Light neuron promises photonic artificial brain

electronics

Editor of the Technological Innovation Website – 11/24/2022

Neur

The optical microcavity works like a neuron – a neuron that fires pulses of light, not electrical pulses.
[Imagem: Mateusz Krol]

light neuron

Neuromorphic computing, or neurocomputing, has just gained yet another hardware alternative.

And this time, an alternative that promises full speed.

Researchers at the University of Warsaw in Poland have used photons – the basic units of light – to create an artificial neuron that could become the building block of a future photonic neural network processor.

And even before this futuristic possibility becomes a reality, the photonic neuron could become the basis for implementing traditional artificial intelligence in hardware.

“We propose to use a new computational paradigm based on encoding information with pulses of light that trigger a signal only when it reaches the neuron successively, at the right time,” explained Professor Andrzej Opala.

Currently, neural networks use layers of interconnected neurons that fire pulses based on the importance assigned to each connection, the so-called “weights”. In the optical neural network just developed, neurons fire (that is, become active) in response to a string of pulses, which can have different intensities and time intervals.

As with biological neurons, which are excited by electrical pulses, there is a certain threshold above which the career of light pulses reaching the optic neuron triggers a signal that will be passed on to other neurons, allowing the creation of a neural network. fully working with light.

Neur

Photonic circuits run at the speed of light and consume a fraction of the energy of their electronic counterparts.
[Imagem: Krzysztof Tyszka et al. – 10.1002/lpor.202100660]

light neural network

As photons virtually do not interact with each other, the team had to make use of quasiparticles, entities that form when fundamental particles, such as photons, electrons, holes, etc., somehow form a set that begins to exhibit its own behavior.

In this case, the team used xcitons, which arise from the interaction between photons and electrons, and polyritons, which combine an excited molecular state with a photon. In the first case, there is a transformation of light into electricity, while in the second case, light and matter have similar properties.

Polyritons make it possible to imitate a biological system because only stimulation with the appropriate number of photons, above a certain threshold, leads to the formation of a Bose-Einstein condensate and then the emission of a photon, on the picosecond scale, which becomes the signal for the next neuron.

“In our research, we propose a solution in which photons interact strongly with very low-mass particles, called xcitons,” detailed researcher Barbara Pietka. “This strong interaction is possible when photons and xcitons are trapped together in so-called optical microcavities, which forces the repetitive exchange of energy between them. This kind of synergy, generated in the microcavity between a photon and a xciton, is so persistent that physicists call it a quasiparticle and refer to it as a polyriton.”

“We were the first to realize that when polyritons are excited by laser pulses, they emit pulses of light in a way that mimics the firing of biological neurons,” concluded Professor Magdalena Furman.

But there are still challenges to be overcome. Polyriton-based firing only works when an intermediate Bose-Einstein condensate has formed; although of very short duration, this only happens in cryogenic temperatures. “Our further aim is to transfer the experiment from cryogenic conditions to room temperature,” said Professor Jacek Szczytko.

Bibliography:

Article: Leaky Integrate-and-Fire Mechanism in Exciton-Polariton Condensates for Photonic Spiking Neuron
Authors: Krzysztof Tyszka, Magdalena Furman, Rafal Mirek, Mateusz Krl, Andrzej Opala, Bartlomiej Seredynski, Jan Suffczynski, Wojciech Pacuski, Michal Matuszewski, Jacek Szczytko, Barbara Pietka
Magazine: Laser & Photonics Review
DOI: 10.1002/lpor.202100660

Follow Technological Innovation Site on Google News

Other news about:

More topics

Source link

About Admin

Check Also

Traditional app advertises one free app a day until Christmas » iPhone Blog

Those who are older in the iPhone world remember the time when there were always …

Leave a Reply

Your email address will not be published. Required fields are marked *