WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION
Interpretable and explainable machine learning have proven to be promising approaches to verify the quality of a data-driven model in general as well as to obtain more information about the quality of certain observations in practise. In this paper, we use these approaches for an application in the...
Published in: | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
---|---|
Main Authors: | , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Copernicus Publications
2020
|
Subjects: | |
Online Access: | https://doi.org/10.5194/isprs-annals-V-2-2020-1005-2020 https://doaj.org/article/b5a7a66e69eb4f7f90581f0a726b1903 |
id |
ftdoajarticles:oai:doaj.org/article:b5a7a66e69eb4f7f90581f0a726b1903 |
---|---|
record_format |
openpolar |
spelling |
ftdoajarticles:oai:doaj.org/article:b5a7a66e69eb4f7f90581f0a726b1903 2023-05-15T16:35:58+02:00 WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION J. Kierdorf J. Garcke J. Behley T. Cheeseman R. Roscher 2020-08-01T00:00:00Z https://doi.org/10.5194/isprs-annals-V-2-2020-1005-2020 https://doaj.org/article/b5a7a66e69eb4f7f90581f0a726b1903 EN eng Copernicus Publications https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/V-2-2020/1005/2020/isprs-annals-V-2-2020-1005-2020.pdf https://doaj.org/toc/2194-9042 https://doaj.org/toc/2194-9050 doi:10.5194/isprs-annals-V-2-2020-1005-2020 2194-9042 2194-9050 https://doaj.org/article/b5a7a66e69eb4f7f90581f0a726b1903 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol V-2-2020, Pp 1005-1012 (2020) Technology T Engineering (General). Civil engineering (General) TA1-2040 Applied optics. Photonics TA1501-1820 article 2020 ftdoajarticles https://doi.org/10.5194/isprs-annals-V-2-2020-1005-2020 2022-12-31T12:12:14Z Interpretable and explainable machine learning have proven to be promising approaches to verify the quality of a data-driven model in general as well as to obtain more information about the quality of certain observations in practise. In this paper, we use these approaches for an application in the marine sciences to support the monitoring of whales. Whale population monitoring is an important element of whale conservation, where the identification of whales plays an important role in this process, for example to trace the migration of whales over time and space. Classical approaches use photographs and a manual mapping with special focus on the shape of the whale flukes and their unique pigmentation. However, this is not feasible for comprehensive monitoring. Machine learning methods, especially deep neural networks, have shown that they can efficiently solve the automatic observation of a large number of whales. Despite their success for many different tasks such as identification, further potentials such as interpretability and their benefits have not yet been exploited. Our main contribution is an analysis of interpretation tools, especially occlusion sensitivity maps, and the question of how the gained insights can help a whale researcher. For our analysis, we use images of humpback whale flukes provided by the Kaggle Challenge ”Humpback Whale Identification”. By means of spectral cluster analysis of heatmaps, which indicate which parts of the image are important for a decision, we can show that the they can be grouped in a meaningful way. Moreover, it appears that characteristics automatically determined by a neural network correspond to those that are considered important by a whale expert. Article in Journal/Newspaper Humpback Whale Directory of Open Access Journals: DOAJ Articles ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences V-2-2020 1005 1012 |
institution |
Open Polar |
collection |
Directory of Open Access Journals: DOAJ Articles |
op_collection_id |
ftdoajarticles |
language |
English |
topic |
Technology T Engineering (General). Civil engineering (General) TA1-2040 Applied optics. Photonics TA1501-1820 |
spellingShingle |
Technology T Engineering (General). Civil engineering (General) TA1-2040 Applied optics. Photonics TA1501-1820 J. Kierdorf J. Garcke J. Behley T. Cheeseman R. Roscher WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
topic_facet |
Technology T Engineering (General). Civil engineering (General) TA1-2040 Applied optics. Photonics TA1501-1820 |
description |
Interpretable and explainable machine learning have proven to be promising approaches to verify the quality of a data-driven model in general as well as to obtain more information about the quality of certain observations in practise. In this paper, we use these approaches for an application in the marine sciences to support the monitoring of whales. Whale population monitoring is an important element of whale conservation, where the identification of whales plays an important role in this process, for example to trace the migration of whales over time and space. Classical approaches use photographs and a manual mapping with special focus on the shape of the whale flukes and their unique pigmentation. However, this is not feasible for comprehensive monitoring. Machine learning methods, especially deep neural networks, have shown that they can efficiently solve the automatic observation of a large number of whales. Despite their success for many different tasks such as identification, further potentials such as interpretability and their benefits have not yet been exploited. Our main contribution is an analysis of interpretation tools, especially occlusion sensitivity maps, and the question of how the gained insights can help a whale researcher. For our analysis, we use images of humpback whale flukes provided by the Kaggle Challenge ”Humpback Whale Identification”. By means of spectral cluster analysis of heatmaps, which indicate which parts of the image are important for a decision, we can show that the they can be grouped in a meaningful way. Moreover, it appears that characteristics automatically determined by a neural network correspond to those that are considered important by a whale expert. |
format |
Article in Journal/Newspaper |
author |
J. Kierdorf J. Garcke J. Behley T. Cheeseman R. Roscher |
author_facet |
J. Kierdorf J. Garcke J. Behley T. Cheeseman R. Roscher |
author_sort |
J. Kierdorf |
title |
WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
title_short |
WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
title_full |
WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
title_fullStr |
WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
title_full_unstemmed |
WHAT IDENTIFIES A WHALE BY ITS FLUKE? ON THE BENEFIT OF INTERPRETABLE MACHINE LEARNING FOR WHALE IDENTIFICATION |
title_sort |
what identifies a whale by its fluke? on the benefit of interpretable machine learning for whale identification |
publisher |
Copernicus Publications |
publishDate |
2020 |
url |
https://doi.org/10.5194/isprs-annals-V-2-2020-1005-2020 https://doaj.org/article/b5a7a66e69eb4f7f90581f0a726b1903 |
genre |
Humpback Whale |
genre_facet |
Humpback Whale |
op_source |
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol V-2-2020, Pp 1005-1012 (2020) |
op_relation |
https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/V-2-2020/1005/2020/isprs-annals-V-2-2020-1005-2020.pdf https://doaj.org/toc/2194-9042 https://doaj.org/toc/2194-9050 doi:10.5194/isprs-annals-V-2-2020-1005-2020 2194-9042 2194-9050 https://doaj.org/article/b5a7a66e69eb4f7f90581f0a726b1903 |
op_doi |
https://doi.org/10.5194/isprs-annals-V-2-2020-1005-2020 |
container_title |
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
container_volume |
V-2-2020 |
container_start_page |
1005 |
op_container_end_page |
1012 |
_version_ |
1766026285048397824 |