English

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

101
2023-09-27 15:24:41
See translation

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

Related Recommendations
  • Deere Laser's self-developed laser induced sintering technology, with LIF equipment orders exceeding 100GW in production capacity

    Recently, Deere Laser received mass production orders and bid confirmations for laser induced sintering (LIF) equipment from multiple top customers, with a cumulative production capacity exceeding 100GW.As an innovative technology iteratively developed by the company based on its own LIR technology and LIA technology, LIF technology has won industry recognition for its excellent efficiency improve...

    2023-09-18
    See translation
  • Feasibility Study on Composite Manufacturing of Laser Powder Bed Melting and Cold Casting

    It is reported that researchers from the Technical University of Munich in Germany have reported a feasibility study on the composite manufacturing of EN AC-42000 alloy by combining laser powder bed melting and cold casting. The related research titled "Feasibility study on hybrid manufacturing combining laser based powder bed fusion and chill casting on the example of EN AC-42000 alloy" was publi...

    2024-06-06
    See translation
  • Researchers have created an X Lidar lidar to help airports operate during volcanic eruptions

    Engineer and inventor Ezequiel Pawelko is one of the creators of X Lidar, a laser technology that can detect volcanic ash in the atmosphere, draw safe flight paths, and maintain airport operations during volcanic eruptions. Nowadays, he is engaged in other applications such as detecting space debris, monitoring natural resources and fisheries, preventing fires, and drawing radiation and wind maps ...

    2023-12-27
    See translation
  • MKS Malaysia Penang Supercenter Factory Holds Groundbreaking Ceremony

    Recently, MKS Instruments held a groundbreaking and celebration ceremony for its Supercenter factory in Penang, Malaysia.This important moment has been witnessed jointly by the Malaysian Investment Development Authority (MIDA) and Invest Penang, which will help meet the growing demand for semiconductor equipment for wafer manufacturing in the region and globally. This advanced factory, covering ...

    2024-11-01
    See translation
  • Abnormal relativistic emission generated by strong interaction between laser and plasma reflector

    The interaction between strong laser pulses and plasma mirrors has been a focus of recent physical research, as they generate interesting effects. Experiments have shown that these interactions can generate a nonlinear physical process called high-order harmonics, characterized by emitting extreme ultraviolet radiation and brief flashes of laser light.Researchers from the Czech Extreme Light Infra...

    2023-12-04
    See translation