaris
  • Key: lifelong RL, variational bayesian
  • OpenReview: 7, 6, 6
  • ExpEnv: mujoco, meta-world

• Plan To Predict: Learning an Uncertainty-Foreseeing Model For Model-Based Reinforcement Learning

  • Zifan Wu, Chao Yu, Chen Chen, Jianye Hao, Hankz Hankui Zhuo
  • Key: treat the model rollout process as a sequential decision making problem
  • OpenReview: 7, 7, 6, 6
  • ExpEnv: mujoco, d4rl

• Joint Model-Policy Optimization of a Lower Bound for Model-Based RL

  • Benjamin Eysenbach, Alexander Khazatsky, Sergey Levine, Russ Salakhutdinov
  • Key: unified objective for model-based RL
  • OpenReview: 8, 8, 7, 6
  • ExpEnv: gridworld, mujoco, ROBEL manipulation

• RAMBO-RL: Robust Adversarial Model-Based Offline Reinforcement Learning

  • Marc Rigter, Bruno Lacerda, Nick Hawes
  • Key: offline rl, model-based rl, two-player game, adversarial model training
  • OpenReview: 6, 6, 6, 4
  • ExpEnv: d4rl

• Conservative Dual Policy Optimization for Efficient Model-Based Reinforcement Learning

  • Shenao Zhang
  • Key: posterior sampling RL, referential update, constrained conservative update
  • OpenReview: 7, 7, 5, 5
  • ExpEnv: mujoco, N-Chain MDPs

• Bayesian Optimistic Optimization: Optimistic Exploration for Model-based Reinforcement Learning

  • Chenyang Wu, Tianci Li, Zongzhang Zhang, Yang Yu
  • Key: optimism in the face of uncertainty(OFU), BOO Regret
  • OpenReview: 6, 6, 5
  • ExpEnv: RiverSwim, Chain, Random MDPs

• Model-based RL with Optimistic Posterior Sampling: Structural Conditions and Sample Complexity

  • Alekh Agarwal, Tong Zhang
  • Key: posterior sampling RL, Bellman error decoupling framework
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: None

• Exponential Family Model-Based Reinforcement Learning via Score Matching

  • Gene Li, Junbo Li, Nathan Srebro, Zhaoran Wang, Zhuoran Yang
  • Key: optimistic model-based, score matching
  • OpenReview: 7, 7, 6
  • ExpEnv: None

• Deep Hierarchical Planning from Pixels

  • Danijar Hafner, Kuang-Huei Lee, Ian Fischer, Pieter Abbeel
  • Key: hierarchical RL, long-horizon and sparse reward tasks
  • OpenReview: 6, 6, 5
  • ExpEnv: atari, deepmind control suite, deepmind lab, crafter

• Continuous MDP Homomorphisms and Homomorphic Policy Gradient

  • Sahand Rezaei-Shoshtari, Rosie Zhao, Prakash Panangaden, David Meger, Doina Precup
  • Key: Homomorphic Policy Gradient, Continuous MDP Homomorphisms, Lax Bisimulation Loss
  • OpenReview: 7, 7, 7
  • ExpEnv: deepmind control suite

ICML 2022

Toggle

• DreamerPro: Reconstruction-Free Model-Based Reinforcement Learning with Prototypical Representations

  • Fei Deng, Ingook Jang, Sungjin Ahn
  • Key: dreamer, prototypes
  • ExpEnv: deepmind control suite

• Denoised MDPs: Learning World Models Better Than the World Itself

  • Tongzhou Wang, Simon Du, Antonio Torralba, Phillip Isola, Amy Zhang, Yuandong Tian
  • Key: representation learning, denoised model
  • ExpEnv: deepmind control suite, RoboDesk

• Model-based Meta Reinforcement Learning using Graph Structured Surrogate Models and Amortized Policy Search

  • Qi Wang, Herke van Hoof
  • Key: graph structured surrogate model, meta training
  • ExpEnv: atari, mujoco

• Towards Adaptive Model-Based Reinforcement Learning

  • Yi Wan, Ali Rahimi-Kalahroudi, Janarthanan Rajendran, Ida Momennejad, Sarath Chandar, Harm van Seijen
  • Key: local change adaptation
  • ExpEnv: GridWorldLoCA, ReacherLoCA, MountaincarLoCA

