All-optical machine learning using diffractive deep neural networks

Acoustic-holography-patterned primary hepatocytes possess liver functions

Acoustic holography (AH), a promising approach for cell patterning, emerges as a powerful tool for constructing novel invitro 3D models that mimic organs and cancers features. However, understanding changes in cell function post-AH remains limited. Furthermore, replicating complex physiological and pathological processes solely with cell lines proves challenging. Here, we employed acoustical holographic lattice to assemble primary hepatocytes directly isolated from mice into a cell cluster matrix to construct a liver-shaped tissue sample. For the first time, we evaluated the liver functions of AH-patterned primary hepatocytes. The patterned model exhibited large numbers of self-assembled spheroids and superior multifarious core hepatocyte functions compared to cells in 2D and traditional 3D culture models. AH offers a robust protocol for long-term in vitro culture of primary cells, underscoring its potential for future applications in disease pathogenesis research, drug testing, and organ replacement therapy.

Terahertz spoof plasmonic neural network for diffractive information recognition and processing

All-optical diffractive neural networks, as analog artificial intelligence accelerators, leverage parallelism and analog computation for complex data processing. However, their low space transmission efficiency or large spatial dimensions hinder miniaturization and broader application. Here, we propose a terahertz spoof plasmonic neural network on a planar diffractive platform for direct multi-target recognition. Our approach employs a spoof surface plasmon polariton coupler array to construct a diffractive network layer, resulting in a compact, efficient, and easily integrable architecture. We designed three schemes: basis vector classification, multi-user recognition, and MNIST handwritten digit classification. Experimental results reveal that the terahertz spoof plasmonic neural network successfully classifies basis vectors, recognizes multi-user orientation information, and directly processes handwritten digits using a designed input framework comprising a metal grating array, transmitters, and receivers. This work broadens the application of terahertz plasmonic metamaterials, paving the way for terahertz on-chip integration, intelligent communication, and advanced computing systems.

Femtosecond laser direct nanolithography of perovskite hydration for temporally programmable holograms

Modern nanofabrication technologies have propelled significant advancement of high-resolution and optically thin holograms. However, it remains a long-standing challenge to tune the complex hologram patterns at the nanoscale for temporal light field control. Here, we report femtosecond laser direct lithography of perovskites with nanoscale feature size and pixel-level temporal dynamics control for temporally programmable holograms. Specifically, under tightly focused laser irradiation, the organic molecules of layered perovskites (PEA)2PbI4 can be exfoliated with nanometric thickness precision and subwavelength lateral size. This creates inorganic lead halide capping nanostructures that retard perovskite hydration, enabling tunable hydration time constant. Leveraging advanced inverse design methods, temporal holograms in which multiple independent images are multiplexed with low cross talk are demonstrated. Furthermore, cascaded holograms are constructed to form temporally holographic neural networks with programmable optical inference functionality. Our work opens up new opportunities for tunable photonic devices with broad impacts on holography display and storage, high-dimensional optical encryption and artificial intelligence.

Pyramid diffractive optical networks for unidirectional image magnification and demagnification

Diffractive deep neural networks (D2NNs) are composed of successive transmissive layers optimized using supervised deep learning to all-optically implement various computational tasks between an input and output field-of-view. Here, we present a pyramid-structured diffractive optical network design (which we term P-D2NN), optimized specifically for unidirectional image magnification and demagnification. In this design, the diffractive layers are pyramidally scaled in alignment with the direction of the image magnification or demagnification. This P-D2NN design creates high-fidelity magnified or demagnified images in only one direction, while inhibiting the image formation in the opposite direction-achieving the desired unidirectional imaging operation using a much smaller number of diffractive degrees of freedom within the optical processor volume. Furthermore, the P-D2NN design maintains its unidirectional image magnification/demagnification functionality across a large band of illumination wavelengths despite being trained with a single wavelength. We also designed a wavelength-multiplexed P-D2NN, where a unidirectional magnifier and a unidirectional demagnifier operate simultaneously in opposite directions, at two distinct illumination wavelengths. Furthermore, we demonstrate that by cascading multiple unidirectional P-D2NN modules, we can achieve higher magnification factors. The efficacy of the P-D2NN architecture was also validated experimentally using terahertz illumination, successfully matching our numerical simulations. P-D2NN offers a physics-inspired strategy for designing task-specific visual processors.

Pre-sensor computing with compact multilayer optical neural network

Moving computation units closer to sensors is becoming a promising approach to addressing bottlenecks in computing speed, power consumption, and data storage. Pre-sensor computing with optical neural networks (ONNs) allows extensive processing. However, the lack of nonlinear activation and dependence on laser input limits the computational capacity, practicality, and scalability. A compact and passive multilayer ONN (MONN) is proposed, which has two convolution layers and an inserted nonlinear layer, performing pre-sensor computations with designed passive masks and a quantum dot film for incoherent light. MONN has an optical length as short as 5 millimeters, two orders of magnitude smaller than state-of-the-art lens-based ONNs. MONN outperforms linear single-layer ONN across various vision tasks, off-loading up to 95% of computationally expensive operations into optics from electronics. Motivated by MONN, a paradigm is emerging for mobile vision, fulfilling the demands for practicality, miniaturization, and low power consumption.

Unlocking ultra-high holographic information capacity through nonorthogonal polarization multiplexing

Contemporary studies in polarization multiplexing are hindered by the intrinsic orthogonality constraints of polarization states, which restrict the scope of multiplexing channels and their practical applications. This research transcends these barriers by introducing an innovative nonorthogonal polarization-basis multiplexing approach. Utilizing spatially varied eigen-polarization states within metaatoms, we successfully reconstruct globally nonorthogonal channels that exhibit minimal crosstalk. This method not only facilitates the generation of free-vector holograms, achieving complete degrees-of-freedom in three nonorthogonal channels with ultra-low energy leakage, but it also significantly enhances the dimensions of the Jones matrix, expanding it to a groundbreaking 10 × 10 scale. The fusion of a controllable eigen-polarization engineering mechanism with a vectorial diffraction neural network culminates in the experimental creation of 55 intricate holographic patterns across these expanded channels. This advancement represents a profound shift in the field of polarization multiplexing, unlocking opportunities in advanced holography and quantum encryption, among other applications.

