Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization

Deep neural networks (DNNs) can be useful within the marine robotics field, but their utility value is restricted by their black-box nature. Explainable artificial intelligence methods attempt to understand how such black-boxes make their decisions. In this work, linear model trees (LMTs) are used t...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Vilde B. Gjærum, Inga Strümke, Ole Andreas Alsos, Anastasios M. Lekkas
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	deep reinforcement learning autonomous surface vessel explainable artificial intelligence linear model trees Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581
Acceso en línea:	https://doaj.org/article/1cef3c3316ff413abd15bb5f091ca66f
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:1cef3c3316ff413abd15bb5f091ca66f
record_format	dspace
spelling	oai:doaj.org-article:1cef3c3316ff413abd15bb5f091ca66f2021-11-25T18:03:56ZExplaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization10.3390/jmse91111782077-1312https://doaj.org/article/1cef3c3316ff413abd15bb5f091ca66f2021-10-01T00:00:00Zhttps://www.mdpi.com/2077-1312/9/11/1178https://doaj.org/toc/2077-1312Deep neural networks (DNNs) can be useful within the marine robotics field, but their utility value is restricted by their black-box nature. Explainable artificial intelligence methods attempt to understand how such black-boxes make their decisions. In this work, linear model trees (LMTs) are used to approximate the DNN controlling an autonomous surface vessel (ASV) in a simulated environment and then run in parallel with the DNN to give explanations in the form of feature attributions in real-time. How well a model can be understood depends not only on the explanation itself, but also on how well it is presented and adapted to the receiver of said explanation. Different end-users may need both different types of explanations, as well as different representations of these. The main contributions of this work are (1) significantly improving both the accuracy and the build time of a greedy approach for building LMTs by introducing ordering of features in the splitting of the tree, (2) giving an overview of the characteristics of the seafarer/operator and the developer as two different end-users of the agent and receiver of the explanations, and (3) suggesting a visualization of the docking agent, the environment, and the feature attributions given by the LMT for when the developer is the end-user of the system, and another visualization for when the seafarer or operator is the end-user, based on their different characteristics.Vilde B. GjærumInga StrümkeOle Andreas AlsosAnastasios M. LekkasMDPI AGarticledeep reinforcement learningautonomous surface vesselexplainable artificial intelligencelinear model treesNaval architecture. Shipbuilding. Marine engineeringVM1-989OceanographyGC1-1581ENJournal of Marine Science and Engineering, Vol 9, Iss 1178, p 1178 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	deep reinforcement learning autonomous surface vessel explainable artificial intelligence linear model trees Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581
spellingShingle	deep reinforcement learning autonomous surface vessel explainable artificial intelligence linear model trees Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581 Vilde B. Gjærum Inga Strümke Ole Andreas Alsos Anastasios M. Lekkas Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
description	Deep neural networks (DNNs) can be useful within the marine robotics field, but their utility value is restricted by their black-box nature. Explainable artificial intelligence methods attempt to understand how such black-boxes make their decisions. In this work, linear model trees (LMTs) are used to approximate the DNN controlling an autonomous surface vessel (ASV) in a simulated environment and then run in parallel with the DNN to give explanations in the form of feature attributions in real-time. How well a model can be understood depends not only on the explanation itself, but also on how well it is presented and adapted to the receiver of said explanation. Different end-users may need both different types of explanations, as well as different representations of these. The main contributions of this work are (1) significantly improving both the accuracy and the build time of a greedy approach for building LMTs by introducing ordering of features in the splitting of the tree, (2) giving an overview of the characteristics of the seafarer/operator and the developer as two different end-users of the agent and receiver of the explanations, and (3) suggesting a visualization of the docking agent, the environment, and the feature attributions given by the LMT for when the developer is the end-user of the system, and another visualization for when the seafarer or operator is the end-user, based on their different characteristics.
format	article
author	Vilde B. Gjærum Inga Strümke Ole Andreas Alsos Anastasios M. Lekkas
author_facet	Vilde B. Gjærum Inga Strümke Ole Andreas Alsos Anastasios M. Lekkas
author_sort	Vilde B. Gjærum
title	Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
title_short	Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
title_full	Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
title_fullStr	Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
title_full_unstemmed	Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization
title_sort	explaining a deep reinforcement learning docking agent using linear model trees with user adapted visualization
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/1cef3c3316ff413abd15bb5f091ca66f
work_keys_str_mv	AT vildebgjærum explainingadeepreinforcementlearningdockingagentusinglinearmodeltreeswithuseradaptedvisualization AT ingastrumke explainingadeepreinforcementlearningdockingagentusinglinearmodeltreeswithuseradaptedvisualization AT oleandreasalsos explainingadeepreinforcementlearningdockingagentusinglinearmodeltreeswithuseradaptedvisualization AT anastasiosmlekkas explainingadeepreinforcementlearningdockingagentusinglinearmodeltreeswithuseradaptedvisualization
_version_	1718411666626969600

Explaining a Deep Reinforcement Learning Docking Agent Using Linear Model Trees with User Adapted Visualization

Ejemplares similares