Menu
Published: 25 May 2022 904 views
In recent years computer vision has made fantastic progress in employing deep learning to create descriptive representations of image data. These methods have enabled accurate predictions (e.g. classification, segmentation etc.) to be made from images, where the learned representations are optimised specifically for these tasks. While effective in many situations, such approaches may struggle to generalise to new examples, particularly where data is drawn from different distributions to the training set. Moreover, learned representations lack both direct interpretability and uncertainty, which limits our ability to understand or resolve reasons for failure.
The aim of this project is to create a machine learning framework that learns human interpretable representations of images, where variations are explicitly decomposed into separate semantically meaningful components such as shape, identify, pose, texture, lighting etc. Important considerations include:devising appropriate model formulations, effectively integrate prior knowledge into the learning process;and exploring how to capture the uncertainty, or ambiguity, of describing an image using these explicit factors.
This research project will employ state of the art deep learning methods to advance our understanding of how interpretable image representations can be learned and used for both discriminative (e.g. classification) and creative (e.g. image editing) tasks. The specific application area for this project is flexible and will depend on the student’s interest. Potential topics include: analysing photographs of human/animal facesor bodies, landscapes and cityscapes,or building models of neurodegenerative disease from brain MRI.
This is a computational project, which would suit a student with good mathematical and programming skills and a keen interest in probabilistic machine learning and computer vision. Students will be expected to present their work at top-tier computer vision, medical image analysis or machine learning venues such as CVPR, ICCV/ECCV, MICCAI, NeurIPS etc.
The University of Sussex is a leading research-intensive university near Brighton. We have both an international and local outlook, with staff and students from more than 100 countries and frequent engagement in community activities and services.
| Application Deadline | 10 Jun 2022 |
| Country to study | United Kingdom |
| School to study | University of Sussex |
| Type | PhD |
| Sponsor | University of Sussex |
| Gender | Men and Women |
The university will provide a Standard UKRI stipend of £16,062, a research training grant of £1,650 per annum, and full tuition fees up to the overseas rate for 3.5 years.
Apply online for a full time PhD in Informatics using our step-by-step guide (http://www.sussex.ac.uk/study/phd/apply). Here you will also find details of our entry requirements.
Please clearly state on your application form that you are applying for the EPSRC DTP 2022 under the supervision of Dr Ivor Simpson ([email protected])
