k-space Neural computation with magnEtic exciTations (k-NET)

Basic data for this project

Description

Artificial neural networks are computing systems inspired by biological neural networks. They emulate the brain by using nonlinear elements that act as neurons interconnected through artificial synapses. Current architectures are facing challenges: the number of synapses implemented is very limited compared with the tens of thousands in the human brain. Furthermore, changing the weight of each connection requires additional memory elements. The EU-funded k-NET project will circumvent these issues. It proposes new architecture based on the idea that dynamical hyperconnectivity can be implemented not in real space but in reciprocal or k-space. To demonstrate this novel approach, researchers will select ferromagnetic nanostructures in which the populations of spin waves – the elementary excitations – play the role of neurons.