Large-scale Annotation of Biomedical data and Expert Label Synthesis
摘要截稿:
全文截稿: 2018-06-01
开会时间: 2018-09-16
会议难度:
CCF分类: 无
会议地点: Granada, Spain
Overview
With the widespread use of data-intensive supervised machine learning methods within the medical image computing community, a growing pressure has mounted to generate vast quantities of quality annotations. Unsurprisingly, in response to the need for very large volumes of training data for deep learning systems, the demand for new methodology to gather vast amounts of annotations in efficient, coherent and safe ways has only surged.
In this context, the proposed third edition of the LABELS workshop (https://www.miccailabels.org) focuses on bringing together researchers in the field to discuss and principles of training data acquisition and the careful design of labelling procedures. A second goal is to promote the development and scientific exchange of algorithms that focus on assisting the annotation process by making them, for example, more general, more accurate, faster or more intuitive for the medical experts. To this end, we propose a diverse program including keynote talks from world-renowned experts and paper submissions addressing the labelling/annotation task by means of methods from domains such as:
– Active learning
– Semi-supervised learning
– Reinforcement learning
– Domain adaptation and transfer learning
– Crowdsourcing
– Learning from multiple annotation modalities
– Deep learning architectures
– Fusion of labels from different sources
– Data augmentation
– Modelling of label uncertainty
– Visualization and human-computer interaction for annotation generation
– Verification and validation of annotations