Fiber optic cables have bottleneck issues. They can transport encoded data at the speed of light, but converting encoded data into understandable information often requires slower, much more energy-hungry equipment. However, although built from previous innovations in the field known as passive neural networks, a team at the Shanghai University of Science and Technology (USST) in China is developing a microscopic workaround. A part of the energy. Plus, each chip is hardly salt-sized.
Recent advances detailed in research published in journals Natural Photonics It relies on a form of neural networking, originally developed by researchers at the University of California, Los Angeles in 2018.All optical diffraction deep neural networks“This method uses a layer of patterned 3D printed passive components. The system is then trained to use photons of light to complete complex calculations.
As New Scientist To explain, the USST team recently used this concept as a starting point to create a “passive and well-trained neural network” that physically manipulates light to perform computational analysis. However, all the light on which the data is encoded is transmitted through a wider fiber-optic gaze than one human hair. Therefore, the AI chip had to be made equally small to read each photon.
Researchers relied on “3D two-photon nanolithography” to construct each smallest chip using ultra-thin polymer layers. The chip was then attached to the end of the fiber optic wire. There, they processed the data as it passed through the cable at light speed. To test the present invention, the team encoded images of numbers into photons and sent them through fiber optic wires. The AI chip then read the data successfully and recreated it with each number to a minimum. This kind of image recognition is now a basic feature in many AI systems, with salt-sized chips able to do it in a trillion second. They also do so using only thousands of energy amounts as today’s AI-based image recognition technology.
The system is not perfect yet. Small chip defects can dismantle the entire system, and each chip must be specially customized according to the job required. Still, the inventors believe that the technology can ultimately provide “unprecedented features.” These may include situations such as endoscopic imaging and quantum computing.
Other deals, reviews, and purchase guides
