Current browse context:
cs.LG
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: Data-efficient Hindsight Off-policy Option Learning
(Submitted on 30 Jul 2020 (v1), last revised 15 Jun 2021 (this version, v2))
Abstract: We introduce Hindsight Off-policy Options (HO2), a data-efficient option learning algorithm. Given any trajectory, HO2 infers likely option choices and backpropagates through the dynamic programming inference procedure to robustly train all policy components off-policy and end-to-end. The approach outperforms existing option learning methods on common benchmarks. To better understand the option framework and disentangle benefits from both temporal and action abstraction, we evaluate ablations with flat policies and mixture policies with comparable optimization. The results highlight the importance of both types of abstraction as well as off-policy training and trust-region constraints, particularly in challenging, simulated 3D robot manipulation tasks from raw pixel inputs. Finally, we intuitively adapt the inference step to investigate the effect of increased temporal abstraction on training with pre-trained options and from scratch.
Submission history
From: Markus Wulfmeier [view email][v1] Thu, 30 Jul 2020 16:52:33 GMT (11942kb,D)
[v2] Tue, 15 Jun 2021 15:55:50 GMT (21297kb,D)
Link back to: arXiv, form interface, contact.