TU Darmstadt / ULB / TUbiblio

Learning Coupled Forward-Inverse Models with Combined Prediction Errors

Koert, Dorothea ; Maeda, Guilherme ; Neumann, Gerhard ; Peters, Jan (2022)
Learning Coupled Forward-Inverse Models with Combined Prediction Errors.
International Conference on Robotics and Automation (ICRA) 2018. Brisbane, QLD, Australia (21.-25.05.2018)
doi: 10.26083/tuprints-00020546
Conference or Workshop Item, Secondary publication, Postprint

Abstract

Challenging tasks in unstructured environments require robots to learn complex models. Given a large amount of information, learning multiple simple models can offer an efficient alternative to a monolithic complex network. Training multiple models-that is, learning their parameters and their responsibilities-has been shown to be prohibitively hard as optimization is prone to local minima. To efficiently learn multiple models for different contexts, we thus develop a new algorithm based on expectation maximization (EM). In contrast to comparable concepts, this algorithm trains multiple modules of paired forward-inverse models by using the prediction errors of both forward and inverse models simultaneously. In particular, we show that our method yields a substantial improvement over only considering the errors of the forward models on tasks where the inverse space contains multiple solutions.

Item Type: Conference or Workshop Item
Erschienen: 2022
Creators: Koert, Dorothea ; Maeda, Guilherme ; Neumann, Gerhard ; Peters, Jan
Type of entry: Secondary publication
Title: Learning Coupled Forward-Inverse Models with Combined Prediction Errors
Language: English
Date: 2022
Place of Publication: Darmstadt
Publisher: IEEE
Book Title: 2018 IEEE International Conference on Robotics and Automation (ICRA)
Collation: 7 Seiten
Event Title: International Conference on Robotics and Automation (ICRA) 2018
Event Location: Brisbane, QLD, Australia
Event Dates: 21.-25.05.2018
DOI: 10.26083/tuprints-00020546
URL / URN: https://tuprints.ulb.tu-darmstadt.de/20546
Corresponding Links:
Origin: Secondary publication service
Abstract:

Challenging tasks in unstructured environments require robots to learn complex models. Given a large amount of information, learning multiple simple models can offer an efficient alternative to a monolithic complex network. Training multiple models-that is, learning their parameters and their responsibilities-has been shown to be prohibitively hard as optimization is prone to local minima. To efficiently learn multiple models for different contexts, we thus develop a new algorithm based on expectation maximization (EM). In contrast to comparable concepts, this algorithm trains multiple modules of paired forward-inverse models by using the prediction errors of both forward and inverse models simultaneously. In particular, we show that our method yields a substantial improvement over only considering the errors of the forward models on tasks where the inverse space contains multiple solutions.

Status: Postprint
URN: urn:nbn:de:tuda-tuprints-205461
Classification DDC: 000 Generalities, computers, information > 004 Computer science
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Intelligent Autonomous Systems
TU-Projects: EC/H2020|640554|SKILLS4ROBOTS
Date Deposited: 18 Nov 2022 14:02
Last Modified: 21 Nov 2022 10:47
PPN:
Corresponding Links:
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details