Neuromimetic photonics

Neuromimetic systems  imitate the functionalities of biological neurons or the architectures of biological systems for neuro-inspired computation and information processing. These systems could allow to fulfill complex tasks (visual or vocal recognition, associative memories, logic...) getting inspiration from the way the brain process information while being energy efficient, fast and not very sensitive to noisy environments. In this project we investigate how neuromimetic (or neuromorphic) systems can be implemented in photonics for e.g. artificial neural networks, reservoir computing…

  Objectives

Our main system are excitable micropillar lasers with integrated saturable absorber. We have shown that these microlasers can be integrated and display interesting and ultra-fast neuromimetic properties such as refractory periods (absolute and relative), temporal summation, regenerative spiking, to mention a few Our aim is to investigate the physics of these building blocks and how they could be used as nodes in a network to efficiently process information and compute (artificial neural networks).