Welcome to the Qinsun Instruments Co., LTD! Set to the home page | Collect this site
The service hotline


Related Articles

Product Photo

Contact Us

Qinsun Instruments Co., LTD!
Address:NO.258 Banting Road., Jiuting Town, Songjiang District, Shanghai

Your location: Home > Related Articles > University of California, Los Angeles builds a fully optical diffraction depth neural network for 3D printing

University of California, Los Angeles builds a fully optical diffraction depth neural network for 3D printing

Author:QINSUN Released in:2024-03 Click:20

Researchers from the University of California, Los Angeles have created an artificial neural network using a 3D printer that can analyze large amounts of data and recognize objects at the speed of light. This system is called Diffractive Deep Neural Network (D2NN). It uses the light scattered by objects to identify them. Researchers at the University of California, Los Angeles work together using passive diffraction layers based on deep learning design.

The researchers first created a computer simulation design and then used a 3D printer to create thin polymer wafers measuring 8 centimeters square. Each of these wafers has an uneven surface to help diffract light from objects.

The 3D printed wafer uses terahertz frequency for penetration. Each layer is composed of tens of thousands of pixels, through which light can pass. This design assigns a pixel to each type of object, and light from the object diffracts towards the pixel assigned to it. This technology enables D2NN to recognize objects within the same amount of time as computers need to see them.

This network is trained to learn the diffracted light generated by each object, and when the light from that object passes through the device, it uses an AI branch called deep learning. Deep learning continuously teaches machines through repetition and the emergence of patterns over time. During the experiment, the device was able to accurately recognize handwritten numbers and clothing categories.

The device is also trained as an imaging lens, working in a similar way to a typical camera lens. Since the device was created using a 3D printer, D2NN can be made with larger and additional layers, forming a device with hundreds of millions of artificial neurons. Larger devices can simultaneously discover more objects, making it possible to perform more complex data analysis.

Another key advantage of D2NN is cost, and researchers suggest that the replication cost of the device can be less than $50.