Reconfigurable metamaterial processing units that solve arbitrary linear calculus equations

Calculus equations serve as fundamental frameworks in mathematics, enabling describing an extensive range of natural phenomena and scientific principles, such as thermodynamics and electromagnetics. Analog computing with electromagnetic waves presents an intriguing opportunity to solve calculus equations with unparalleled speed, while facing an inevitable tradeoff in computing density and equation reconfigurability. Here, we propose a reconfigurable metamaterial processing unit (MPU) that solves arbitrary linear calculus equations at a very fast speed. Subwavelength kernels based on inverse-designed pixel metamaterials are used to perform calculus operations on time-domain signals. In addition, feedback mechanisms and reconfigurable components are used to formulate and solve calculus equations with different orders and coefficients. A prototype of this MPU with a compact planar size of 0.93λ0×0.93λ0 (λ0 is the free-space wavelength) is constructed and evaluated in microwave frequencies. Experimental results demonstrate the MPU's ability to successfully solve arbitrary linear calculus equations. With the merits of compactness, easy integration, reconfigurability, and reusability, the proposed MPU provides a potential route for integrated analog computing with high speed of signal processing.

Nonlinear encoding in diffractive information processing using linear optical materials

Nonlinear encoding of optical information can be achieved using various forms of data representation. Here, we analyze the performances of different nonlinear information encoding strategies that can be employed in diffractive optical processors based on linear materials and shed light on their utility and performance gaps compared to the state-of-the-art digital deep neural networks. For a comprehensive evaluation, we used different datasets to compare the statistical inference performance of simpler-to-implement nonlinear encoding strategies that involve, e.g., phase encoding, against data repetition-based nonlinear encoding strategies. We show that data repetition within a diffractive volume (e.g., through an optical cavity or cascaded introduction of the input data) causes the loss of the universal linear transformation capability of a diffractive optical processor. Therefore, data repetition-based diffractive blocks cannot provide optical analogs to fully connected or convolutional layers commonly employed in digital neural networks. However, they can still be effectively trained for specific inference tasks and achieve enhanced accuracy, benefiting from the nonlinear encoding of the input information. Our results also reveal that phase encoding of input information without data repetition provides a simpler nonlinear encoding strategy with comparable statistical inference accuracy to data repetition-based diffractive processors. Our analyses and conclusions would be of broad interest to explore the push-pull relationship between linear material-based diffractive optical systems and nonlinear encoding strategies in visual information processors.

Direct electromagnetic information processing with planar diffractive neural network

Diffractive neural network in electromagnetic wave-driven system has attracted great attention due to its ultrahigh parallel computing capability and energy efficiency. However, recent neural networks based on the diffractive framework still face the bottlenecks of misalignment and relatively large size limiting their further applications. Here, we propose a planar diffractive neural network (pla-NN) with a highly integrated and conformal architecture to achieve direct signal processing in the microwave frequency. On the basis of printed circuit fabrication process, the misalignment could be effectively circumvented while enabling flexible extension for multiple conformal and stacking designs. We first conduct validation on the fashion-MNIST dataset and experimentally build up a system using the proposed network architecture for direct recognition of different geometry structures in the electromagnetic space. We envision that the presented architecture, once combined with the advanced dynamic maneuvering techniques and flexible topology, would exhibit unlimited potentials in the areas of high-performance computing, wireless sensing, and flexible wearable electronics.

Polarization and wavelength routers based on diffractive neural network

In the field of information processing, all-optical routers are significant for achieving high-speed, high-capacity signal processing and transmission. In this study, we developed three types of structurally simple and flexible routers using the deep diffractive neural network (D2NN), capable of routing incident light based on wavelength and polarization. First, we implemented a polarization router for routing two orthogonally polarized light beams. The second type is the wavelength router that can route light with wavelengths of 1550, 1300, and 1100 nm, demonstrating outstanding performance with insertion loss as low as 0.013 dB and an extinction ratio of up to 18.96 dB, while also maintaining excellent polarization preservation. The final router is the polarization-wavelength composite router, capable of routing six types of input light formed by pairwise combinations of three wavelengths (1550, 1300, and 1100 nm) and two orthogonal linearly polarized lights, thereby enhancing the information processing capability of the device. These devices feature compact structures, maintaining high contrast while exhibiting low loss and passive characteristics, making them suitable for integration into future optical components. This study introduces new avenues and methodologies to enhance performance and broaden the applications of future optical information processing systems.

Influence of discontinuities on photonic waveguides

Fabrication-induced imperfections in photonic wire waveguides, such as roughness, stitching errors, and discontinuities, degrade their performance and thereby lower the yield of large-scale systems. This degradation is primarily due to the high insertion losses induced by imperfections, which scale nonlinearly with the index contrast in wire waveguides. Here we investigate the influence of discontinuities in photonic waveguides and later show a platform that is robust to fabrication imperfections. Our platform is based on an array of silicon nano-pillars, arranged to form a sub-wavelength (SW) grating waveguide. We focus on investigating the robustness by considering an abrupt break in the waveguide, as an extreme case of discontinuity. We show that sub-wavelength silicon waveguides are robust against unwanted large discontinuities relative to the operating wavelength. We measure a transmission loss of <2.2 dB at 1550 n m, for a discontinuity of length 2.1 μ m, when compared to more than 7 d B of loss in conventional silicon wire waveguides for the same discontinuity. Our results show that this mode of protection is broadband, covering the entire telecommunication band (λ =1500-1600 nm). We believe that this investigation of the influence of discontinuities in photonic waveguides could be a step toward the realization of low-loss optical waveguides.

Computational Investigation of the Potential and Limitations of Machine Learning with Neural Network Circuits Based on Synaptic Transistors

Synaptic transistors have been proposed to implement neuron activation functions of neural networks (NNs). While promising to enable compact, fast, inexpensive, and energy-efficient dedicated NN circuits, they also have limitations compared to digital NNs (realized as codes for digital processors), including shape choices of the activation function using particular types of transistor implementation, and instabilities due to noise and other factors present in analog circuits. We present a computational study of the effects of these factors on NN performance and find that, while accuracy competitive with traditional NNs can be realized for many applications, there is high sensitivity to the instability in the shape of the activation function, suggesting that, when highly accurate NNs are required, high-precision circuitry should be developed beyond what has been reported for synaptic transistors to date.

