東華大學圖書館 |

Deep Learning Applied to Animal Linguistics = = Deep Learning Mit Anwendung in Der Tierlinguistik.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Deep Learning Applied to Animal Linguistics =/
其他題名:	Deep Learning Mit Anwendung in Der Tierlinguistik.
作者:	Bergler, Christian.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2023,
面頁冊數:	454 p.
附註:	Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
Contained By:	Dissertations Abstracts International85-08B.
標題:	Deep learning. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30790049
ISBN:	9798381454895

Deep Learning Applied to Animal Linguistics = = Deep Learning Mit Anwendung in Der Tierlinguistik.
Bergler, Christian.

Deep Learning Applied to Animal Linguistics =Deep Learning Mit Anwendung in Der Tierlinguistik. - Ann Arbor : ProQuest Dissertations & Theses, 2023 - 454 p.

Source: Dissertations Abstracts International, Volume: 85-08, Section: B.

Thesis (Ph.D.)--Friedrich-Alexander-Universitaet Erlangen-Nuernberg (Germany), 2023.

Even nowadays, people have only a very limited understanding about animal communication. Scientists are still far from identifying statistically relevant, animal-specific, and recurring linguistic paradigms. However, combined with the associated situation-specific behavioral observations, these patterns represent an indispensable basis for decoding animal communication. In order to derive statistically significant communicative and behavioral hypotheses, sufficiently large audiovisual data volumes are essential covering the animalspecific communicative and behavioral repertoire in a representative, natural, and realistic manner. Hence, passive audiovisual monitoring techniques are increasingly deployed to obtain more natural insights, since the recording is performed in an unobtrusive fashion to minimize disruptive factors and simultaneously maximizing the probability of observing the entire inventory of natural communicative and behavioral paradigms in adequate numbers. Nevertheless, time- and human-resource constraints hamper scientists to efficiently process large-scale noise-heavy data archives, incorporating massive amounts of hidden audiovisual information, to derive an overall and bigger picture about animal linguistics. Thus, in order to perform a deep and detailed data analysis to derive real-world representations, the support of machine-based data-driven algorithms is a fundamental prerequisite. In the scope of this doctoral thesis, a hybrid approach between machine (deep) learning and animal bioacoustics is presented, applying a wide variety of different and novel algorithms to analyze large-scale, noise-heavy, audiovisual, and animal-specific data repositories in order to provide completely new insights into the field of animal linguistics. Due to their complex social, communicative, and cognitive abilities, the largest member of the dolphin family - the killer whale (Orcinus orca) - was chosen as target species and prototype for this study. In northern British Columbia one of the largest animal-specific bioacoustic archives - the Orchive- was acquired by the OrcaLab and used as major data foundation, further extended by additional acoustic and behavioral data material, collected during project-internal fieldwork expeditions along the West Coast of Canada in 2017, 2018, 2019, and 2022. A broad spectrum of publicly available deep learning-based algorithms is presented, originally developed on killer whales, but also transferable to other vocalizing animal species, while addressing the following essential acoustic and image-related biological research questions: (1) signal segmentation - robust, efficient, and fully-automated detection of killer whale sound types, (2) sound denoising - signal enhancement of diverse killer whale vocalizations, (3) call type identification - supervised, semi-supervised, and unsupervised deep architectures to recognize vocal killer whale paradigms, (4) sound type separation - signal segregation of overlapping killer whale vocalizations, (5) individual recognition - image-based deep learning framework to identify killer whale individuals, (6) sound source localization - underwater identification of vocalizing killer whale individuals, (7) signal generation - artificial and representative killer whale signal production, and (8) animal independence - adaption and generalization of developed killer whale-related deep learning concepts to other species-specific bioacoustic data volumes. All the inventive and publicly available machine (deep) learning frameworks demonstrate auspicious results and provide totally unprecedented analysis techniques, facilitating more profound interpretations of massive, animal-specific, and audiovisual data volumes, all together building the imperative foundation to significantly push not only the communicative and behavioral understanding of killer whales, but the entire research field of animal bioacoustics and linguistics.

