Deep learning for mathematical imaging: CMIT Summer Research Internship 2022
The Centre for Mathematical Imaging Techniques (CMIT) ran a summer intern programme for undergraduate students on "deep learning for mathematical imaging”. The training programme included invited lecturers from Paris, Cambridge and Liverpool. The internship provides opportunities for existing year 1 or 2 students to engage in hands-on research experience. Students worked in small groups and as CMIT has established an informal agreement with the City University of Hong Kong we could exchange two students to offer an enriching experience. This year we took on 27 students and offered each one the opportunity to work with us on implementing and testing convolutional neural network approaches (UNet) to two current mathematical imaging problems: medical image segmentation and discrete tomography.
We organized a series of synchronous online lectures on machine and deep learning, UNets, and implementation details in PyTorch and Numpy/Tensorflow which were kindly given by Dr Liam Burrows (University of Liverpool), Dr Angelica Aviles-Rivero (University of Cambridge), and Dr Taibou Birgui Sekou (Nfinite, France). After these lectures, the students worked in small groups on the problems and presented their results in mid-September. The students Keyuan Bo, Shixin Chen, Mengqi Ding, Yongqi Li, Gulfam Mahmud, Shavarsh Melikyan, Luming Pan, Kun Shi, Xiaoyuan Wang, Yuran Wang, Megxin Xi, Yihe Xia, Zi Yang, Zhijie Yao, and Chuning Zhang finished the planned projects, passed with distinction, and obtained a certificate. Congratulations!
Based on the positive experience and popular demand we will offer this internship again next summer to enable students to gain added experience in research in the fast growing area of data science.
Profesor Ke Chen and Dr Andreas Alpers