I studied Mathematics at the University of Cambridge, before working as a Software Engineer for 4 years and then completing an MSc in Statistics and Operational Research at the University of Edinburgh. At COMPASS, I’m currently researching deep kernel machines – a particular kind of infinite width limit neural network. Kernel methods are infamously difficult to scale, but by applying variational inference as well as work from the Gaussian Process literature, we have developed a practical training algorithm for our deep kernel machines. We hope that by studying these kernel machines and extending them to new architectures, we can gain new insights into representation learning. My project is supervised by Dr. Laurence Aitchison.