ISBN: 9798381454895Subjects--Topical Terms:

3554982
Deep learning.

Deep Learning Applied to Animal Linguistics = = Deep Learning Mit Anwendung in Der Tierlinguistik.
LDR:09360nmm a2200385 4500 001 2398545
005 20240812064704.5
006 m o d
007 cr#unu||||||||
008 251215s2023 ||||||||||||||||| ||eng d
020 $a 9798381454895
035 $a (MiAaPQ)AAI30790049
035 $a (MiAaPQ)FAUErlangen23926
035 $a AAI30790049
040 $a MiAaPQ $c MiAaPQ
100 1 $a Bergler, Christian. $3 3768460
245 1 0 $a Deep Learning Applied to Animal Linguistics = $b Deep Learning Mit Anwendung in Der Tierlinguistik.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2023
300 $a 454 p.
500 $a Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
500 $a Advisor: Noth, Elmar;Orozco-Arroyave, Juan Rafael.
502 $a Thesis (Ph.D.)--Friedrich-Alexander-Universitaet Erlangen-Nuernberg (Germany), 2023.
520 $a Even nowadays, people have only a very limited understanding about animal communication. Scientists are still far from identifying statistically relevant, animal-specific, and recurring linguistic paradigms. However, combined with the associated situation-specific behavioral observations, these patterns represent an indispensable basis for decoding animal communication. In order to derive statistically significant communicative and behavioral hypotheses, sufficiently large audiovisual data volumes are essential covering the animalspecific communicative and behavioral repertoire in a representative, natural, and realistic manner. Hence, passive audiovisual monitoring techniques are increasingly deployed to obtain more natural insights, since the recording is performed in an unobtrusive fashion to minimize disruptive factors and simultaneously maximizing the probability of observing the entire inventory of natural communicative and behavioral paradigms in adequate numbers. Nevertheless, time- and human-resource constraints hamper scientists to efficiently process large-scale noise-heavy data archives, incorporating massive amounts of hidden audiovisual information, to derive an overall and bigger picture about animal linguistics. Thus, in order to perform a deep and detailed data analysis to derive real-world representations, the support of machine-based data-driven algorithms is a fundamental prerequisite. In the scope of this doctoral thesis, a hybrid approach between machine (deep) learning and animal bioacoustics is presented, applying a wide variety of different and novel algorithms to analyze large-scale, noise-heavy, audiovisual, and animal-specific data repositories in order to provide completely new insights into the field of animal linguistics. Due to their complex social, communicative, and cognitive abilities, the largest member of the dolphin family - the killer whale (Orcinus orca) - was chosen as target species and prototype for this study. In northern British Columbia one of the largest animal-specific bioacoustic archives - the Orchive- was acquired by the OrcaLab and used as major data foundation, further extended by additional acoustic and behavioral data material, collected during project-internal fieldwork expeditions along the West Coast of Canada in 2017, 2018, 2019, and 2022. A broad spectrum of publicly available deep learning-based algorithms is presented, originally developed on killer whales, but also transferable to other vocalizing animal species, while addressing the following essential acoustic and image-related biological research questions: (1) signal segmentation - robust, efficient, and fully-automated detection of killer whale sound types, (2) sound denoising - signal enhancement of diverse killer whale vocalizations, (3) call type identification - supervised, semi-supervised, and unsupervised deep architectures to recognize vocal killer whale paradigms, (4) sound type separation - signal segregation of overlapping killer whale vocalizations, (5) individual recognition - image-based deep learning framework to identify killer whale individuals, (6) sound source localization - underwater identification of vocalizing killer whale individuals, (7) signal generation - artificial and representative killer whale signal production, and (8) animal independence - adaption and generalization of developed killer whale-related deep learning concepts to other species-specific bioacoustic data volumes. All the inventive and publicly available machine (deep) learning frameworks demonstrate auspicious results and provide totally unprecedented analysis techniques, facilitating more profound interpretations of massive, animal-specific, and audiovisual data volumes, all together building the imperative foundation to significantly push not only the communicative and behavioral understanding of killer whales, but the entire research field of animal bioacoustics and linguistics.
520 $a Auch heutzutage hat die Menschheit nur ein sehr begrenztes Verständnis über Tierkommunikation. Wissenschaftler sind noch weit davon entfernt statistisch aussagekräftige, tierartenspezifische und wiederkehrende sprachliche Muster zu identifizieren. Zusammen mit den zugehörigen situationsspezifischen Verhaltensbeobachtungen stellen diese Paradigmen jedoch eine unverzichtbare Grundlage für die Entschlüsselung von Tierkommunikation dar. Um kommunikative und verhaltensbezogene Hypothesen in statistisch signifkanten Mengen abzuleiten, sind ausreichend grose audiovisuelle Datenmengen von essentieller Bedeutung, welche das tierartenspezifische Repertoire aus Kommunikation und Verhalten auf eine möglichst repräsentative, natürliche und realistische Art und Weise abdecken. Passive audiovisuelle Beobachtungstechniken finden zunehmend Einsatz und gewährleisten natürliche Einblicke aufgrund einer dezenten und unauffälligen Datenakquise, bedingt durch die Minimierung von Störfaktoren und einer zeitgleichen Erhöhung der Wahrscheinlichkeit dadurch das gesamte Inventar an Kommunikation und Verhalten in ausreichend grosen Stichproben zu beobachten. Zeit- und Resourcenmängel erschweren es Wissenschaftlern jedoch umfassende und verrauschte Datenarchive mit riesigen Mengen an verborgenen audiovisuellen Informationen effizient zu verarbeiten, um ein umfassendes Bild der Tier-Linguistik ableiten zu können. Die Unterstützung durch computergestützte und datengetriebene Algorithmen ist eine essentielle Vorraussetzung für die Durchführung detaillierter Datenanalysen, um dadurch realitätsnahe Repräsentationen zu gewinnen. Im Rahmen dieser Doktorarbeit wird ein hybrider Ansatz zwischen maschinellem (tiefen) Lernen und der Tier-Bioakustik durch die Anwendung eines breiten Spektrums an unterschiedlichen und neuartigen Algorithmen vorgestellt, um riesige, verrauschte, audiovisuelle, und tierartenbezogene Datensammlungen zu untersuchen, wodurch völlig neue Einblicke in die Tier-Linguistik ermöglicht werden. Aufgrund der komplexen sozialen, kommunikativen und kognitiven Fähigkeiten, wurde das gröste Mitglied der Delphinfamilie - der Killerwal (Orcinus orca) - als Zielspezies und Prototyp ausgewählt. Hauptdatengrundlage ist das im Norden von British Columbia durch das OrcaLab aufgenommene - Orchive- eines der grösten existierenden tierspezifischen bioakustischen Datensammlungen, erweitert durch zusätzliches akustisches und verhaltensrelevantes Datenmaterial aus einer projektinternen Forschungsexpedition in 2017, 2018, 2019, und 2022 entlang der Westküste von Kanada. Es wird ein breites Portfolio an öffentlich zugänglichen Deep Learning-basierten Algorithmen vorgestellt, diese essentiell wichtige akustische und bildbezogene biologische Forschungsfragen zu Killerwalen behandeln, aber auch auf andere kommunikative Tierarten übertragbar sind: (1) Signalsegmentierung - robuste, effiziente und vollautomatisierte Erkennung von Orca-Vokalisation, (2) Soundaufbereitung - entrauschen verschiedener Orca-Vokalisationsmuster, (3) Identifikation von Ruftypen - überwachte, semi-überwachte und unüberwachte Architekturen zur Erkennung von Orca-Kommunikationsmustern, (4) Signalzerlegung - separieren von überlappenden Orca-Signalen, (5) Individuenerkennung - bildbasierte Identifikation von Orca-Einzeltieren, (6) Lokalisation von Soundquellen - Unterwasserortung von vokalisierenden Orca-Individuen, (7) Signalerzeugung - künstliche Erzeugung von Orca-Signalen, und (8) Tierunabhängigkeit - modifizieren und generalisieren entwickelter Orca-spezifischer Deep Learning Konzepte auf Datenmengen anderer Tiergattungen. All die neuartigen und öffentlich verfügbaren maschinellen (tiefen) Lernverfahren zeigen vielversprechende Ergebnisse und liefern völlig ungesehene, sowie zwingend erforderliche, Analyseverfahren, um ausführliche Interpretationen von riesigen, tierspezifischen und audiovisuellen Daten zu ermöglichen, die nicht nur das Verständnis in Bezug auf Kommunikation und Verhalten von Killerwalen, sondern das gesamte Forschungsgebiet der Tier-Bioakustik und Tier-Linguistik, masgeblich vorantreiben.
590 $a School code: 0575.
650 4 $a Deep learning. $3 3554982
650 4 $a Verbal communication. $3 3560678
650 4 $a Whales & whaling. $3 3695308
650 4 $a Funding. $3 3557916
650 4 $a Neural networks. $3 677449
650 4 $a Signal processing. $3 533904
650 4 $a Animal behavior. $3 587929
650 4 $a Linguistics. $3 524476
650 4 $a Pattern recognition. $3 3560648
650 4 $a Behavioral sciences. $3 529833
650 4 $a Biological oceanography. $3 2122748
650 4 $a Communication. $3 524709
650 4 $a Electrical engineering. $3 649834
690 $a 0290
690 $a 0800
690 $a 0602
690 $a 0416
690 $a 0459
690 $a 0544
710 2 $a Friedrich-Alexander-Universitaet Erlangen-Nuernberg (Germany). $3 3564652
773 0 $t Dissertations Abstracts International $g 85-08B.
790 $a 0575
791 $a Ph.D.
792 $a 2023
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30790049