Super-resolution diffractive neural network for all-optical direction of arrival estimation beyond diffraction limits

Wireless sensing of the wave propagation direction from radio sources lays the foundation for communication, radar, navigation, etc. However, the existing signal processing paradigm for the direction of arrival estimation requires the radio frequency electronic circuit to demodulate and sample the multichannel baseband signals followed by a complicated computing process, which places the fundamental limit on its sensing speed and energy efficiency. Here, we propose the super-resolution diffractive neural networks (S-DNN) to process electromagnetic (EM) waves directly for the DOA estimation at the speed of light. The multilayer meta-structures of S-DNN generate super-oscillatory angular responses in local angular regions that can perform the all-optical DOA estimation with angular resolutions beyond the diffraction limit. The spatial-temporal multiplexing of passive and reconfigurable S-DNNs is utilized to achieve high-resolution DOA estimation over a wide field of view. The S-DNN is validated for the DOA estimation of multiple radio sources over 5 GHz frequency bandwidth with estimation latency over two to four orders of magnitude lower than the state-of-the-art commercial devices in principle. The results achieve the angular resolution over an order of magnitude, experimentally demonstrated with four times, higher than diffraction-limited resolution. We also apply S-DNN's edge computing capability, assisted by reconfigurable intelligent surfaces, for extremely low-latency integrated sensing and communication with low power consumption. Our work is a significant step towards utilizing photonic computing processors to facilitate various wireless sensing and communication tasks with advantages in both computing paradigms and performance over electronic computing.

Nanowatt all-optical 3D perception for mobile robotics

Three-dimensional (3D) perception is vital to drive mobile robotics' progress toward intelligence. However, state-of-the-art 3D perception solutions require complicated postprocessing or point-by-point scanning, suffering computational burden, latency of tens of milliseconds, and additional power consumption. Here, we propose a parallel all-optical computational chipset 3D perception architecture (Aop3D) with nanowatt power and light speed. The 3D perception is executed during the light propagation over the passive chipset, and the captured light intensity distribution provides a direct reflection of the depth map, eliminating the need for extensive postprocessing. The prototype system of Aop3D is tested in various scenarios and deployed to a mobile robot, demonstrating unprecedented performance in distance detection and obstacle avoidance. Moreover, Aop3D works at a frame rate of 600 hertz and a power consumption of 33.3 nanowatts per meta-pixel experimentally. Our work is promising toward next-generation direct 3D perception techniques with light speed and high energy efficiency.

Generating Multi-Depth 3D Holograms Using a Fully Convolutional Neural Network

Efficiently generating 3D holograms is one of the most challenging research topics in the field of holography. This work introduces a method for generating multi-depth phase-only holograms using a fully convolutional neural network (FCN). The method primarily involves a forward-backward-diffraction framework to compute multi-depth diffraction fields, along with a layer-by-layer replacement method (L2RM) to handle occlusion relationships. The diffraction fields computed by the former are fed into the carefully designed FCN, which leverages its powerful non-linear fitting capability to generate multi-depth holograms of 3D scenes. The latter can smooth the boundaries of different layers in scene reconstruction by complementing information of occluded objects, thus enhancing the reconstruction quality of holograms. The proposed method can generate a multi-depth 3D hologram with a PSNR of 31.8 dB in just 90 ms for a resolution of 2160 × 3840 on the NVIDIA Tesla A100 40G tensor core GPU. Additionally, numerical and experimental results indicate that the generated holograms accurately reconstruct clear 3D scenes with correct occlusion relationships and provide excellent depth focusing.

Chiral Kirigami for Bend-Tolerant Reconfigurable Hologram with Continuously Variable Chirality Measures

Despite the commonality of static holograms, the holography with multiple information layers and reconfigurable grey-scale images at communication frequencies remain a confluence of scientific challenges. One well-known difficulty is the simultaneous modulation of phase and amplitude of electromagnetic wavefronts with a high modulation depth. A less appreciated challenge is scrambling of the information and images with hologram bending. Here, this work shows that chirality-guided pixelation of plasmonic kirigami sheets enables tunable multiplexed holography at terahertz (THz) frequencies. The convex and concave structures with slanted Au strips exhibit gradual variations in geometries facilitating modulation of light ellipticity reaching 40 deg. Real-time switching of 3D images of the letter "M" and the Mona Lisa demonstrates the possibility of complex grey-scale information content and importance of continuously variable mirror asymmetry. Microscale chirality measures of each pixel experiences little change with bending while retaining controllable reconfigurability upon stretching, which translates to remarkable resilience of chiral holograms to bending. Simplicity of their design with local chirality measures opens the door to information technologies with fault-tolerant THz encryption, wearable holographic devices, and new communication technologies.

Photonics for Neuromorphic Computing: Fundamentals, Devices, and Opportunities

In the dynamic landscape of Artificial Intelligence (AI), two notable phenomena are becoming predominant: the exponential growth of large AI model sizes and the explosion of massive amount of data. Meanwhile, scientific research such as quantum computing and protein synthesis increasingly demand higher computing capacities. As the Moore's Law approaches its terminus, there is an urgent need for alternative computing paradigms that satisfy this growing computing demand and break through the barrier of the von Neumann model. Neuromorphic computing, inspired by the mechanism and functionality of human brains, uses physical artificial neurons to do computations and is drawing widespread attention. This review studies the expansion of optoelectronic devices on photonic integration platforms that has led to significant growth in photonic computing, where photonic integrated circuits (PICs) have enabled ultrafast artificial neural networks (ANN) with sub-nanosecond latencies, low heat dissipation, and high parallelism. In particular, various technologies and devices employed in neuromorphic photonic AI accelerators, spanning from traditional optics to PCSEL lasers are examined. Lastly, it is recognized that existing neuromorphic technologies encounter obstacles in meeting the peta-level computing speed and energy efficiency threshold, and potential approaches in new devices, fabrication, materials, and integration to drive innovation are also explored. As the current challenges and barriers in cost, scalability, footprint, and computing capacity are resolved one-by-one, photonic neuromorphic systems are bound to co-exist with, if not replace, conventional electronic computers and transform the landscape of AI and scientific computing in the foreseeable future.

Ultracompact and multifunctional integrated photonic platform

Realizing a multifunctional integrated photonic platform is one of the goals for future optical information processing, which usually requires large size to realize due to multiple integration challenges. Here, we realize a multifunctional integrated photonic platform with ultracompact footprint based on inverse design. The photonic platform is compact with 86 inverse designed-fixed couplers and 91 phase shifters. The footprint of each coupler is 4 μm by 2 μm, while the whole photonic platform is 3 mm by 0.2 mm-one order of magnitude smaller than previous designs. One-dimensional Floquet Su-Schrieffer-Heeger model and Aubry-André-Harper model are performed with measured fidelities of 97.90 (±0.52) % and 99.34 (±0.44) %, respectively. We also demonstrate a handwritten digits classification task with the test accuracy of 87% using on-chip training. Moreover, the scalability of this platform has been proved by demonstrating more complex computing tasks. This work provides an effective method to realize an ultrasmall integrated photonic platform.

