Alexander Wang sac pas cher

Switch language

replique montre rolex homme

Highlight dans Nature

processor neuromorphic

COMPUTER SCIENCE - 10 February 2017

Computer designs that mimic how the brain processes information can lead to a vastly improved efficiency when handling complex problems. Researchers have implemented a brain-inspired architecture using photonic hardware. The computational capability was evaluated through a standard speech recognition task, moreover at a processing speed up to one million words per second.

When dealing with complex problems such as image or speech recognition, traditional computers are limited in terms of computational efficiency and energy consumption. Computers today use an architecture where memory is physically separated from the processing unit and instructions are executed step by step. Processing temporal information as the brain does, however, mixes memory and processing to achieve higher computational efficiency and flexibility. Implementing a braininspired computer in photonic hardware, which processes information via light, can lead to further improvements by making use of low-power optical components and providing high speed through the use of broadband telecom devices. We demonstrate the capabilities of such a device, based on a brain-inspired paradigm known as reservoir computing. Our approach exhibits state-of-the-art speed on speech recognition tasks, identifying up to one million words per second with very low error rates.

Our design uses off-the-shelf components to implement a reservoir computing architecture that relies on electro-optical phase delay dynamics, which encodes information in the phase of light waves as opposed to their intensity to provide more accurate and faster  rocessing. We demonstrate speech recognition using a standard database of recorded words, which can be processed and identified by our system after a learning procedure. In addition to our speed performance, we find improvements in computing efficiency compared to other recent implementations of photonic reservoir computing.
Currently the physical parameters of this architecture need to be fine-tuned for the task at hand; a truly universal machine requires no such optimization. Improvements in the complexity of our architecture could achieve not only greater universality but also more computational power. Another big challenge is unsupervised learning, where there is no information about what the result should be.

Phys. Rev. X 7, 011015 (2017) : High-Speed Photonic Reservoir Computing Using a Time-Delay-Based Architecture: 2 Million Words per Second Classification
Laurent Larger, Antonio Baylón-Fuentes, Romain Martinenghi, Vladimir S. Udaltsov, Yanne K. Chembo, and Maxime Jacquot
FEMTO-ST Institute/Optics Department, CNRS & University Bourgogne Franche-Comté, 15B avenue des Montboucons, 25030 Besançon Cedex, France
Vavilov Optical State Institute, Saint-Petersburg, Russia

Other publications: 

  • Chimera states in Nonlinear Delay Dynamics: Larger et al., Nat. Commun. 2015, 7752.
  • Discovery of the Chimera states: Larger et al., Phys. Rev. Lett. 2013, 054103.
  • 1st demo. of photonic neuromorphic computing: Larger et al., Opt. Ex. 2012, 3241.
  • Fast neuromorphic photonic computer: Martinenghi et al. Phys. Rev. Lett. 2012, 244101.
fake Alexander Wang taschen
Alexander Wang bolsos españa