CMIH are very pleased to be forging international links with
industry, in particular CMIH postdoc Yu Wang is currently
undertaking a 3 month placement with Microsoft Research Asia, in
Beijing. She will be working alongside Dr. David Wipf, Lead
Researcher with the Visual Computing Group, investigating Bayesian
Deep Learning.

Deep Neural Networks (DNN) have seen shown as a recent success of
empirical achievements on a wide range of machine learning
problems. However, using back propagation for neural net learning
still has some inherent disadvantages, e.g., having to tune a
large number of hyperparameters. In contrast, this research
investigates the effectiveness of a Bayesian treatment, i.e., the
Bayesian approach to learning neural networks, with the hope to
regularize the learning procedure in a principled way. By
exploiting the Bayesian DNN approach, the overfitting problems
above can be hopefully resolved in a statistical approach.
Specifically, this work looks at applying novel Bayesian Neural
Network designs on medical image segmentation, recognition, and
classification problems.