Memory-less scattering imaging with ultrafast convolutional optical neural networks

The optical memory effect in complex scattering media including turbid tissue and speckle layers has been a critical foundation for macroscopic and microscopic imaging methods. However, image reconstruction from strong scattering media without the optical memory effect has not been achieved. Here, we demonstrate image reconstruction through scattering layers where no optical memory effect exists, by developing a multistage convolutional optical neural network (ONN) integrated with multiple parallel kernels operating at the speed of light. Training this Fourier optics-based, parallel, one-step convolutional ONN with the strong scattering process for direct feature extraction, we achieve memory-less image reconstruction with a field of view enlarged by a factor up to 271. This device is dynamically reconfigurable for ultrafast multitask image reconstruction with a computational power of 1.57 peta-operations per second (POPS). Our achievement establishes an ultrafast and high energy-efficient optical machine learning platform for graphic processing.

Information-hiding cameras: Optical concealment of object information into ordinary images

We introduce an information-hiding camera integrated with an electronic decoder that is jointly optimized through deep learning. This system uses a diffractive optical processor, which transforms and hides input images into ordinary-looking patterns that deceive/mislead observers. This information-hiding transformation is valid for infinitely many combinations of secret messages, transformed into ordinary-looking output images through passive light-matter interactions within the diffractive processor. By processing these output patterns, an electronic decoder network accurately reconstructs the original information hidden within the deceptive output. We demonstrated our approach by designing information-hiding diffractive cameras operating under various lighting conditions and noise levels, showing their robustness. We further extended this framework to multispectral operation, allowing the concealment and decoding of multiple images at different wavelengths, performed simultaneously. The feasibility of our framework was also validated experimentally using terahertz radiation. This optical encoder-electronic decoder-based codesign provides a high speed and energy efficient information-hiding camera, offering a powerful solution for visual information security.

All-optical phase conjugation using diffractive wavefront processing

Optical phase conjugation (OPC) is a nonlinear technique used for counteracting wavefront distortions, with applications ranging from imaging to beam focusing. Here, we present a diffractive wavefront processor to approximate all-optical phase conjugation. Leveraging deep learning, a set of diffractive layers was optimized to all-optically process an arbitrary phase-aberrated input field, producing an output field with a phase distribution that is the conjugate of the input wave. We experimentally validated this wavefront processor by 3D-fabricating diffractive layers and performing OPC on phase distortions never seen during training. Employing terahertz radiation, our diffractive processor successfully performed OPC through a shallow volume that axially spans tens of wavelengths. We also created a diffractive phase-conjugate mirror by combining deep learning-optimized diffractive layers with a standard mirror. Given its compact, passive and multi-wavelength nature, this diffractive wavefront processor can be used for various applications, e.g., turbidity suppression and aberration correction across different spectral bands.

Large-scale photonic computing with nonlinear disordered media

Neural networks find widespread use in scientific and technological applications, yet their implementations in conventional computers have encountered bottlenecks due to ever-expanding computational needs. Photonic computing is a promising neuromorphic platform with potential advantages of massive parallelism, ultralow latency and reduced energy consumption but mostly for computing linear operations. Here we demonstrate a large-scale, high-performance nonlinear photonic neural system based on a disordered polycrystalline slab composed of lithium niobate nanocrystals. Mediated by random quasi-phase-matching and multiple scattering, linear and nonlinear optical speckle features are generated as the interplay between the simultaneous linear random scattering and the second-harmonic generation, defining a complex neural network in which the second-order nonlinearity acts as internal nonlinear activation functions. Benchmarked against linear random projection, such nonlinear mapping embedded with rich physical computational operations shows improved performance across a large collection of machine learning tasks in image classification, regression and graph classification. Demonstrating up to 27,648 input and 3,500 nonlinear output nodes, the combination of optical nonlinearity and random scattering serves as a scalable computing engine for diverse applications.

An Ultrathin, Fast-Response, Large-Scale Liquid-Crystal-Facilitated Multi-Functional Reconfigurable Metasurface for Comprehensive Wavefront Modulation

The rapid advancement of prevailing communication/sensing technologies necessitates cost-effective millimeter-wave arrays equipped with a massive number of phase-shifting cells to perform complicated beamforming tasks. Conventional approaches employing semiconductor switch/varactor components or tunable materials encounter obstacles such as quantization loss, high cost, high complexity, and limited adaptability for realizing large-scale arrays. Here, a low-cost, ultrathin, fast-response, and large-scale solution relying on metasurface concepts combined together with liquid crystal (LC) materials requiring a layer thickness of only 5 µm is reported. Rather than immersing resonant structures in LCs, a joint material-circuit-based strategy is devised, via integrating deep-subwavelength-thick LCs into slow-wave structures, to achieve constitutive metacells with continuous phase shifting and stable reflectivity. An LC-facilitated reconfigurable metasurface sub-system containing more than 2300 metacells is realized with its unprecedented comprehensive wavefront manipulation capacity validated through various beamforming functions, including beam focusing/steering, reconfigurable vortex beams, and tunable holograms, demonstrating a milli-second-level function-switching speed. The proposed methodology offers a paradigm shift for modulating electromagnetic waves in a non-resonating broadband fashion with fast-response and low-cost properties by exploiting functionalized LC-enabled metasurfaces. Moreover, this extremely agile metasurface-enabled antenna technology will facilitate a transformative impact on communication/sensing systems and empower new possibilities for wavefront engineering and diffractive wave calculation/inference.

Light-Induced Conductance Potentiation and Depression in an All-Optically Controlled Memristor

Optoelectronic memristors are new multifunctional devices with both electrically tunable and light-tunable synaptic plasticity, attracting great attention as key promising devices for optoelectronic neuromorphic computing systems. At present, the conductance modulation in most optoelectronic memristors is conducted in a hybrid photoelectric mode, suffering some problems such as heat generation and control complexity. Here, an optoelectronic memristor based on the p+-Si/n-ZnO heterojunction is proposed where the conductance can be reversibly modulated in an all-optically controlled mode. The electron detrapping/trapping mechanism at the p+-Si/n-ZnO interface barrier region is presented to explain the light-induced conductance potentiation/depression behavior. Furthermore, some synaptic functions, including excitatory postsynaptic current (EPSC), inhibitory postsynaptic current (IPSC), and paired-pulse facilitation (PPF), are successfully mimicked in the p+-Si/n-ZnO heterojunction memristor, instructing its application potential for optoelectronic neuromorphic computing.

