Rivan Rughubar (UCT)

Approximating classical kernels on NISQ computers

Abstract:

The talk will aim to demonstrate how a kernel function can be approximated on a NISQ computer and to demonstrate the limitations of this method.

Kernel methods are used throughout classical and quantum machine learning. Over the last few years there has also been much exploration into the link between variational quantum circuits and their kernels and feature maps. Building quantum circuits which exactly implement a given kernel function is not a trivial task. We will look at a method for approximating kernel functions which can be implemented on NISQ computers. This method also attempts to avoid the black box problem in hopes that it can be iterated on and used to approximate a larger family of functions in the future.

Please, download the announcement here.