Efficient Optical Computing with Exciton-Polaritons

Explore a 57-minute lecture on efficient optical computing with exciton-polaritons, delivered by Prof. Michał Matuszewski from the Institute of Physics PAS. Delve into the world of photonic information processing and its advantages, including high speed, parallelization, low communication losses, and high bandwidth. Discover recent demonstrations of hardware neural network systems utilizing strong optical nonlinearity derived from exciton-polariton interactions. Learn about these quantum superpositions of light and matter, their excellent transport properties, and strong interactions. Understand how semiconductor microcavity systems can be used to construct energy-efficient, all-optical neural networks. Examine why polaritonics may be superior to standard nonlinear optical phenomena in achieving high performance. The lecture covers topics such as machine learning, neuromorphic computing, integrated optical neural networks, diffractive neural networks, optoelectronic neural networks, all-optical neural networks, experiments in reservoir computing, binarized neural networks, and spiking neurons. Gain insights into the future outlook of polariton neural networks, the limits of neural network efficiency, and reservoir computing with exciton-polaritons.

Intro
Collaboration
Machine learning
Neuromorphic computing
Integrated optical neural networks
Diffractive neural networks
Optoelectronic neural networks
All-optical neural networks
Experiments: reservoir computing
Experiments: binarized neural network
Experiments: spiking neurons
Polariton neural networks: outlook
Limits of neural network efficiency
Reservoir computing with exciton-polaritons