Multi-View Representation Learning in Computer Vision, Deep Learning, Computer Vision, Multi-View Data – PhD (Funded)

The University of Exeter’s College of Engineering, Mathematics and Physical Sciences, is inviting applications for a fully funded PhD studentship to commence in March 2020 or as soon as possible thereafter. For eligible students the studentship will cover UK/EU/International tuition fees plus an annual tax-free stipend of at least £15,009 for 3.5 years full-time, or pro rata for part-time study. The student would be based in the Innovation Centre Phase 1 in the College of Engineering, Mathematics and Physical Sciences at the Streatham Campus in Exeter.

Project Description:
Multi-view representation learning is concerned with the problem of learning representations (or features) of the multi-view (multi-modal) data that facilitate extracting readily useful information when developing prediction models. This learning mechanism has attracted much attention since multi-view data have become increasingly available in real world applications where examples are described by multi-modal measurements of an underlying signal, such as text + image, sketch + image, audio + video, and many others. Generally, data from different views usually contain complementary information, and multi-view representation learning exploits this point to learn more comprehensive and robust representations than those of single-view learning methods. Since the performance of machine learning methods is heavily dependent on the expressive power of data representation, multi-view representation learning has become a very promising topic with wide applicability. Following the success of deep neural networks, several deep multi-view learning methods are recently proposed based on deep learning. These methods usually apply multi-view learning criteria on top of multiple single-view deep networks; generally, a two-stage scheme of iterative learning is usually adopted to train the network parameters, where view specific features are learned until the very top layers. This is usually done by following a sequential step of multi-view criteria followed by the objectives of the specified learning tasks or regularized learning objectives that seek a balance between multi-view criteria and the final tasks of interest. In this project, our objective is to explore and develop different deep learning models that can solve various computer vision problems involving multi-view data, such as sketch and text-based image retrieval, unsupervised image clustering and classification etc.

This award provides annual funding to cover UK/EU/ International tuition fees and a tax-free stipend. of at least £15,009 per year.

The studentship will be awarded on the basis of merit for 3.5 years of full-time study to commence in March 2020.