Learning by Doing : Controlling a Dynamical System using Causality, Control, and Reinforcement Learning
(2022) 35th Conference on Neural Information Processing Systems, NeurIPS 2021 176. p.246-258- Abstract
Questions in causality, control, and reinforcement learning go beyond the classical machine learning task of prediction under i.i.d. observations. Instead, these fields consider the problem of learning how to actively perturb a system to achieve a certain effect on a response variable. Arguably, they have complementary views on the problem: In control, one usually aims to first identify the system by excitation strategies to then apply model-based design techniques to control the system. In (non-model-based) reinforcement learning, one directly optimizes a reward. In causality, one focus is on identifiability of causal structure. We believe that combining the different views might create synergies and this competition is meant as a... (More)
Questions in causality, control, and reinforcement learning go beyond the classical machine learning task of prediction under i.i.d. observations. Instead, these fields consider the problem of learning how to actively perturb a system to achieve a certain effect on a response variable. Arguably, they have complementary views on the problem: In control, one usually aims to first identify the system by excitation strategies to then apply model-based design techniques to control the system. In (non-model-based) reinforcement learning, one directly optimizes a reward. In causality, one focus is on identifiability of causal structure. We believe that combining the different views might create synergies and this competition is meant as a first step toward such synergies. The participants had access to observational and (offline) interventional data generated by dynamical systems. Track CHEM considers an open-loop problem in which a single impulse at the beginning of the dynamics can be set, while Track ROBO considers a closed-loop problem in which control variables can be set at each time step. The goal in both tracks is to infer controls that drive the system to a desired state. Code is open-sourced (github.com/LearningByDoingCompetition/learningbydoingcomp) to reproduce the winning solutions of the competition and to facilitate trying out new methods on the competition tasks.
(Less)
- author
- Weichwald, Sebastian ; Mogensen, Søren Wengel LU ; Lee, Tabitha Edith ; Baumann, Dominik ; Kroemer, Oliver ; Guyon, Isabelle ; Trimpe, Sebastian ; Peters, Jonas and Pfister, Niklas
- organization
- publishing date
- 2022
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Causality, chemical reactions, control, dynamical systems, reinforcement learning, robotics, system identification
- host publication
- Proceedings of Machine Learning Research
- volume
- 176
- pages
- 13 pages
- publisher
- ML Research Press
- conference name
- 35th Conference on Neural Information Processing Systems, NeurIPS 2021
- conference location
- Virtual, Online
- conference dates
- 2021-12-06 - 2021-12-14
- external identifiers
-
- scopus:85163821753
- language
- English
- LU publication?
- yes
- id
- 20ceb65b-a41c-46af-80bf-edb5eb19bc8c
- date added to LUP
- 2023-10-30 11:17:16
- date last changed
- 2023-10-30 11:17:16
@inproceedings{20ceb65b-a41c-46af-80bf-edb5eb19bc8c,
abstract = {{<p>Questions in causality, control, and reinforcement learning go beyond the classical machine learning task of prediction under i.i.d. observations. Instead, these fields consider the problem of learning how to actively perturb a system to achieve a certain effect on a response variable. Arguably, they have complementary views on the problem: In control, one usually aims to first identify the system by excitation strategies to then apply model-based design techniques to control the system. In (non-model-based) reinforcement learning, one directly optimizes a reward. In causality, one focus is on identifiability of causal structure. We believe that combining the different views might create synergies and this competition is meant as a first step toward such synergies. The participants had access to observational and (offline) interventional data generated by dynamical systems. Track CHEM considers an open-loop problem in which a single impulse at the beginning of the dynamics can be set, while Track ROBO considers a closed-loop problem in which control variables can be set at each time step. The goal in both tracks is to infer controls that drive the system to a desired state. Code is open-sourced (github.com/LearningByDoingCompetition/learningbydoingcomp) to reproduce the winning solutions of the competition and to facilitate trying out new methods on the competition tasks.</p>}},
author = {{Weichwald, Sebastian and Mogensen, Søren Wengel and Lee, Tabitha Edith and Baumann, Dominik and Kroemer, Oliver and Guyon, Isabelle and Trimpe, Sebastian and Peters, Jonas and Pfister, Niklas}},
booktitle = {{Proceedings of Machine Learning Research}},
keywords = {{Causality; chemical reactions; control; dynamical systems; reinforcement learning; robotics; system identification}},
language = {{eng}},
pages = {{246--258}},
publisher = {{ML Research Press}},
title = {{Learning by Doing : Controlling a Dynamical System using Causality, Control, and Reinforcement Learning}},
volume = {{176}},
year = {{2022}},
}