Kernel methods in the age of deep learning: from multimodal fusion to quantum machine learning

In this keynote I will explore the potential that kernel-based machine learning offers in the age of deep machine learning, in which Neural Tangent Kernels & Deep Kernel Learning have opened doors to new avenues for understanding deep neural networks. I begin with an introduction to key kernel concepts before exploring developments undertaken by my group and collaborators. I will explore the potential for kernel methods to address issues of multimodal data fusion such as heterogeny and missing data (including an exploration of the neutral point method and the tomographic kernel combination methods).

I will end with an exploration of how Quantum Machine Learning in a NISQ-based setting can be seen as a form of Kernel Machine Learning, with a look at my recent work with Verona University in developing path kernels for quantum machine learning to replicate deep neural network capabilities in the quantum setting.

For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this submission.