Sitzungen | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
FM: Fall Meeting
FM 90: Special Session: Quantum Physics for AI & AI for Quantum Physics
FM 90.1: Invited Talk
Freitag, 27. September 2019, 11:00–11:30, Audi Max
How to use quantum light to machine learn graph-structured data — •Maria Schuld1,2, Kamil Bradler1, Robert Israel1, Daiqin Su1, and Brajesh Gupt1 — 1Xanadu, Toronto, Canada — 2University of KwaZulu-Natal, Durban, South Africa
A device called a 'Gaussian Boson Sampler' has initially been proposed as a near-term demonstration of classically intractable quantum computation. But these devices can also be used to decide whether two graphs are similar to each other, which is the central challenge when doing machine learning on data represented by graphs, such as molecules and social networks. In this talk, I will show how to construct a graph similarity measure - or 'graph kernel' as it is known in machine learning - using samples from an optical Gaussian Boson Sampler. Combining this with standard machine learning methods allows us to predict features of the graph using example data. I will present promising benchmark results comparing the 'quantum kernel' to 'classical kernels' and motivate theoretically why such a continuous-variable quantum computer can actually extract interesting properties. The work is an example of how to use first-generation quantum technologies for machine learning tasks.