• Efficient Model-based Multi-agent Reinforcement Learning via Optimistic Equilibrium Computation

  • Pier Giuseppe Sessa, Maryam Kamgarpour, Andreas Krause
  • Key: model-based multi-agent, confidence bound
  • ExpEnv: SMART

• Regularizing a Model-based Policy Stationary Distribution to Stabilize Offline Reinforcement Learning

  • Shentao Yang, Yihao Feng, Shujian Zhang, Mingyuan Zhou
  • Key: offline rl, model-based rl, stationary distribution regularization
  • ExpEnv: d4rl

• Design-Bench: Benchmarks for Data-Driven Offline Model-Based Optimization

  • Brandon Trabucco, Xinyang Geng, Aviral Kumar, Sergey Levine
  • Key: benchmark, offline MBO
  • ExpEnv: Design-Bench Benchmark Tasks

• Temporal Difference Learning for Model Predictive Control

  • Nicklas Hansen, Hao Su, Xiaolong Wang
  • Key: td-learning, MPC
  • ExpEnv: deepmind control suite, Meta-World

ICLR 2022

Toggle

• Revisiting Design Choices in Offline Model Based Reinforcement Learning

  • Cong Lu, Philip Ball, Jack Parker-Holder, Michael Osborne, Stephen J. Roberts
  • Key: model-based offline, uncertainty quantification
  • OpenReview: 8, 8, 6, 6, 6
  • ExpEnv: d4rl dataset

• Value Gradient weighted Model-Based Reinforcement Learning

  • Claas A Voelcker, Victor Liao, Animesh Garg, Amir-massoud Farahmand
  • Key: Value-Gradient weighted Model loss
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: mujoco

• Planning in Stochastic Environments with a Learned Model

  • Ioannis Antonoglou, Julian Schrittwieser, Sherjil Ozair, Thomas K Hubert, David Silver
  • Key: MCTS, stochastic MuZero
  • OpenReview: 10, 8, 8, 5
  • ExpEnv: 2048 game, Backgammon, Go

• Policy improvement by planning with Gumbel

  • Ivo Danihelka, Arthur Guez, Julian Schrittwieser, David Silver
  • Key: Gumbel AlphaZero, Gumbel MuZero
  • OpenReview: 8, 8, 8, 6
  • ExpEnv: go, chess, atari

• Model-Based Offline Meta-Reinforcement Learning with Regularization

  • Sen Lin, Jialin Wan, Tengyu Xu, Yingbin Liang, Junshan Zhang
  • Key: model-based offline Meta-RL
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: d4rl dataset

• On-Policy Model Errors in Reinforcement Learning

  • Lukas Froehlich, Maksym Lefarov, Melanie Zeilinger, Felix Berkenkamp
  • Key: model errors, on-policy corrections
  • OpenReview: 8, 6, 6, 5
  • ExpEnv: mujoco, pybullet

• A Relational Intervention Approach for Unsupervised Dynamics Generalization in Model-Based Reinforcement Learning

  • Jiaxian Guo, Mingming Gong, Dacheng Tao
  • Key: relational intervention, dynamics generalization
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: Pendulum, mujoco

• Information Prioritization through Empowerment in Visual Model-based RL

  • Homanga Bharadhwaj, Mohammad Babaeizadeh, Dumitru Erhan, Sergey Levine
  • Key: mutual information, visual model-based RL
  • OpenReview: 8, 8, 8, 6
  • ExpEnv: deepmind control suite, Kinetics dataset

• Transfer RL across Observation Feature Spaces via Model-Based Regularization

  • Yanchao Sun, Ruijie Zheng, Xiyao Wang, Andrew E Cohen, Furong Huang
  • Key: latent dynamics model, transfer RL
  • OpenReview: 8, 6, 5, 5
  • ExpEnv: CartPole, Acrobot and Cheetah-Run, mujoco, 3DBall

• Learning State Representations via Retracing in Reinforcement Learning

  • Changmin Yu, Dong Li, Jianye HAO, Jun Wang, Neil Burgess
  • Key: representation learning, learning via retracing
  • OpenReview: 8, 6, 5, 3
  • ExpEnv: deepmind control suite

