Autoregressive Dynamics Models for Offline Policy Evaluation and Optimization

by   Michael R. Zhang, et al.

Standard dynamics models for continuous control make use of feedforward computation to predict the conditional distribution of next state and reward given current state and action using a multivariate Gaussian with a diagonal covariance structure. This modeling choice assumes that different dimensions of the next state and reward are conditionally independent given the current state and action and may be driven by the fact that fully observable physics-based simulation environments entail deterministic transition dynamics. In this paper, we challenge this conditional independence assumption and propose a family of expressive autoregressive dynamics models that generate different dimensions of the next state and reward sequentially conditioned on previous dimensions. We demonstrate that autoregressive dynamics models indeed outperform standard feedforward models in log-likelihood on heldout transitions. Furthermore, we compare different model-based and model-free off-policy evaluation (OPE) methods on RL Unplugged, a suite of offline MuJoCo datasets, and find that autoregressive dynamics models consistently outperform all baselines, achieving a new state-of-the-art. Finally, we show that autoregressive dynamics models are useful for offline policy optimization by serving as a way to enrich the replay buffer through data augmentation and improving performance using model-based planning.


page 1

page 2

page 3

page 4


DARA: Dynamics-Aware Reward Augmentation in Offline Reinforcement Learning

Offline reinforcement learning algorithms promise to be applicable in se...

Unifying Model-Based and Neural Network Feedforward: Physics-Guided Neural Networks with Linear Autoregressive Dynamics

Unknown nonlinear dynamics often limit the tracking performance of feedf...

Parallel and Flexible Sampling from Autoregressive Models via Langevin Dynamics

This paper introduces an alternative approach to sampling from autoregre...

Reliable Offline Model-based Optimization for Industrial Process Control

In the research area of offline model-based optimization, novel and prom...

DROMO: Distributionally Robust Offline Model-based Policy Optimization

We consider the problem of offline reinforcement learning with model-bas...

Revisit Policy Optimization in Matrix Form

In tabular case, when the reward and environment dynamics are known, pol...

Autoregressive Text Generation Beyond Feedback Loops

Autoregressive state transitions, where predictions are conditioned on p...