Advances In Quantum Machine Learning




Advances in Quantum Machine Learning

8 October 2019


Added 01-Jan-1970

The first quantum machine learning algorithms suggesting advantage compared to their classical counterparts relied on speedups in subroutines such as matrix inversion and unstructured search. Despite theoretical advantage over the best known classical algorithms these methods can be very sensitive to noise, making them unsuitable for deployment on noisy intermediate scale quantum (NISQ) computers. An alternative approach to learning with quantum computers is to parametrize quantum circuits and train them using data to perform a specific tasks. Similar to classical neural networks, these algorithms exhibit a high noise tolerance making them suitable for implementation on NISQ era devices. With quantum computers now available to test algorithms there has been an explosion of interest in parametrized quantum circuits with applications in chemistry modelling, computer vision, generative modelling, optimization and even factoring. Ed will give an introductory overview to this new field and discuss some recent advances.

Edward Grant is Chief Science Officer at Rahko, a London based quantum software company, using quantum machine learning to assist discovery in chemistry, materials and beyond. Edward is also a PhD candidate in Simone Severini' s group at University College London where he develops NISQ era quantum machine learning algorithms.