• Model-augmented Prioritized Experience Replay

  • Youngmin Oh, Jinwoo Shin, Eunho Yang, Sung Ju Hwang
  • Key: prioritized experience replay, mbrl
  • OpenReview: 8, 8, 6, 5
  • ExpEnv: pybullet

• Evaluating Model-Based Planning and Planner Amortization for Continuous Control

  • Arunkumar Byravan, Leonard Hasenclever, Piotr Trochim, Mehdi Mirza, Alessandro Davide Ialongo, Yuval Tassa, Jost Tobias Springenberg, Abbas Abdolmaleki, Nicolas Heess, Josh Merel, Martin Riedmiller
  • Key: model predictive control
  • OpenReview: 8, 6, 6, 6
  • ExpEnv: mujoco

• Gradient Information Matters in Policy Optimization by Back-propagating through Model

  • Chongchong Li, Yue Wang, Wei Chen, Yuting Liu, Zhi-Ming Ma, Tie-Yan Liu
  • Key: two-model-based method, analyze model error and policy gradient
  • OpenReview: 8, 8, 6, 6
  • ExpEnv: mujoco

• Pareto Policy Pool for Model-based Offline Reinforcement Learning

  • Yijun Yang, Jing Jiang, Tianyi Zhou, Jie Ma, Yuhui Shi
  • Key: model-based offline, model return-uncertainty trade-off
  • OpenReview: 8, 8, 6, 5
  • ExpEnv: d4rl dataset

• Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage

  • Masatoshi Uehara, Wen Sun
  • Key: model-based offline theory, PAC bounds
  • OpenReview: 8, 6, 6, 5
  • ExpEnv: None

• Know Thyself: Transferable Visual Control Policies Through Robot-Awareness

  • Edward S. Hu, Kun Huang, Oleh Rybkin, Dinesh Jayaraman
  • Key: world models that transfer to new robots
  • OpenReview: 8, 6, 6, 5
  • ExpEnv: mujoco, WidowX and Franka Panda robot

NeurIPS 2021

Toggle

• On Effective Scheduling of Model-based Reinforcement Learning

  • Hang Lai, Jian Shen, Weinan Zhang, Yimin Huang, Xing Zhang, Ruiming Tang, Yong Yu, Zhenguo Li
  • Key: extension of mbpo, hyper-controller learning
  • OpenReview: 8, 6, 6
  • ExpEnv: mujoco, pybullet

• COMBO: Conservative Offline Model-Based Policy Optimization

  • Tianhe Yu, Aviral Kumar, Rafael Rafailov, Aravind Rajeswaran, Sergey Levine, Chelsea Finn
  • Key: offline reinforcement learning, model-based reinforcement learning, deep reinforcement learning
  • OpenReview: 6, 7, 6, 8
  • ExpEnv: d4rl dataset

• Safe Reinforcement Learning by Imagining the Near Future

  • Garrett Thomas, Yuping Luo, Tengyu Ma
  • Key: safe rl, reward penalty, theory about model-based rollouts
  • OpenReview: 8, 6, 6
  • ExpEnv: mujoco

• Model-Based Reinforcement Learning via Imagination with Derived Memory

  • Yao Mu, Yuzheng Zhuang, Bin Wang, Guangxiang Zhu, Wulong Liu, Jianyu Chen, Ping Luo, Shengbo Eben Li, Chongjie Zhang, Jianye HAO
  • Key: extension of dreamer, prediction-reliability weight
  • OpenReview: 6, 6, 6, 6
  • ExpEnv: deepmind control suite

• MobILE: Model-Based Imitation Learning From Observation Alone

  • Rahul Kidambi, Jonathan Chang, Wen Sun
  • Key: imitation learning from observations alone, mbrl
  • OpenReview: 6, 6, 6, 4
  • ExpEnv: cartpole, mujoco

• Model-Based Episodic Memory Induces Dynamic Hybrid Controls

  • Hung Le, Thommen Karimpanal George, Majid Abdolshah, Truyen Tran, Svetha Venkatesh
  • Key: model-based, episodic control
  • OpenReview: 7, 7, 6, 6
  • ExpEnv: 2D maze navigation, cartpole, mountainCar and lunarlander, atari, 3D navigation: gym-miniworld

