Noise as a resource in low power physical computing

There are computational models in which noise, be it electronic, electromagnetic, or quantum, constitutes a computational resource. This phenomenon is known to occur in the brain, where noise allows signals to run on, activate spontaneous firing of neurons, and more. The new frontiers of computing are based on new computational paradigms based on new types of hardware. Among them, hardware that allows direct implementation of neural networks suitable for the study of time series, such as neuromorphic systems based on spiking neurons, and quantum computers stand out for their performances, but at the same time for their very low energy consumption. In the PhysiComp project, we will develop noise-assisted reservoir computing to enable the processing of time series with very low power consumption on both unconventional physical systems suitable for edge computing, such as artificial spiking neuron networks based on CMOS technologies, and quantum computers, including novel hybrid algorithms.