All-optical complex field imaging using diffractive processors

Complex field imaging, which captures both the amplitude and phase information of input optical fields or objects, can offer rich structural insights into samples, such as their absorption and refractive index distributions. However, conventional image sensors are intensity-based and inherently lack the capability to directly measure the phase distribution of a field. This limitation can be overcome using interferometric or holographic methods, often supplemented by iterative phase retrieval algorithms, leading to a considerable increase in hardware complexity and computational demand. Here, we present a complex field imager design that enables snapshot imaging of both the amplitude and quantitative phase information of input fields using an intensity-based sensor array without any digital processing. Our design utilizes successive deep learning-optimized diffractive surfaces that are structured to collectively modulate the input complex field, forming two independent imaging channels that perform amplitude-to-amplitude and phase-to-intensity transformations between the input and output planes within a compact optical design, axially spanning ~100 wavelengths. The intensity distributions of the output fields at these two channels on the sensor plane directly correspond to the amplitude and quantitative phase profiles of the input complex field, eliminating the need for any digital image reconstruction algorithms. We experimentally validated the efficacy of our complex field diffractive imager designs through 3D-printed prototypes operating at the terahertz spectrum, with the output amplitude and phase channel images closely aligning with our numerical simulations. We envision that this complex field imager will have various applications in security, biomedical imaging, sensing and material science, among others.

Bidirectional Optical Neural Networks Based on Free-Space Optics Using Lens Arrays and Spatial Light Modulator

This paper introduces a novel architecture-bidirectional optical neural network (BONN)-for providing backward connections alongside forward connections in artificial neural networks (ANNs). BONN incorporates laser diodes and photodiodes and exploits the properties of Köhler illumination to establish optical channels for backward directions. Thus, it has bidirectional functionality that is crucial for algorithms such as the backpropagation algorithm. BONN has a scaling limit of 96 × 96 for input and output arrays, and a throughput of 8.5 × 1015 MAC/s. While BONN's throughput may rise with additional layers for continuous input, limitations emerge in the backpropagation algorithm, as its throughput does not scale with layer count. The successful BONN-based implementation of the backpropagation algorithm requires the development of a fast spatial light modulator to accommodate frequent data flow changes. A two-mirror-like BONN and its cascaded extension are alternatives for multilayer emulation, and they help save hardware space and increase the parallel throughput for inference. An investigation into the application of the clustering technique to BONN revealed its potential to help overcome scaling limits and to provide full interconnections for backward directions between doubled input and output ports. BONN's bidirectional nature holds promise for enhancing supervised learning in ANNs and increasing hardware compactness.

Tunable-bias based optical neural network for reinforcement learning in path planning

Owing to the high integration, reconfiguration and strong robustness, Mach-Zehnder interferometers (MZIs) based optical neural networks (ONNs) have been widely considered. However, there are few works adding bias, which is important for neural networks, into the ONNs and systematically studying its effect. In this article, we propose a tunable-bias based optical neural network (TBONN) with one unitary matrix layer, which can improve the utilization rate of the MZIs, increase the trainable weights of the network and has more powerful representational capacity than traditional ONNs. By systematically studying its underlying mechanism and characteristics, we demonstrate that TBONN can achieve higher performance by adding more optical biases to the same side beside the inputted signals. For the two-dimensional dataset, the average prediction accuracy of TBONN with 2 biases (97.1%) is 5% higher than that of TBONN with 0 biases (92.1%). Additionally, utilizing TBONN, we propose a novel optical deep Q network (ODQN) algorithm to complete path planning tasks. By implementing simulated experiments, our ODQN shows competitive performance compared with the conventional deep Q network, but accelerates the computation speed by 2.5 times and 4.5 times for 2D and 3D grid worlds, respectively. Further, a more noticeable acceleration will be obtained when applying TBONN to more complex tasks. Also, we demonstrate the strong robustness of TBONN and the imprecision elimination method by using on-chip training.

Experimental reservoir computing with diffractively coupled VCSELs

We present experiments on reservoir computing (RC) using a network of vertical-cavity surface-emitting lasers (VCSELs) that we diffractively couple via an external cavity. Our optical reservoir computer consists of 24 physical VCSEL nodes. We evaluate the system's memory and solve the 2-bit XOR task and the 3-bit header recognition (HR) task with bit error ratios (BERs) below 1% and the 2-bit digital-to-analog conversion (DAC) task with a root mean square error (RMSE) of 0.067.

Monolayer directional metasurface for all-optical image classifier doublet

Diffractive deep neural networks, known for their passivity, high scalability, and high efficiency, offer great potential in holographic imaging, target recognition, and object classification. However, previous endeavors have been hampered by spatial size and alignment. To address these issues, this study introduces a monolayer directional metasurface, aimed at reducing spatial constraints and mitigating alignment issues. Utilizing this methodology, we use MNIST datasets to train diffractive deep neural networks and realize digital classification, revealing that the metasurface can achieve excellent digital image classification results, and the classification accuracy of ideal phase mask plates and metasurface for phase-only modulation can reach 84.73% and 84.85%, respectively. Despite a certain loss of degrees of freedom compared to multi-layer phase mask plates, the single-layer metasurface is easier to fabricate and align, thereby improving spatial utilization efficiency.

An optoacoustic field-programmable perceptron for recurrent neural networks

Recurrent neural networks (RNNs) can process contextual information such as time series signals and language. But their tracking of internal states is a limiting factor, motivating research on analog implementations in photonics. While photonic unidirectional feedforward neural networks (NNs) have demonstrated big leaps, bi-directional optical RNNs present a challenge: the need for a short-term memory that (i) programmable and coherently computes optical inputs, (ii) minimizes added noise, and (iii) allows scalability. Here, we experimentally demonstrate an optoacoustic recurrent operator (OREO) which meets (i, ii, iii). OREO contextualizes the information of an optical pulse sequence via acoustic waves. The acoustic waves link different optical pulses, capturing their information and using it to manipulate subsequent operations. OREO's all-optical control on a pulse-by-pulse basis offers simple reconfigurability and is used to implement a recurrent drop-out and pattern recognition of 27 optical pulse patterns. Finally, we introduce OREO as bi-directional perceptron for new classes of optical NNs.