• A Consciousness-Inspired Planning Agent for Model-Based Reinforcement Learning

  • Mingde Zhao, Zhen Liu, Sitao Luan, Shuyuan Zhang, Doina Precup, Yoshua Bengio
  • Key: mbrl, set representation
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: MiniGrid-BabyAI framework

• Mastering Atari Games with Limited Data

  • Weirui Ye, Shaohuai Liu, Thanard Kurutach, Pieter Abbeel, Yang Gao
  • Key: muzero, self-supervised consistency loss
  • OpenReview: 7, 7, 7, 5
  • ExpEnv: atrai 100k, deepmind control suite

• Online and Offline Reinforcement Learning by Planning with a Learned Model

  • Julian Schrittwieser, Thomas K Hubert, Amol Mandhane, Mohammadamin Barekatain, Ioannis Antonoglou, David Silver
  • Key: muzero, reanalyse, offline
  • OpenReview: 8, 8, 7, 6
  • ExpEnv: atrai dataset, deepmind control suite dataset

• Self-Consistent Models and Values

  • Gregory Farquhar, Kate Baumli, Zita Marinho, Angelos Filos, Matteo Hessel, Hado van Hasselt, David Silver
  • Key: new model learning way
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: tabular MDP, Sokoban, atari

• Proper Value Equivalence

  • Christopher Grimm, Andre Barreto, Gregory Farquhar, David Silver, Satinder Singh
  • Key: value equivalence, value-based planning, muzero
  • OpenReview: 8, 7, 7, 6
  • ExpEnv: four rooms, atari

• MOPO: Model-based Offline Policy Optimization

  • Tianhe Yu, Garrett Thomas, Lantao Yu, Stefano Ermon, James Zou, Sergey Levine, Chelsea Finn, Tengyu Ma
  • Key: model-based, offline
  • OpenReview: None
  • ExpEnv: d4rl dataset, halfcheetah-jump and ant-angle

• RoMA: Robust Model Adaptation for Offline Model-based Optimization

  • Sihyun Yu, Sungsoo Ahn, Le Song, Jinwoo Shin
  • Key: model-based, offline
  • OpenReview: 7, 6, 6
  • ExpEnv: design-bench

• Offline Reinforcement Learning with Reverse Model-based Imagination

  • Jianhao Wang, Wenzhe Li, Haozhe Jiang, Guangxiang Zhu, Siyuan Li, Chongjie Zhang
  • Key: model-based, offline
  • OpenReview: 7, 6, 6, 5
  • ExpEnv: d4rl dataset

• Offline Model-based Adaptable Policy Learning

  • Xiong-Hui Chen, Yang Yu, Qingyang Li, Fan-Ming Luo, Zhiwei Tony Qin, Shang Wenjie, Jieping Ye
  • Key: model-based, offline
  • OpenReview: 6, 6, 6, 4
  • ExpEnv: d4rl dataset

• Weighted model estimation for offline model-based reinforcement learning

  • Toru Hishinuma, Kei Senda
  • Key: model-based, offline, off-policy evaluation
  • OpenReview: 7, 6, 6, 6
  • ExpEnv: pendulum, d4rl dataset

• Reward-Free Model-Based Reinforcement Learning with Linear Function Approximation

  • Weitong Zhang, Dongruo Zhou, Quanquan Gu
  • Key: learning theory, model-based reward-free RL, linear function approximation
  • OpenReview: 6, 6, 5, 5
  • ExpEnv: None

• Provable Model-based Nonlinear Bandit and Reinforcement Learning: Shelve Optimism, Embrace Virtual Curvature

  • Kefan Dong, Jiaqi Yang, Tengyu Ma
  • Key: learning theory, model-based bandit RL, nonlinear function approximation
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: None

• Discovering and Achieving Goals via World Models

  • Russell Mendonca, Oleh Rybkin, Kostas Daniilidis, Danijar Hafner, Deepak Pathak
  • Key: unsupervised goal reaching, goal-conditioned RL
  • OpenReview: 6, 6, 6, 6, 6
  • ExpEnv: walker, quadruped, bins, kitchen

ICLR 2021

Toggle

• Deployment-Efficient Reinforcement Learning via Model-Based Offline Optimization

  • Tatsuya Matsushima, Hiroki Furuta, Yutaka Matsuo, Ofir Nachum, Shixiang Gu
  • Key: model-based, behavior cloning (warmup), trpo
  • OpenReview: 8, 7, 7, 5
  • ExpEnv: d4rl dataset

