The ultimate goal of artificial intelligence is to mimic the human brain to perform decision-making and control directly from high-dimensional sensory input. Diffractive optical networks provide a promising solution for implementing artificial intelligence with high-speed and low-power consumption. Most of the reported diffractive optical networks focus on single or multiple tasks that do not involve environmental interaction, such as object recognition and image classification. In contrast, the networks capable of performing decision-making and control have not yet been developed to our knowledge. Here, we propose using deep reinforcement learning to implement diffractive optical networks that imitate human-level decision-making and control capability. Such networks taking advantage of a residual architecture, allow for finding optimal control policies through interaction with the environment and can be readily implemented with existing optical devices. The superior performance of these networks is verified by engaging three types of classic games, Tic-Tac-Toe, Super Mario Bros., and Car Racing. Finally, we present an experimental demonstration of playing Tic-Tac-Toe by leveraging diffractive optical networks based on a spatial light modulator. Our work represents a solid step forward in advancing diffractive optical networks, which promises a fundamental shift from the target-driven control of a pre-designed state for simple recognition or classification tasks to the high-level sensory capability of artificial intelligence. It may find exciting applications in autonomous driving, intelligent robots, and intelligent manufacturing.

We demonstrate optical networks composed of diffractive layers trained using deep learning to encode the spatial information of objects into the power spectrum of the diffracted light, which are used to classify objects with a single-pixel spectroscopic detector. Using a plasmonic nanoantenna-based detector, we experimentally validated this single-pixel machine vision framework at terahertz spectrum to optically classify the images of handwritten digits by detecting the spectral power of the diffracted light at ten distinct wavelengths, each representing one class/digit. We also coupled this diffractive network-based spectral encoding with a shallow electronic neural network, which was trained to rapidly reconstruct the images of handwritten digits based on solely the spectral power detected at these ten distinct wavelengths, demonstrating task-specific image decompression. This single-pixel machine vision framework can also be extended to other spectral-domain measurement systems to enable new 3D imaging and sensing modalities integrated with diffractive network-based spectral encoding of information.

Diffractive deep neural network is an optical machine learning framework that blends deep learning with optical diffraction and light-matter interaction to engineer diffractive surfaces that collectively perform optical computation at the speed of light. A diffractive neural network is first designed in a computer using deep learning techniques, followed by the physical fabrication of the designed layers of the neural network using e.g., 3-D printing or lithography. Since the connection between the input and output planes of a diffractive neural network is established via diffraction of light through passive layers, the inference process and the associated optical computation does not consume any power except the light used to illuminate the object of interest. Developed by researchers at UCLA, diffractive optical networks provide a low power, low latency and highly-scalable machine learning platform that can find numerous applications in robotics, autonomous vehicles, defense industry, among many others. In addition to providing statistical inference and generalization to classes of data, diffractive neural networks have also been used to design deterministic optical systems such as a thin imaging system.