Reconfigurable origami hologram based on deep neural networks

Reconfigurable and multifunctional metasurfaces are becoming indispensable in a variety of applications due to their capability to execute diverse functions across various states. However, many of these metasurfaces incorporate complex active components, thereby escalating structural complexity and bulk volume. In this research, we propose a reconfigurable passive hologram based solely on an origami structure, enabling the successful generation of holograms depicting the 'Z' and 'L' illuminated by a right-hand circular polarization (RHCP) wave in two distinct states: planar and zigzag configuration, respectively. The transformation between the 2D planar metasurface and the 3D zigzag structure with slant angles of 35 is achieved solely through mechanically stretching and compressing the origami metasurface. The phases on the origami metasurface are trained through a deep neural network which operates on the NVIDIA Tesla k80 GPU, with the total training process costing 11.88 s after 100 epochs. The reconfigurable scheme proposed in this research provides flexibility and ease of implementation in the fields of imaging and data processing.

Large-scale photonic chiplet Taichi empowers 160-TOPS/W artificial general intelligence

The pursuit of artificial general intelligence (AGI) continuously demands higher computing performance. Despite the superior processing speed and efficiency of integrated photonic circuits, their capacity and scalability are restricted by unavoidable errors, such that only simple tasks and shallow models are realized. To support modern AGIs, we designed Taichi-large-scale photonic chiplets based on an integrated diffractive-interference hybrid design and a general distributed computing architecture that has millions-of-neurons capability with 160-tera-operations per second per watt (TOPS/W) energy efficiency. Taichi experimentally achieved on-chip 1000-category-level classification (testing at 91.89% accuracy in the 1623-category Omniglot dataset) and high-fidelity artificial intelligence-generated content with up to two orders of magnitude of improvement in efficiency. Taichi paves the way for large-scale photonic computing and advanced tasks, further exploiting the flexibility and potential of photonics for modern AGI.

Intelligent electroactive material systems with self-adaptive mechanical memory and sequential logic

By synthesizing the requisite functionalities of intelligence in an integrated material system, it may become possible to animate otherwise inanimate matter. A significant challenge in this vision is to continually sense, process, and memorize information in a decentralized way. Here, we introduce an approach that enables all such functionalities in a soft mechanical material system. By integrating nonvolatile memory with continuous processing, we develop a sequential logic-based material design framework. Soft, conductive networks interconnect with embedded electroactive actuators to enable self-adaptive behavior that facilitates autonomous toggling and counting. The design principles are scaled in processing complexity and memory capacity to develop a model 8-bit mechanical material that can solve linear algebraic equations based on analog mechanical inputs. The resulting material system operates continually to monitor the current mechanical configuration and to autonomously search for solutions within a desired error. The methods created in this work are a foundation for future synthetic general intelligence that can empower materials to autonomously react to diverse stimuli in their environment.

Multichannel meta-imagers for accelerating machine vision

Rapid developments in machine vision technology have impacted a variety of applications, such as medical devices and autonomous driving systems. These achievements, however, typically necessitate digital neural networks with the downside of heavy computational requirements and consequent high energy consumption. As a result, real-time decision-making is hindered when computational resources are not readily accessible. Here we report a meta-imager designed to work together with a digital back end to offload computationally expensive convolution operations into high-speed, low-power optics. In this architecture, metasurfaces enable both angle and polarization multiplexing to create multiple information channels that perform positively and negatively valued convolution operations in a single shot. We use our meta-imager for object classification, achieving 98.6% accuracy in handwritten digits and 88.8% accuracy in fashion images. Owing to its compactness, high speed and low power consumption, our approach could find a wide range of applications in artificial intelligence and machine vision applications.

Broadband nonlinear modulation of incoherent light using a transparent optoelectronic neuron array

Nonlinear optical processing of ambient natural light is highly desired for computational imaging and sensing. Strong optical nonlinear response under weak broadband incoherent light is essential for this purpose. By merging 2D transparent phototransistors (TPTs) with liquid crystal (LC) modulators, we create an optoelectronic neuron array that allows self-amplitude modulation of spatially incoherent light, achieving a large nonlinear contrast over a broad spectrum at orders-of-magnitude lower intensity than achievable in most optical nonlinear materials. We fabricated a 10,000-pixel array of optoelectronic neurons, and experimentally demonstrated an intelligent imaging system that instantly attenuates intense glares while retaining the weaker-intensity objects captured by a cellphone camera. This intelligent glare-reduction is important for various imaging applications, including autonomous driving, machine vision, and security cameras. The rapid nonlinear processing of incoherent broadband light might also find applications in optical computing, where nonlinear activation functions for ambient light conditions are highly sought.

A programmable hybrid digital chemical information processor based on the Belousov-Zhabotinsky reaction

The exponential growth of the power of modern digital computers is based upon the miniaturization of vast nanoscale arrays of electronic switches, but this will be eventually constrained by fabrication limits and power dissipation. Chemical processes have the potential to scale beyond these limits by performing computations through chemical reactions, yet the lack of well-defined programmability limits their scalability and performance. Here, we present a hybrid digitally programmable chemical array as a probabilistic computational machine that uses chemical oscillators using Belousov-Zhabotinsky reaction partitioned in interconnected cells as a computational substrate. This hybrid architecture performs efficient computation by distributing information between chemical and digital domains together with inbuilt error correction logic. The efficiency is gained by combining digital logic with probabilistic chemical logic based on nearest neighbour interactions and hysteresis effects. We demonstrated the computational capabilities of our hybrid processor by implementing one- and two-dimensional Chemical Cellular Automata demonstrating emergent dynamics of life-like entities called Chemits. Additionally, we demonstrate hybrid probabilistic logic as a viable logic for solving combinatorial optimization problems.

Cross-layer transmission realized by light-emitting memristor for constructing ultra-deep neural network with transfer learning ability

Deep neural networks have revolutionized several domains, including autonomous driving, cancer detection, and drug design, and are the foundation for massive artificial intelligence models. However, hardware neural network reports still mainly focus on shallow networks (2 to 5 layers). Implementing deep neural networks in hardware is challenging due to the layer-by-layer structure, resulting in long training times, signal interference, and low accuracy due to gradient explosion/vanishing. Here, we utilize negative ultraviolet photoconductive light-emitting memristors with intrinsic parallelism and hardware-software co-design to achieve electrical information's optical cross-layer transmission. We propose a hybrid ultra-deep photoelectric neural network and an ultra-deep super-resolution reconstruction neural network using light-emitting memristors and cross-layer block, expanding the networks to 54 and 135 layers, respectively. Further, two networks enable transfer learning, approaching or surpassing software-designed networks in multi-dataset recognition and high-resolution restoration tasks. These proposed strategies show great potential for high-precision multifunctional hardware neural networks and edge artificial intelligence.