• Control-Aware Representations for Model-based Reinforcement Learning

  • Brandon Cui, Yinlam Chow, Mohammad Ghavamzadeh
  • Key: representation learning, model-based soft actor-critic
  • OpenReview: 6, 6, 6
  • ExpEnv: planar system, inverted pendulum – swingup, cartpole, 3-link manipulator — swingUp & balance

• Mastering Atari with Discrete World Models

  • Danijar Hafner, Timothy Lillicrap, Mohammad Norouzi, Jimmy Ba
  • Key: DreamerV2, many tricks(multiple categorical variables, KL balancing, etc)
  • OpenReview: 9, 8, 5, 4
  • ExpEnv: atari

• Model-Based Visual Planning with Self-Supervised Functional Distances

  • Stephen Tian, Suraj Nair, Frederik Ebert, Sudeep Dasari, Benjamin Eysenbach, Chelsea Finn, Sergey Levine
  • Key: goal-reaching task, dynamics learning, distance learning (goal-conditioned Q-function)
  • OpenReview: 7, 7, 7, 7
  • ExpEnv: sawyer, door sliding

• Model-Based Offline Planning

  • Arthur Argenson, Gabriel Dulac-Arnold
  • Key: model-based, offline
  • OpenReview: 8, 7, 5, 5
  • ExpEnv: RL Unplugged(RLU), d4rl dataset

• Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation

  • Justin Fu, Sergey Levine
  • Key: model-based, offline
  • OpenReview: 8, 6, 6
  • ExpEnv: design-bench

• On the role of planning in model-based deep reinforcement learning

  • Jessica B. Hamrick, Abram L. Friesen, Feryal Behbahani, Arthur Guez, Fabio Viola, Sims Witherspoon, Thomas Anthony, Lars Buesing, Petar Veličković, Théophane Weber
  • Key: discussion about planning in MuZero
  • OpenReview: 7, 7, 6, 5
  • ExpEnv: atari, go, deepmind control suite

• Representation Balancing Offline Model-based Reinforcement Learning

  • Byung-Jun Lee, Jongmin Lee, Kee-Eung Kim
  • Key: Representation Balancing MDP, model-based, offline
  • OpenReview: 7, 7, 7, 6
  • ExpEnv: d4rl dataset

• Model-based micro-data reinforcement learning: what are the crucial model properties and which model to choose?

  • Balázs Kégl, Gabriel Hurtado, Albert Thomas
  • Key: mixture density nets, heteroscedasticity
  • OpenReview: 7, 7, 7, 6, 5
  • ExpEnv: acrobot system

ICML 2021

Toggle

• Conservative Objective Models for Effective Offline Model-Based Optimization

  • Brandon Trabucco, Aviral Kumar, Xinyang Geng, Sergey Levine
  • Key: conservative objective model, offline mbrl
  • ExpEnv: design-bench

• Continuous-Time Model-Based Reinforcement Learning

  • Çağatay Yıldız, Markus Heinonen, Harri Lähdesmäki
  • Key: continuous-time
  • ExpEnv: pendulum, cartPole and acrobot

• Model-Based Reinforcement Learning via Latent-Space Collocation

  • Oleh Rybkin, Chuning Zhu, Anusha Nagabandi, Kostas Daniilidis, Igor Mordatch, Sergey Levine
  • Key: latent space collocation
  • ExpEnv: sparse metaworld tasks

• Model-Free and Model-Based Policy Evaluation when Causality is Uncertain

  • David A Bruns-Smith
  • Key: worst-case bounds
  • ExpEnv: ope-tools

• Muesli: Combining Improvements in Policy Optimization

  • Matteo Hessel, Ivo Danihelka, Fabio Viola, Arthur Guez, Simon Schmitt, Laurent Sifre, Theophane Weber, David Silver, Hado van Hasselt
  • Key: value equivalence
  • ExpEnv: atari

• Vector Quantized Models for Planning

  • Sherjil Ozair, Yazhe Li, Ali Razavi, Ioannis Antonoglou, Aäron van den Oord, Oriol Vinyals
  • Key: VQVAE, MCTS
  • ExpEnv: chess datasets, DeepMind Lab

