Institute Homepage
Institute Homepage DE Sign In

Back

Autonomous Learning Conference Paper 2019

Control What You Can: Intrinsically Motivated Task-Planning Agent

Thumb ticker sm img 20210812 091658 cropped
Robust Machine Learning
Sebastian Blaes
  • Postdoctoral Researcher
no image
Autonomous Learning
Marin Vlastelica Pogancic
Thumb ticker sm aaeaaqaaaaaaaadsaaaajdzkmdeyzti4ltezmzutnddhni1injiwltdlndninzc2zwvhmq
Empirical Inference
Jia-Jie Zhu
Thumb ticker sm georg 2018 crop small
Empirical Inference, Autonomous Learning
Georg Martius
Senior Research Scientist

We present a novel intrinsically motivated agent that learns how to control the environment in the fastest possible manner by optimizing learning progress. It learns what can be controlled, how to allocate time and attention, and the relations between objects using surprise based motivation. The effectiveness of our method is demonstrated in a synthetic as well as a robotic manipulation environment yielding considerably improved performance and smaller sample complexity. In a nutshell, our work combines several task-level planning agent structures (backtracking search on task graph, probabilistic road-maps, allocation of search efforts) with intrinsic motivation to achieve learning from scratch.

Author(s): Sebastian Blaes and Marin Vlastelica and Jia-Jie Zhu and Georg Martius
Book Title: Advances in Neural Information Processing Systems (NeurIPS 2019)
Pages: 12520--12531
Year: 2019
Month: December
Publisher: Curran Associates, Inc.
Project(s):
Bibtex Type: Conference Paper (inproceedings)
Event Name: 33rd Annual Conference on Neural Information Processing Systems
State: Accepted
Electronic Archiving: grant_archive
Links:
Attachments:

BibTex 

@inproceedings{BlaesVlastelicaZhuMartius2019:CWYC,
  title = {Control {W}hat {Y}ou {C}an: {I}ntrinsically Motivated Task-Planning Agent},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2019)},
  abstract = {We present a novel intrinsically motivated agent that learns how to control the environment in the fastest possible manner by optimizing learning progress. It learns what can be controlled, how to allocate time and attention, and the relations between objects using surprise based motivation. The effectiveness of our method is demonstrated in a synthetic as well as a robotic manipulation environment yielding considerably improved performance and smaller sample complexity. In a nutshell, our work combines several task-level planning agent structures (backtracking search on task graph, probabilistic road-maps, allocation of search efforts) with intrinsic motivation to achieve learning from scratch. },
  pages = {12520--12531},
  publisher = {Curran Associates, Inc.},
  month = dec,
  year = {2019},
  slug = {blaesvlastelicazhumartius2019-cwyc},
  author = {Blaes, Sebastian and Vlastelica, Marin and Zhu, Jia-Jie and Martius, Georg},
  month_numeric = {12}
}