Towards ultrafast quantitative phase imaging via differentiable microscopy [Invited]

With applications ranging from metabolomics to histopathology, quantitative phase microscopy (QPM) is a powerful label-free imaging modality. Despite significant advances in fast multiplexed imaging sensors and deep-learning-based inverse solvers, the throughput of QPM is currently limited by the pixel-rate of the image sensors. Complementarily, to improve throughput further, here we propose to acquire images in a compressed form so that more information can be transferred beyond the existing hardware bottleneck of the image sensor. To this end, we present a numerical simulation of a learnable optical compression-decompression framework that learns content-specific features. The proposed differentiable quantitative phase microscopy (∂-QPM) first uses learnable optical processors as image compressors. The intensity representations produced by these optical processors are then captured by the imaging sensor. Finally, a reconstruction network running on a computer decompresses the QPM images post aquisition. In numerical experiments, the proposed system achieves compression of × 64 while maintaining the SSIM of ∼0.90 and PSNR of ∼30 dB on cells. The results demonstrated by our experiments open up a new pathway to QPM systems that may provide unprecedented throughput improvements.

Metasurface-Based Optical Logic Operators Driven by Diffractive Neural Networks

In this paper, a novel optical logic operator based on the multifunctional metasurface driven by all-optical diffractive neural network is reported, which can perform four principal quantum logic operations (Pauli-X, Pauli-Y, Pauli-Z, and Hadamard gates). The two ground states| 0 ⟩ $|0 \rangle $ and| 1 ⟩ $|1 \rangle $  are characterized by two orthogonal linear polarization states. The proposed spatial- and polarization-multiplexed all-optical diffractive neural network only contains a hidden layer physically mapped as a metasurface with simple and compact unit cells, which dramatically reduces the volume and computing resources required for the system. The designed optical quantum operator is proven to achieve high fidelities for all four quantum logical gates, up to 99.96% numerically and 99.88% experimentally. The solution will facilitate the construction of large-scale optical quantum computing systems and scalable optical quantum devices.

Technology and Integration Roadmap for Optoelectronic Memristor

Optoelectronic memristors (OMs) have emerged as a promising optoelectronic Neuromorphic computing paradigm, opening up new opportunities for neurosynaptic devices and optoelectronic systems. These OMs possess a range of desirable features including minimal crosstalk, high bandwidth, low power consumption, zero latency, and the ability to replicate crucial neurological functions such as vision and optical memory. By incorporating large-scale parallel synaptic structures, OMs are anticipated to greatly enhance high-performance and low-power in-memory computing, effectively overcoming the limitations of the von Neumann bottleneck. However, progress in this field necessitates a comprehensive understanding of suitable structures and techniques for integrating low-dimensional materials into optoelectronic integrated circuit platforms. This review aims to offer a comprehensive overview of the fundamental performance, mechanisms, design of structures, applications, and integration roadmap of optoelectronic synaptic memristors. By establishing connections between materials, multilayer optoelectronic memristor units, and monolithic optoelectronic integrated circuits, this review seeks to provide insights into emerging technologies and future prospects that are expected to drive innovation and widespread adoption in the near future.

PMONN: an optical neural network for photonic integrated circuits based on micro-resonator

We propose an improved optical neural network (ONN) circuit architecture based on conventional micro-resonator ONNs, called the Phase-based Micro-resonator Optical Neural Network (PMONN). PMONN's core architecture features a Convolutions and Batch Normalization (CB) unit, comprising a phase-based (PB) convolutional layer, a Depth-Point-Wise (DPW) convolutional layer, and a reconstructed Batch Normalization (RBN) layer. The PB convolution kernel uses modulable phase shifts of Add-drop MRRs as learnable parameters and their optical transfer function as convolution weights. The DPW convolution kernel amplifies PB convolution weights by learning the amplification factors. To address the internal covariate shift during training, the RBN layer normalizes DPW outputs by reconstructing the BN layer of the electronic neural network, which is then merged with the DPW layer in the test stage. We employ the tunable DAs in the architecture to implement the merged layer. PMONN achieves 99.15% and 91.83% accuracy on MNIST and Fashion-MNIST datasets, respectively. This work presents a method for implementing an optical neural network on the improved architecture based on MRRs and increases the flexibility and reusability of the architecture. PMONN has potential applications as the backbone for future optical object detection neural networks.

High fidelity laser beam shaping using liquid crystal on silicon spatial light modulators as diffractive neural networks

Spatial light modulators (SLMs) based on liquid crystal on silicon (LCoS) are powerful tools for laser beam shaping as they can be used to dynamically create almost arbitrary intensity distributions. However, laser beam shaping with LCoS-SLMs often suffers from beam shaping artifacts in part caused by unconsidered properties of the LCoS devices: astigmatism that stems from the non-normal incidence of the laser beam on the SLM and the effect commonly referred to as the '0-th diffraction order' that is caused by both the crosstalk between neighboring pixels and the direct reflection at the cover glass of the SLM. We here present a method to consider and compensate for these inherent properties of LCoS devices by treating the SLM as a diffractive neural network.

Optical authentication scheme based on all-optical neural network

Diffractive deep neural network is architectural designs based on the principles of neural networks, which consists of multiple diffraction layers and has the remarkable ability to perform machine learning tasks at the speed of light. In this paper, a novel optical authentication system was presented that utilizes the diffractive deep neural network principle. By carefully manipulating a light beam with both a public key and a private key, we are able to generate a unique and secure image representation at a precise distance. The generated image can undergo authentication by being processed through the proposed authentication system. Leveraging the utilization of invisible terahertz light, the certification system possesses inherent characteristics of concealment and enhanced security. Additionally, the entire certification process operates solely through the manipulation of the light beam, eliminating the need for electronic calculations. As a result, the system offers rapid certification speed. The proposed optical authentication scheme is further validated through computer simulations, which showcase its robust security and high precision. This method holds immense potential for diverse applications in optical neural network authentication, warranting a broad scope of future prospects.

