Erlangen 2018 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
Q: Fachverband Quantenoptik und Photonik
Q 16: Quantum Information and Simulation
Q 16.2: Vortrag
Montag, 5. März 2018, 14:15–14:30, K 1.020
Open quantum generalisation of classical Hopfield neural networks — •Eliana Fiorelli1,2, Pietro Rotondo1,2, Matteo Marcuzzi1,2, Juan P Garrahan1,2, Markus Muller3, and Igor Lesanovsky1,2 — 1School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD, UK — 2Centre for the Mathematics and Theoretical Physics of Quantum Non-equilibrium Systems, University of Nottingham — 3Department of Physics, Swansea University, Singleton Park, Swansea SA2 8PP, UK
Neural networks (NNs) are artificial networks inspired by the interconnected structure of neurons in animal brains. They are now capable of computational tasks where most ordinary algorithms would fail, such as speech and pattern recognition, with a wide range of applicability both within and outside research. Hopfield NNs [1] constitute a simple, but rich example of how an associative memory can work; they have the ability to retrieve, from a set of stored network states, the one which is closest to the input pattern. In the last decades, many models have been proposed in order to combine the properties of NNs with quantum mechanics, aiming at understanding if NNs computing can take advantage from quantum effects. Here we discuss a quantum generalisation of a classical Hopfield model [2] whose dynamics is governed by purely dissipative, yet quantum, processes. We show that this dynamics may indeed yield an advantage over a purely classical one, leading to a shorter retrieval time. [1] J.J. Hopfield, Proceedings of the National Academy of Sciences, 79, 2554, (1982) [2] P. Rotondo et al., arXiv:1701.01727 (2017).