Neuromorphic computing with devices based on self-assembled nanowire networks
Introduction
Electronic device based on self-assembled networks of nanowires that allows the creation of a «hardware» architecture for neuromorphic computing capable of simultaneously processing multiple space-time inputs. This implementation allows the reduction of the number of parameters to be trained and the energy consumption of artificial neural networks.

Technical features
An innovative «hardware» computing architecture, which allows the processing of multiple space-time inputs through a self-assembled network of high-connectivity nanowires (TRL = 4). Unlike conventional computing architectures, this architecture exploits the emergent behavior of the system in a similar way to what happens in biological networks. The operating principle of this network is based on the mutual interaction between memresistive junctions of nanowires where the state of resistance depends on the history of the applied voltage / current. The system is achievable at low cost and allows the implementation of unconventional computing paradigms such as “reservoir computing” thanks to which it is possible to reduce the number of parameters to be trained in artificial neural networks, reducing the energy consumption of the computation.
Possible Applications
- Neuromorphic computing and artificial intelligence;
- Image and pattern recognition;
- Time-series prediction;
- Robotics.
Advantages
- Reduction of the power consumption of computation;
- Reduction of the number of parameters to be trained in neural networks;
- Simultaneous processing of multiple spatiotemporal input;
- Low-cost computing architecture.