Producción CyT
Domain Adaptation and Representation Transfer
Libro
Autoría:
Lisa Koch ; M. Jorge Cardoso ; FERRANTE, ENZO ; Konstantinos Kamnitsas ; Mobarakol Islam ; Meirui Jiang ; Nicola Rieke ; Sotirios A. Tsaftaris ; Dong YangFecha:
2023Editorial y Lugar de Edición:
Springer NatureISBN:
978-3-031-45857-6Resumen *
Recent breakthroughs in advanced machine learning and deep learning have revolution- ized computer vision and medical imaging, enabling unparalleled accuracy in tasks such as image segmentation, object recognition, disease detection, and image registration. Although these developments have greatly benefited the MICCAI community, many models suffer from limited adaptability when faced with novel scenarios or heteroge- neous input data. To overcome this restriction, researchers have explored techniques such as transfer learning, representation learning, and domain adaptation, allowing for improved model training, effective domain adaptation, and the application of knowledge learned from one domain to tackle challenges in other domains. By expanding the ver- satility and robustness of these cutting-edge methods, researchers hope to increase their clinical utility and broaden their impact across various medical imaging applications.The 5th MICCAI Workshop on Domain Adaptation and Representation Trans- fer (DART 2023), which happened in conjunction with MICCAI 2023 in Vancouver, Canada, aimed at creating a discussion forum to compare, evaluate, and discuss method- ological advancements and ideas that can improve the applicability of machine learning (ML)/deep learning (DL) approaches to clinical settings by making them robust and consistent across different domains.During the fifth edition of DART, 16 full papers were accepted out of 32 submis- sions that underwent a rigorous double-blind peer review process. At least three expert reviewers evaluated each paper, ensuring that only high-quality works were accepted. To maintain objectivity and prevent any biases, reviewer assignments were automated and took into account potential conflicts of interest and recent collaboration history between reviewers and authors. The final decision on acceptance or rejection was made by the area chairs based on the reviews received, and these decisions were always fair and unappealable.Additionally, the workshop organization committee granted the Best Paper Award to the best submission presented at DART 2023. The Best Paper Award was assigned as a result of a secret voting procedure where each member of the committee indicated two papers worthy of consideration for the award. The paper collecting the majority of votes was then chosen by the committee.We believe that the paper selection process implemented during DART 2023, as well as the quality of the submissions, resulted in scientifically validated and interesting contributions for the MICCAI community and, in particular, for researchers working on domain adaptation and representation transfer. Información suministrada por el agente en SIGEVAPalabras Clave
machine learningdomain adaptationdomain shiftartificial intelligence