• PC-MLP: Model-based Reinforcement Learning with Policy Cover Guided Exploration

  • Yuda Song, Wen Sun
  • Key: sample complexity, kernelized nonlinear regulators, linear MDPs
  • ExpEnv: mountain car, antmaze, mujoco

• Temporal Predictive Coding For Model-Based Planning In Latent Space

  • Tung Nguyen, Rui Shu, Tuan Pham, Hung Bui, Stefano Ermon
  • Key: temporal predictive coding with a RSSM, latent space
  • ExpEnv: deepmind control suite

• Model-based Reinforcement Learning for Continuous Control with Posterior Sampling

  • Ying Fan, Yifei Ming
  • Key: regret bound of psrl, mpc
  • ExpEnv: continuous cartpole, pendulum swingup, mujoco

• A Sharp Analysis of Model-based Reinforcement Learning with Self-Play

  • Qinghua Liu, Tiancheng Yu, Yu Bai, Chi Jin
  • Key: learning theory, multi-agent, model-based self play, two-player zero-sum Markov games
  • ExpEnv: None

Other

• Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Driving

  • Yuqi Wang, Jiawei He, Lue Fan, Hongxin Li, Yuntao Chen, Zhaoxiang Zhang CVPR 2024
  • Key: AutoDrive world modeling
  • ExpEnv: nuScenes

• DriveWorld: 4D Pre-trained Scene Understanding via World Models for Autonomous Driving

  • Chen Min, Dawei Zhao, Liang Xiao, Jian Zhao, Xinli Xu, Zheng Zhu, Lei Jin, Jianshu Li, Yulan Guo, Junliang Xing, Liping Jing, Yiming Nie, Bin Dai CVPR 2024
  • Key: AutoDrive world modeling
  • ExpEnv: nuScenes, OpenScene

• Masked Trajectory Models for Prediction, Representation, and Control

  • Philipp Wu, Arjun Majumdar, Kevin Stone, Yixin Lin, Igor Mordatch, Pieter Abbeel, Aravind Rajeswaran ICLR 2023 Workshop RRL
  • Key: offline RL, learning for control, sequence modeling
  • ExpEnv: d4rl

• World Models via Policy-Guided Trajectory Diffusion

  • Marc Rigter, Jun Yamada, Ingmar Posner Arxiv 2023
  • Key: Diffusion model, world model
  • ExpEnv: deepmind control suite, gridworld

• Model-Based Epistemic Variance of Values for Risk-Aware Policy Optimization

  • Carlos E. Luis, Alessandro G. Bottero, Julia Vinogradska, Felix Berkenkamp, Jan Peters Arxiv 2023
  • Key: cumulative rewards uncertainty estimation in MBRL
  • ExpEnv: mujoco

• Sample-Efficient Learning to Solve a Real-World Labyrinth Game Using Data-Augmented Model-Based Reinforcement Learning

  • Thomas Bi, Raffaello D'Andrea. Arxiv 2023
  • Key: Data-Augmented, DreamerV3
  • ExpEnv: Real-World Labyrinth Game

• Mastering Diverse Domains through World Models

  • Danijar Hafner, Jurgis Pasukonis, Jimmy Ba, Timothy Lillicrap. Arxiv 2023
  • Key: DreamerV3, scaling property to world model
  • ExpEnv: deepmind control suite, atari, DMLab, minecraft

• Theoretically Guaranteed Policy Improvement Distilled from Model-Based Planning

  • Chuming Li, Ruonan Jia, Jiawei Yao, Jie Liu, Yinmin Zhang, Yazhe Niu, Yaodong Yang, Yu Liu, Wanli Ouyang. IJCAI Workshop 2023
  • Key: extended policy improvement, model regularization, planning theorem
  • ExpEnv: mujoco

Tutorial

• [Video] Csaba Szepesvári - The challenges of model-based reinforcement learning and how to overcome them
• [Blog] Model-Based Reinforcement Learning: Theory and Practice

Codebase

• mbrl-lib - Meta: Library for Model Based RL
• DI-engine - OpenDILab: Decision AI Engine

Contributing

Our purpose is to make this repo even better. If you are interested in contributing, please refer to HERE for instructions in contribution.

License

Awesome Model-Based RL is released under the Apache 2.0 license.

(Back to top)