Deep learning models are often inefficient and expensive to train, often requiring tens or hundreds of thousands of training images labelled by an expert. It is impossible to have hand labelled images for each abnormal image. A domain shift paradigm refers to a novel paradigm of learning, specifically the recognition of new concepts from only a simple description. It is expected that a classifier trained on only the training-class examples will work for test data comprised of unseen classes, by exploiting the side information regarding the semantic relationship between training and unseen classes.
The project will investigate effective domain shift techniques as fundamental research and their applications in real world problems. Regarding the applications to ultrasound image analysis, a real scenario based on ultrasound video streams will be selected to be used as a test bed, which will be completed in collaboration with hospital radiologists.
A first degree in a relevant scientific discipline, such as computer science, engineering, mathematics, physics, or medicine. Desirable skills include mathematics, statistics, machine learning, computer vision, and software engineering.
English language requirement
IELTS score must be at least 6.0. Other equivalent qualifications will be accepted. Full details of the University’s policy are available online.
Edinburgh Napier University is committed to promoting equality and diversity in our staff and student community https://www.napier.ac.uk/about-us/university-governance/equality-and-diversity-information