Photonic neuromorphic architecture for tens-of-task lifelong learning

Scalable, high-capacity, and low-power computing architecture is the primary assurance for increasingly manifold and large-scale machine learning tasks. Traditional electronic artificial agents by conventional power-hungry processors have faced the issues of energy and scaling walls, hindering them from the sustainable performance improvement and iterative multi-task learning. Referring to another modality of light, photonic computing has been progressively applied in high-efficient neuromorphic systems. Here, we innovate a reconfigurable lifelong-learning optical neural network (L2ONN), for highly-integrated tens-of-task machine intelligence with elaborated algorithm-hardware co-design. Benefiting from the inherent sparsity and parallelism in massive photonic connections, L2ONN learns each single task by adaptively activating sparse photonic neuron connections in the coherent light field, while incrementally acquiring expertise on various tasks by gradually enlarging the activation. The multi-task optical features are parallelly processed by multi-spectrum representations allocated with different wavelengths. Extensive evaluations on free-space and on-chip architectures confirm that for the first time, L2ONN avoided the catastrophic forgetting issue of photonic computing, owning versatile skills on challenging tens-of-tasks (vision classification, voice recognition, medical diagnosis, etc.) with a single model. Particularly, L2ONN achieves more than an order of magnitude higher efficiency than the representative electronic artificial neural networks, and 14× larger capacity than existing optical neural networks while maintaining competitive performance on each individual task. The proposed photonic neuromorphic architecture points out a new form of lifelong learning scheme, permitting terminal/edge AI systems with light-speed efficiency and unprecedented scalability.

Diffractive optical computing in free space

Structured optical materials create new computing paradigms using photons, with transformative impact on various fields, including machine learning, computer vision, imaging, telecommunications, and sensing. This Perspective sheds light on the potential of free-space optical systems based on engineered surfaces for advancing optical computing. Manipulating light in unprecedented ways, emerging structured surfaces enable all-optical implementation of various mathematical functions and machine learning tasks. Diffractive networks, in particular, bring deep-learning principles into the design and operation of free-space optical systems to create new functionalities. Metasurfaces consisting of deeply subwavelength units are achieving exotic optical responses that provide independent control over different properties of light and can bring major advances in computational throughput and data-transfer bandwidth of free-space optical processors. Unlike integrated photonics-based optoelectronic systems that demand preprocessed inputs, free-space optical processors have direct access to all the optical degrees of freedom that carry information about an input scene/object without needing digital recovery or preprocessing of information. To realize the full potential of free-space optical computing architectures, diffractive surfaces and metasurfaces need to advance symbiotically and co-evolve in their designs, 3D fabrication/integration, cascadability, and computing accuracy to serve the needs of next-generation machine vision, computational imaging, mathematical computing, and telecommunication technologies.

Deep photonic network platform enabling arbitrary and broadband optical functionality

Expanding applications in optical communications, computing, and sensing continue to drive the need for high-performance integrated photonic components. Designing these on-chip systems with arbitrary functionality requires beyond what is possible with physical intuition, for which machine learning-based methods have recently become popular. However, computational demands for physically accurate device simulations present critical challenges, significantly limiting scalability and design flexibility of these methods. Here, we present a highly-scalable, physics-informed design platform for on-chip optical systems with arbitrary functionality, based on deep photonic networks of custom-designed Mach-Zehnder interferometers. Leveraging this platform, we demonstrate ultra-broadband power splitters and a spectral duplexer, each designed within two minutes. The devices exhibit state-of-the-art experimental performance with insertion losses below 0.66 dB, and 1-dB bandwidths exceeding 120 nm. This platform provides a tractable path towards systematic, large-scale photonic system design, enabling custom power, phase, and dispersion profiles for high-throughput communications, quantum information processing, and medical/biological sensing applications.

Optical ReLU-like activation function based on a semiconductor laser with optical injection

Artificial neural networks usually consist of successive linear multiply-accumulate operations and nonlinear activation functions. However, most optical neural networks only achieve the linear operation in the optical domain, while the optical implementation of activation function remains challenging. Here we present an optical ReLU-like activation function (with 180° rotation) based on a semiconductor laser subject to the optical injection in an experiment. The ReLU-like function is achieved in a broad regime above the Hopf bifurcation of the injection-locking diagram and is operated in the continuous-wave mode. In particular, the slope of the activation function is reconfigurable by tuning the frequency difference between the master laser and the slave laser.

Orbital angular momentum-mediated machine learning for high-accuracy mode-feature encoding

Machine learning with optical neural networks has featured unique advantages of the information processing including high speed, ultrawide bandwidths and low energy consumption because the optical dimensions (time, space, wavelength, and polarization) could be utilized to increase the degree of freedom. However, due to the lack of the capability to extract the information features in the orbital angular momentum (OAM) domain, the theoretically unlimited OAM states have never been exploited to represent the signal of the input/output nodes in the neural network model. Here, we demonstrate OAM-mediated machine learning with an all-optical convolutional neural network (CNN) based on Laguerre-Gaussian (LG) beam modes with diverse diffraction losses. The proposed CNN architecture is composed of a trainable OAM mode-dispersion impulse as a convolutional kernel for feature extraction, and deep-learning diffractive layers as a classifier. The resultant OAM mode-dispersion selectivity can be applied in information mode-feature encoding, leading to an accuracy as high as 97.2% for MNIST database through detecting the energy weighting coefficients of the encoded OAM modes, as well as a resistance to eavesdropping in point-to-point free-space transmission. Moreover, through extending the target encoded modes into multiplexed OAM states, we realize all-optical dimension reduction for anomaly detection with an accuracy of 85%. Our work provides a deep insight to the mechanism of machine learning with spatial modes basis, which can be further utilized to improve the performances of various machine-vision tasks by constructing the unsupervised learning-based auto-encoder.

All-optical image denoising using a diffractive visual processor

Image denoising, one of the essential inverse problems, targets to remove noise/artifacts from input images. In general, digital image denoising algorithms, executed on computers, present latency due to several iterations implemented in, e.g., graphics processing units (GPUs). While deep learning-enabled methods can operate non-iteratively, they also introduce latency and impose a significant computational burden, leading to increased power consumption. Here, we introduce an analog diffractive image denoiser to all-optically and non-iteratively clean various forms of noise and artifacts from input images - implemented at the speed of light propagation within a thin diffractive visual processor that axially spans <250 × λ, where λ is the wavelength of light. This all-optical image denoiser comprises passive transmissive layers optimized using deep learning to physically scatter the optical modes that represent various noise features, causing them to miss the output image Field-of-View (FoV) while retaining the object features of interest. Our results show that these diffractive denoisers can efficiently remove salt and pepper noise and image rendering-related spatial artifacts from input phase or intensity images while achieving an output power efficiency of ~30-40%. We experimentally demonstrated the effectiveness of this analog denoiser architecture using a 3D-printed diffractive visual processor operating at the terahertz spectrum. Owing to their speed, power-efficiency, and minimal computational overhead, all-optical diffractive denoisers can be transformative for various image display and projection systems, including, e.g., holographic displays.