Goal-Conditioned Generators of Deep Policies
-
Published:2023-06-26
Issue:6
Volume:37
Page:7503-7511
-
ISSN:2374-3468
-
Container-title:Proceedings of the AAAI Conference on Artificial Intelligence
-
language:
-
Short-container-title:AAAI
Author:
Faccio Francesco,Herrmann Vincent,Ramesh Aditya,Kirsch Louis,Schmidhuber Jürgen
Abstract
Goal-conditioned Reinforcement Learning (RL) aims at learning optimal policies, given goals encoded in special command inputs. Here we study goal-conditioned neural nets (NNs) that learn to generate deep NN policies in form of context-specific weight matrices, similar to Fast Weight Programmers and other methods from the 1990s. Using context commands of the form ``generate a policy that achieves a desired expected return,'' our NN generators combine powerful exploration of parameter space with generalization across commands to iteratively find better and better policies. A form of weight-sharing HyperNetworks and policy embeddings scales our method to generate deep NNs. Experiments show how a single learned policy generator can produce policies that achieve any return seen during training. Finally, we evaluate our algorithm on a set of continuous control tasks where it exhibits competitive performance.
Our code is public.
Publisher
Association for the Advancement of Artificial Intelligence (AAAI)
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献