Poster
Off-Policy Deep Reinforcement Learning without Exploration
Scott Fujimoto · David Meger · Doina Precup

Tue Jun 11th 06:30 -- 09:00 PM @ Pacific Ballroom #38

Many practical applications of reinforcement learning constrain agents to learn from a fixed batch of data which has already been gathered, without offering further possibility for data collection. In this paper, we demonstrate that due to errors introduced by extrapolation, standard off-policy deep reinforcement learning algorithms, such as DQN and DDPG, are incapable of learning with data uncorrelated to the distribution under the current policy, making them ineffective for this fixed batch setting. We introduce a novel class of off-policy algorithms, batch-constrained reinforcement learning, which restricts the action space in order to force the agent towards behaving close to on-policy with respect to a subset of the given data. We present the first continuous control deep reinforcement learning algorithm which can learn effectively from arbitrary, fixed batch data, and empirically demonstrate the quality of its behavior in several tasks.

Author Information

Scott Fujimoto (McGill University)
David Meger (McGill University)
Doina Precup (McGill University / DeepMind)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors