dc.contributor.advisor | Jenssen, Robert | |
dc.contributor.author | Choi, Changkyu | |
dc.date.accessioned | 2023-05-26T08:32:51Z | |
dc.date.available | 2023-05-26T08:32:51Z | |
dc.date.issued | 2023-06-09 | |
dc.description.abstract | Marine environment monitoring has become increasingly significant due to the excessive exploitation of oceans, which detrimentally impacts ecosystems. Deep learning provides an effective monitoring approach by automating the analysis of vast amounts of observed image data, enabling stakeholders to make informed decisions regarding fishing quotas or conservation efforts.
The success of deep learning is often attributed to its capacity to extract relevant features from data, without the need for handcrafted rules or heuristics. However, this capability is not without limitations, as the intricate feature extraction process of deep learning-based systems poses fundamental challenges.
A lack of annotated data presents an inherent challenge for deep learning. The widespread success of deep learning has primarily relied on the ample availability of annotated data, while deep learning models encounter difficulties when learning from limited annotations. However, obtaining annotated data is expensive, particularly in the context of marine environment monitoring, as it is often a manual process demanding the expertise of domain specialists.
Another challenge of deep learning is a lack of explainability. The black-box nature of deep learning models can make it difficult to understand how they arrive at their decisions. This hinders their adoption in critical decision-making processes, as stakeholders may be hesitant to trust models whose decision-making rationale is not transparent or interpretable.
To address the challenges and further advance deep learning methodologies, this thesis proposes three novel deep learning methods, highlighting marine environment monitoring as an application domain. The dependence on annotated data is addressed through two novel semi-supervised methods demonstrated in different image tasks. The central operational logic in both methods entails alternating between supervised learning and unsupervised deep clustering within a single network, merging data structure uncovered through unsupervised clustering with a small amount of ground-truth class information.
Both methods employ multi-frequency echosounder data to demonstrate their effectiveness in marine environment monitoring, outperforming conventional approaches.
Moreover, a new explainable deep learning method is proposed to address the lack of explainability. This method generates explanations for its decisions while adhering to user-centered explanation principles, such as minimality, sufficiency, and interactivity. The information-bottleneck framework provides a theoretical ground to pursue minimality and sufficiency, while interactivity is accomplished by integrating additional domain knowledge into the training process, enabling the generated explanations to evolve accordingly. The method is validated using a variety of marine image datasets, encompassing multi-frequency echosounder data and seal pup images on sea ice.
While the monitoring of marine environments is a significant focus, the primary aim of the thesis is to contribute to the advancements of deep learning methodologies. As such, the proposed methods are designed to be generic and possess the potential for broader applicability across various domains. We believe that the methods presented in this thesis hold the promise of fostering a more effective, user-centered, and transparent approach to deep learning, as well as facilitating our efforts to preserve the marine environment and promote sustainable ocean stewardship. | en_US |
dc.description.doctoraltype | ph.d. | en_US |
dc.description.popularabstract | Marine environments face threats, and monitoring them is vital. Deep learning can analyze image data but grapples with challenges like requiring annotated data and limited explainability. This thesis enhances deep learning for marine monitoring by addressing these issues.
A novel semi-supervised learning algorithm is introduced to tackle annotated data dependence, generalizing to diverse image tasks. These methods outperform conventional approaches while significantly reducing annotated data use. Furthermore, a new self-explainable deep learning method addresses explainability limitations, aiming to generate user-centered explanations aligned with various domain knowledge.
The thesis initiates an effort to bridge deep learning and marine environment monitoring, emphasizing the importance of collaboration between experts in both fields for marine environment preservation. | en_US |
dc.description.sponsorship | This work was supported in part by the Ubiquitous Cognitive Computer Vision for Marine Services (COGMAR) under Research Council of Norway (RCN) Grant 270966, and in part by the Centre for Research-based Innovation Visual Intelligence under RCN Grant 309439. | en_US |
dc.identifier.isbn | 978-82-8236-522-2 trykk | |
dc.identifier.isbn | 978-82-8236-523-9 - pdf | |
dc.identifier.uri | https://hdl.handle.net/10037/29267 | |
dc.language.iso | eng | en_US |
dc.publisher | UiT Norges arktiske universitet | en_US |
dc.publisher | UiT The Arctic University of Norway | en_US |
dc.relation.haspart | <p>Paper I: Choi, C., Kampffmeyer, M., Handegard, N.O., Salberg, A.B., Brautaset, O., Eikvil, L. & Jenssen, R. (2021). Semisupervised Target Classification in Multi-frequency Echosounder Data. <i>ICES Journal of Marine Science, 78</i>(7), 2615–2627. Also available in Munin at <a href=https://hdl.handle.net/10037/22715>https://hdl.handle.net/10037/22715</a>.
<p>Paper II: Choi, C., Kampffmeyer, M., Handegard, N.O., Salberg, A.B. & Jenssen, R. (2023). Deep Semi-supervised Semantic Segmentation in Multi-frequency Echosounder Data. <i>IEEE Journal of Oceanic Engineering, 48</i>(2), 384 - 400. Also available in Munin at <a href=https://hdl.handle.net/10037/29001>https://hdl.handle.net/10037/29001</a>.
<p>Paper III: Choi, C., Yu, S., Kampffmeyer, M., Handegard, N.O., Salberg, A.B. & Jenssen, R. Deep Deterministic Information-Bottleneck Explainability on Marine Image Data. (Submitted manuscript). | en_US |
dc.relation.isbasedon | Multi-frequency echosounder data: Johnsen, E., Rieucau, G., Ona, E. & Skaret, G. (2017). Collective structures anchor massive schools of lesser sandeel to the seabed, increasing vulnerability to fishery. <i>Marine Ecology Progress Series, 573</i>, 29-236, available at <a href=https://doi.org/10.3354/meps12156> https://doi.org/10.3354/meps12156</a>. | en_US |
dc.relation.isbasedon | Areal images of seal pups on sea ice: Biuw, M., Øigård, T.A., Nilssen, K.T., Stenson, G., Lindblom, L., Poltermann, M., Kristiansen, M. & Haug, T. (2022). Recent Harp and Hooded Seal Pup Production Estimates in the Greenland Sea Suggest Ecology-Driven Declines. <i>NAMMCO Scientific Publications, 12</i>, available at <a href=https://doi.org/10.7557/3.5821>https://doi.org/10.7557/3.5821</a>. | en_US |
dc.rights.accessRights | openAccess | en_US |
dc.rights.holder | Copyright 2023 The Author(s) | |
dc.subject.courseID | DOKTOR-004 | |
dc.subject | VDP::Mathematics and natural science: 400::Information and communication science: 420::Algorithms and computability theory: 422 | en_US |
dc.subject | VDP::Matematikk og Naturvitenskap: 400::Informasjons- og kommunikasjonsvitenskap: 420::Algoritmer og beregnbarhetsteori: 422 | en_US |
dc.subject | VDP::Technology: 500::Marine technology: 580::Other marine technology: 589 | en_US |
dc.subject | VDP::Teknologi: 500::Marin teknologi: 580::Annen marin teknologi: 589 | en_US |
dc.subject | VDP::Mathematics and natural science: 400::Information and communication science: 420::Mathematical modeling and numerical methods: 427 | en_US |
dc.subject | VDP::Matematikk og Naturvitenskap: 400::Informasjons- og kommunikasjonsvitenskap: 420::Matematisk modellering og numeriske metoder: 427 | en_US |
dc.subject | VDP::Mathematics and natural science: 400::Mathematics: 410::Applied mathematics: 413 | en_US |
dc.subject | VDP::Matematikk og Naturvitenskap: 400::Matematikk: 410::Anvendt matematikk: 413 | en_US |
dc.subject | VDP::Mathematics and natural science: 400::Information and communication science: 420::Simulation, visualization, signal processing, image processing: 429 | en_US |
dc.subject | VDP::Matematikk og Naturvitenskap: 400::Informasjons- og kommunikasjonsvitenskap: 420::Simulering, visualisering, signalbehandling, bildeanalyse: 429 | en_US |
dc.title | Advancing Deep Learning for Marine Environment Monitoring | en_US |
dc.type | Doctoral thesis | en_US |
dc.type | Doktorgradsavhandling | en_US |