Learning What To Do by Simulating the Past

by   David Lindner, et al.

Since reward functions are hard to specify, recent work has focused on learning policies from human feedback. However, such approaches are impeded by the expense of acquiring such feedback. Recent work proposed that agents have access to a source of information that is effectively free: in any environment that humans have acted in, the state will already be optimized for human preferences, and thus an agent can extract information about what humans want from the state. Such learning is possible in principle, but requires simulating all possible past trajectories that could have led to the observed state. This is feasible in gridworlds, but how do we scale it to complex tasks? In this work, we show that by combining a learned feature encoder with learned inverse models, we can enable agents to simulate human actions backwards in time to infer what they must have done. The resulting algorithm is able to reproduce a specific skill in MuJoCo environments given a single state sampled from the optimal policy for that skill.


page 8

page 19

page 20

page 21

page 22

page 23

page 24


PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training

Conveying complex objectives to reinforcement learning (RL) agents can o...

Guided Policy Search for Parameterized Skills using Adverbs

We present a method for using adverb phrases to adjust skill parameters ...

Deep Reinforcement Learning from Policy-Dependent Human Feedback

To widen their accessibility and increase their utility, intelligent age...

Preferences Implicit in the State of the World

Reinforcement learning (RL) agents optimize only the features specified ...

Safe Deep RL in 3D Environments using Human Feedback

Agents should avoid unsafe behaviour during both training and deployment...

Mutual Reinforcement Learning

Recently, collaborative robots have begun to train humans to achieve com...

Learning from Physical Human Feedback: An Object-Centric One-Shot Adaptation Method

For robots to be effectively deployed in novel environments and tasks, t...