Offline Reinforcement Learning with Implicit Q-Learning

10/12/2021
by   Ilya Kostrikov, et al.
0

Offline reinforcement learning requires reconciling two conflicting aims: learning a policy that improves over the behavior policy that collected the dataset, while at the same time minimizing the deviation from the behavior policy so as to avoid errors due to distributional shift. This trade-off is critical, because most current offline reinforcement learning methods need to query the value of unseen actions during training to improve the policy, and therefore need to either constrain these actions to be in-distribution, or else regularize their values. We propose an offline RL method that never needs to evaluate actions outside of the dataset, but still enables the learned policy to improve substantially over the best behavior in the data through generalization. The main insight in our work is that, instead of evaluating unseen actions from the latest policy, we can approximate the policy improvement step implicitly by treating the state value function as a random variable, with randomness determined by the action (while still integrating over the dynamics to avoid excessive optimism), and then taking a state conditional upper expectile of this random variable to estimate the value of the best actions in that state. This leverages the generalization capacity of the function approximator to estimate the value of the best available action at a given state without ever directly querying a Q-function with this unseen action. Our algorithm alternates between fitting this upper expectile value function and backing it up into a Q-function. Then, we extract the policy via advantage-weighted behavioral cloning. We dub our method implicit Q-learning (IQL). IQL demonstrates the state-of-the-art performance on D4RL, a standard benchmark for offline reinforcement learning. We also demonstrate that IQL achieves strong performance fine-tuning using online interaction after offline initialization.

READ FULL TEXT
research
06/09/2022

Mildly Conservative Q-Learning for Offline Reinforcement Learning

Offline reinforcement learning (RL) defines the task of learning from a ...
research
03/28/2023

Offline RL with No OOD Actions: In-Sample Learning via Implicit Value Regularization

Most offline reinforcement learning (RL) methods suffer from the trade-o...
research
06/06/2023

Boosting Offline Reinforcement Learning with Action Preference Query

Training practical agents usually involve offline and online reinforceme...
research
07/25/2023

Offline Reinforcement Learning with On-Policy Q-Function Regularization

The core challenge of offline reinforcement learning (RL) is dealing wit...
research
07/12/2023

Budgeting Counterfactual for Offline RL

The main challenge of offline reinforcement learning, where data is limi...
research
09/29/2022

Offline Reinforcement Learning via High-Fidelity Generative Behavior Modeling

In offline reinforcement learning, weighted regression is a common metho...
research
07/01/2022

Offline Policy Optimization with Eligible Actions

Offline policy optimization could have a large impact on many real-world...

Please sign up or login with your details

Forgot password? Click here to reset