Physicists Use Self-Propelled Particles to Create Unconventional Neural Network
-
Physicists created a neural network using synthetic self-propelled colloidal particles rather than electricity and microchips.
-
The particles are controlled by lasers and their rotational motions are used to perform calculations, implementing a type of reservoir computing.
-
The system exhibits noise from the Brownian motion of particles in water, which normally requires large reservoirs to mitigate.
-
They developed a method to reduce noise by using past states of the reservoir, enabling smaller reservoirs.
-
This advances information processing with active matter and provides a technique to optimize physical reservoir computing.