Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models | HTML5

Model-based reinforcement learning (RL) algorithms can attain excellent sample efficiency, but often lag behind the best model-free algorithms in terms of asymptotic performance. This is especially true with high-capac…

Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models Kurtland Chua Roberto Calandra Rowan McAllister Sergey Levine Berkeley Artificial Intelligence Research University of California, Berkeley {kchua, roberto.calandra, rmcallister, svlevine}@berkeley.edu Abstract Model-based reinforcement learning (RL) algorithms can attain excellent sample efficiency, but often lag behind the best model-free algorithms in terms of asymptotic performance. This is especially true with high-capacity parametric function approximators, such as deep networks. In this paper, we study

related reading

• [1709.06560] Deep Reinforcement Learning that Mattersarxiv.org
• A Free Lunch from the Noise:Provable and Practical Exploration for Representation Learningarxiv.org
• A Gallery of Methods Beyond RL — Part I: Sampling Methodsshengyu-feng.github.io
• Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Reviewarxiv.org
• c21f4ce780c5c9d774f79841b81fdc6d-Paper.pdfproceedings.neurips.cc
• [2606.05555] Representation Learning Enables Scalable Multitask Deep Reinforcement Learningarxiv.org
• [2103.06326] S4RL: Surprisingly Simple Self-Supervision for Offline Reinforcement Learningarxiv.org
• [1710.10044] Distributional Reinforcement Learning with Quantile Regressionarxiv.org
• Exploration for the Efficient Deployment of Reinforcement Learning Agentsopenreview.net
• f514cec81cb148559cf475e7426eed5e-Paper.pdfproceedings.neurips.cc
• [2602.11399] Can We Really Learn One Representation to Optimize All Rewards?arxiv.org
• Robust Regression for Safe Exploration in Controlarxiv.org