Agnostic System Identification for Model-Based Reinforcement Learning

Abstract

A fundamental problem in control is to learn a model of a system from observations that is useful for controller synthesis. To provide good performance guarantees, existing methods must assume that the real system is in the class of models considered during learning. We present an iterative method with strong guarantees even in the agnostic case where the system is not in the class. In particular, we show that any no-regret online learning algorithm can be used to obtain a near-optimal policy, provided some model achieves low training error and access to a good exploration distribution. Our approach applies to both discrete and continuous domains. We demonstrate its efficacy and scalability on a challenging helicopter domain from the literature.

8 pages, published in ICML 2012

Cited by in corpus (22)

• MOReL : Model-Based Offline Reinforcement Learning
• EPOpt: Learning Robust Neural Network Policies Using Model Ensembles
• Plan Online, Learn Offline: Efficient Learning and Exploration via Model-Based Control
• A Game Theoretic Framework for Model Based Reinforcement Learning
• COMBO: Conservative Offline Model-Based Policy Optimization
• Preparing for the Unknown: Learning a Universal Policy with Online System Identification
• What are the Statistical Limits of Offline RL with Linear Function Approximation?
• Information Theoretic Regret Bounds for Online Nonlinear Control
• The Effect of Planning Shape on Dyna-style Planning in High-dimensional State Spaces
• Learning Policies for Contextual Submodular Prediction
• Neuroprosthetic decoder training as imitation learning
• Gradient-Aware Model-based Policy Search
• Off-Dynamics Reinforcement Learning: Training for Transfer with Domain Classifiers
• Sample Complexity of Sparse System Identification Problem
• Online Learning with Continuous Variations: Dynamic Regret and Reductions
• Importance Sampling Policy Evaluation with an Estimated Behavior Policy
• Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage
• Learning Active Task-Oriented Exploration Policies for Bridging the Sim-to-Real Gap
• Few Shot System Identification for Reinforcement Learning
• Offline Reinforcement Learning: Fundamental Barriers for Value Function Approximation
• Learning the Reward Function for a Misspecified Model
• Sample Complexity of Estimating the Policy Gradient for Nearly Deterministic Dynamical Systems