Projekt

Zur Projekt-Website

Initiative:	Integration molekularer Komponenten in funktionale makroskopische Systeme (beendet, nur noch Fortsetzungsanträge)
Bewilligung:	02.12.2015
Laufzeit:	3 Jahre

The goal of this project is to develop a new nanophotonics based platform for neuro-inspired information processing. Dense arrays of semiconductor microlasers and single photon sources with quantum dots in the active layer will take a role comparable to neurons in the brain. Neuron-connectivity is being established via diffractive coupling by an external spatial light modulator. Similar to the brain's primary sensory cortex, computation is provided by induced macroscopic network-dynamics, which allows for efficient information processing with diverse applications such as pattern classification, nonlinear prediction and ultra-fast control loops. A particularly attractive aspect of the scheme is that it merges the inherently parallel concepts of reservoir computing and photonics within a compact and scalable physical machine learning implementation. As such, an ultra-fast (GHz bandwidth) and versatile platform complementary to recent large-scale electronic approaches (e.g. human brain project, IBM or Google) will be developed.

• Prof. Dr. Stephan Reitzenstein

  Technische Universität Berlin
  Fakultät II
  Institut für Festkörperphysik
  Sekretariat EW 5-3
  Berlin
  

• Daniel Brunner, Ph.D.

  CNRS - Centre National de la Recherche
  Scientifique
  Département d'Optique
  Institut FEMTO-ST
  UMR CNRS 6174
  Office N1-BUR-04
  Besancon (cedex)
  Frankreich
  

• Resonance fluorescence of a site-controlled quantum dot realized by the buriedstressor growth technique

• Developing a photonic hardware platform for brain-inspired computing based on 5 x 5 VCSEL arrays

• Development of Highly Homogenous Quantum Dot Micropillar Arrays for Optical Reservoir Computing

• Reinforcement learning in a large-scale photonic recurrent neural network

• Boolean learning under noise-perturbations in hardware neural networks