日本語

Scientists demonstrate a new optical neural network training method that can crush electronic microprocessors

931
2023-09-27 15:24:41
翻訳を見る

The current deep neural network system (such as ChatGPT) can quickly improve energy efficiency by 100 times in training, and "future improvements will greatly increase by several orders of magnitude. Scientists from MIT and other institutions have demonstrated a new optical neural network training method that can crush state-of-the-art electronic microprocessors.

Moreover, the computational density of the demonstrated system is about two orders of magnitude higher than that of Nvidia, Google, or Graphcore systems.

Basically, this means that the most advanced models can be trained with 100 times less energy and occupy less space at the same speed.

Artificial neural networks mimic the way biological brains process information. These artificial intelligence systems aim to learn, combine, and summarize information from big datasets, reshaping the field of information processing. Current applications include images, objects, speech recognition, games, medicine, and physical chemistry.

The current artificial intelligence model has reached hundreds of billions of artificial neurons, showing exponential growth and posing challenges to current hardware capabilities.

This paper demonstrates that optical neural network (ONN) methods with high clock speed, parallelism, and low loss data transmission can overcome current limitations.

Our technology opens up a path for large-scale optoelectronic processors to accelerate machine learning tasks from data centers to decentralized edge devices, "the paper wrote.

The ONN method is expected to alleviate the bottlenecks of traditional processors, such as the number of transistors, data mobility energy consumption, and semiconductor size. ONN uses light, which can carry a large amount of information simultaneously due to its wide bandwidth and low data transmission loss. In addition, many photonic circuits can be integrated to expand the system.

In order to move light for calculation, the team led by MIT utilized many laser beams, which were described as "using mass-produced micrometer scale vertical cavity surface emitting lasers for neuron coding".

The researchers explained, "Our scheme is similar to the 'axon synapse dendrite' structure in biological neurons
They believe that the demonstrated system can be expanded through mature wafer level manufacturing processes and photon integration.

Dirk Englund, Associate Professor of Electrical Engineering and Computer Science at the Massachusetts Institute of Technology and the head of this work, explained to SciTechDaily that the size of models such as ChatGPT is limited by the capabilities of today's supercomputers. Therefore, training larger models is not economically feasible.

He claimed, "Our new technology can make it possible to cross machine learning models, otherwise it would not be possible in the near future.

This paper titled "Deep Learning Using Coherent VCSEL Neural Networks" was published by a large team of scientists. This work has received support from the Army Research Office, NTT Research, and NTT Netcast Awards, as well as financial support from the Volkswagen Foundation. The three researchers of the team have applied for patents related to this technology.

Source: Laser Network

関連のおすすめ
  • X photon 3D nanolithography

    Virtual and Physical Prototypes: X-ray laser direct writing 3D nanolithography.Multi-photon polymerization (MPP), also known as 3D nanoprinting, has been investigated using wavelength-tunable femtosecond lasers. At a fixed pulse width of 100 fs, any spectral color in the range of 500nm to 1200nm can be used, which reveals the interaction of more subtle photophysical mechanisms than two-photon phot...

    2023-09-11
    翻訳を見る
  • Research Progress in High Efficiency Supercontinuum Spectra in Specific Wavebands Made by Shanghai Optics and Machinery High Power Laser Unit Technology Laboratory

    Recently, the High Power Laser Unit Technology Laboratory of Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, has made progress in research on high efficiency supercontinuum in specific bands. The relevant research results were published in the Journal of Lightwave Technology under the title of "Strong Anti Stokes and flat supercontinuum in specified band based on non ...

    2023-10-17
    翻訳を見る
  • Coherent Company Announces the Launch of High Power Non Cooled G10 Pumped Laser Module for Submarine and Ground Applications

    Coherent, a leading supplier of high-performance optical network solutions, announced today the launch of a new high-power non cooled pump laser module based on the latest G10 series semiconductor laser tube technology. These new modules are specifically developed for high reliability submarine applications as well as single chip and dual chip ground applications.The new non cooled pump laser modu...

    2024-03-23
    翻訳を見る
  • Researchers use lasers to measure and manipulate magnetic ripple interactions

    One vision for computing the future is to use ripples in magnetic fields as the fundamental mechanism. In this application, magnetic oscillators can be comparable to electricity and serve as the foundation of electronic products.In traditional digital technology, this magnetic system is expected to be much faster than today's technology, from laptops and smartphones to telecommunications. In quant...

    2024-03-05
    翻訳を見る
  • Optimizing the phase focusing of laser accelerators

    With the help of on-chip accelerator technology, researchers at Stanford University are getting closer to manufacturing a miniature electron accelerator that can have various applications in industrial, medical, and physical research.Scientists have proven that silicon dielectric laser accelerators can now be used to accelerate and limit electrons, thereby producing concentrated high-energy electr...

    2024-02-29
    翻訳を見る