To address biomedical data analysis tasks by learning from noisy, limited, or imprecise expert annotations, researchers have recently started to develop weakly-supervised deep learning (WSDL) techniques, which are of great interest in the field of biomedical engineering. WSDL can not only significantly relieve the human efforts for annotating structured biomedical data (such as signals, images, and videos) but also enable the corresponding deep neural network models to learn from much larger-scale data with a reduced annotation cost.
With the fast growth of advanced deep learning techniques, such as generative adversarial networks (GAN), graph neural networks (GNN), vision transformers (ViT), and deep reinforcement learning (DRL) models, research studies started to focus on solving problems in WSDL and applying WSDL techniques to biomedical analysis tasks.
Topics for this Special Issue include, but are not limited to:
– DRL-based WSDL and its applications to the biomedical field.
– GAN-based WSDL and its applications to the biomedical field.
– GNN-based WSDL and its applications to the biomedical field.
– Graph Convolutional Network-based WSDL and its applications to the biomedical field.
– Methodological studies on WSDL and its applications to the biomedical field.
– Multi-modal WSDL theory and its applications to the biomedical field.
– Multi-task WSDL theory and its applications to the biomedical field.
– Robust WSDL theory and framework and its applications to the biomedical field.
– Spatial/temporal WSDL and its applications to the biomedical field.
– ViT-based WSDL and its applications to the biomedical field.
Guest Editor: Yu-Dong Zhang – yudongzhang@ieee.org
University of Leicester
Submission Information:
January 16, 2023