Researchers from the University of California, Los Angeles, used a 3D printer to create an artificial neural network capable of analyzing large amounts of data and identifying objects at the speed of light. The system is called the deep refractive neural network (D2NN). Uses the light scattered by an object to define the object.
UCLA researchers based the system on a deep learning-based design using passive refraction layers working together. The researchers first created a design that simulated a computer and then used a 3D printer to create thin polymer foils measuring eight square centimeters. Each of these chips has uneven surfaces to aid in scattering the light coming from an object.
The 3D printed chips It is penetrated using terahertz frequencies. Each layer is made up of tens of thousands of pixels that light can travel through. The design has each type of object assigned a pixel to it, with the light coming from the object skewed towards the pixel assigned to its type. This technology allows D2NN to identify an object for the same amount of time that a computer takes to see the object.
The network was trained to know the diffracted light produced by each object while the light from that object passes through the device using a branch of artificial intelligence called deep learning. Deep learning teaches machines through iteration and over time as patterns emerge. During the experiments, the device was able to accurately identify handwritten numbers and clothing items.
The device has also been trained to function as a photographic lens, similar to how a typical viewfinder would. Since the device was built with a 3D printer, the D2NN can be fabricated with larger and additional layers resulting in a device with hundreds of millions of artificial neurons. Larger devices can find many objects simultaneously with the ability to perform more complex data analysis. Another important aspect of D2NN is the cost, as the researcher said the device could be reproduced for less than $ 50.