Pith. sign in

REVIEW 4 major objections 4 minor 105 references

Large-scale artificial intelligence with 41 million nanophotonic neurons on a metasurface

T0 review · 4 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single untrained metasurface layer with 41 million meta-atoms can match deep networks like ResNet, ViT, and SAM on medical image tasks.

desk verdict A 41M-meta-atom random-projection metasurface that is a real hardware milestone, but the parity claim with deep models rests on non-standard baselines. read the letter →

arxiv 2504.20416 v1 pith:RAYLWTHR submitted 2025-04-29 physics.optics

Mingcheng Luo , Meirui Jiang , Bhavin J. Shastri , Nansen Zhou , Wenfei Guo , Jianmin Xiong , Dongliang Wang , Renjie Zhou
show 3 more authors
Chester Shu Qi Dou Chaoran Huang
This is my paper · ORCID
classification physics.optics
keywords opticalneuralnetworkneuromorphiccomputingmetasurfacerandomprojectiontangentkernelmachinevisionmedicalimageclassificationwholeslideanalysis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a passive optical layer can replace the representational work of a deep network. A single metasurface with 41 million randomly patterned meta-atoms projects an image into a high-dimensional optical feature space, and a compact digital readout of fewer than 10,000 trained weights turns that projection into a prediction. On several medical and video benchmarks, the system is reported to reach the accuracy of ResNet-50, ViT, and SAM while using over 1,000 times less computing time and energy than GPUs. The reason to care is that the optical layer itself is never trained; if the claim holds, scaling optical AI no longer requires fabricating programmable or precisely layer-aligned optical hardware.

What carries the argument

The central object is the metasurface random-projection layer: 41 million silicon meta-atoms with Gaussian-distributed phase and amplitude responses, used as an untrained, fixed optical transform. Each meta-atom acts as an optical neuron whose scattered secondary wavefront contributes to the field at the sensor plane; a Fourier lens then emphasizes high-frequency content, and the sensor's square-law detection supplies the network nonlinearity and spatial downsampling. The paper interprets the resulting wide layer through the neural tangent kernel: at tens of millions of neurons, a Gaussian-initialized network behaves like an infinitely wide network, so the untrained optical transform already provides a rich feature map and only the compact digital backend needs training.

What would settle it

Evaluate the same datasets with the meta-ONN and with publicly released ResNet-50, ViT, and SAM checkpoints under the benchmarks' standard metrics, including per-class AUC for ChestX-ray8; if the optical system falls materially short of those baselines, or if an end-to-end timing measurement does not reproduce the claimed speed and energy advantage, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a single-layer metasurface optical neural network can match deep, large-scale neural networks on real-world tasks. The authors experimentally realize a 10 mm2 silicon metasurface with 6,400 by 6,400 meta-atoms whose transmission coefficients are sampled from a Gaussian distribution and are never trained. Light from an encoded image propagates through the metasurface and a Fourier lens to a 600 by 800 sensor, producing a nonlinear, downsampled feature vector; a digital layer with 192 to 9,600 weights performs the final classification. Reported results include 99.3 percent on MNIST, 98.0 percent on COVID-19 chest X-rays, 85.4 percent average accuracy on NIH ChestX-ray8, 98.8 percent with an IoU of 0.61 on RSNA hemorrhage detection, 99.1 percent action accuracy on KTH video through a recurrent optical setup, and an AUC of 97.0 percent with an IoU of 0.60 on CAMELYON16 whole-slide images. On these tasks the system is said to be comparable to ResNet-50, ViT, and SAM, with an energy efficiency of 241 TOPS/W and a claimed reduction of over 1,000 times in computing time and energy versus GPUs.

Load-bearing premise

The central parity claim rests on the digital comparison models being trained and evaluated under fair, standard protocols, and on the reported accuracy numbers being measured with metrics that are comparable to published benchmarks for the same datasets.

Editorial extensions

If this is right

  • Because the optical layer is fixed, the same metasurface chip can be reused across tasks by retraining only a readout with fewer than 10,000 weights, compressing the digital model by factors of 10^5 to 10^6.
  • Gigapixel pathology becomes practical in the projected system: a whole-slide image can be analyzed in about a second per slide, compared with over an hour for a SAM-based pipeline.
  • The system can recover from physical perturbation: after a 10 micrometer misalignment of the metasurface, retraining the small digital layer restores accuracy in 234 milliseconds.
  • The optical layer can be embedded in recurrent architectures for video, reaching 99.1 percent action accuracy at a projected 1,968 frames per second.
  • Energy per operation can fall below one photon per complex-valued multiplication, enabling operation in low-illumination settings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the optical layer is truly a fixed random projection, a decisive scaling test is to increase the camera pixel count while holding the metasurface fixed: effective feature dimensionality should rise with readout resolution rather than with the 41 million meta-atom count, and accuracy should track the readout.
  • The same chip could serve as a universal analog feature extractor in front of any small trained model, which suggests benchmark suites that compare one shared optical front-end against learned front-ends across many tasks.
  • The reported throughput and speed figures partly rest on component-level projections stated in the paper, so the strongest test of the 1,000-times claim is an end-to-end timing measurement of the integrated optical path.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The manuscript reports a free-space optical neural network built from a single metasurface containing 41 million random Gaussian-phase silicon meta-atoms, a Fourier lens, and a camera, followed by a compact digital readout. The authors argue that this random projection acts as an infinitely wide kernel machine, and they present experimental results on MNIST, COVID-19 chest X-rays, NIH ChestX-ray8, RSNA intracranial hemorrhage detection, KTH action recognition, and CAMELYON16 whole-slide images. They claim that the meta-ONN matches or approaches the accuracy of ResNet-50, ViT, and SAM while reducing computing time and energy by orders of magnitude, and that the system is the first ONN to close the performance gap with large-scale deep learning models.

Significance. If the parity claims hold, this would be a notable advance in optical neuromorphic hardware: the 41-million-meta-atom scale, the universal kernel-machine framing, the broad set of real-world medical and video tasks, and the reported energy efficiency (241 TOPS/W including peripherals) are all significant. The paper also contains genuine experimental assets: a fabricated 10 mm^2 metasurface, quantitative phase imaging of the fabricated sample, per-task optical experiments, and a compact readout with only hundreds to thousands of trained weights. The RNN extension and the sub-photon-per-multiplication demonstration are interesting in their own right. However, the central significance claim rests on comparisons to in-house digital baselines and on non-standard evaluation metrics, so the strength of the contribution depends on whether those comparisons can be made rigorous.

major comments (4)
  1. [Section 2.2, NIH ChestX-ray8] The reported 'average accuracy' of 85.4% on the eight-label, heavily imbalanced NIH ChestX-ray8 dataset is not a standard benchmark metric; published evaluations use per-label mean AUC with patient-level splits. As presented, the comparison with ViT (85.8% accuracy) is not informative because average accuracy can be dominated by the majority label, and it is not comparable to published AUC results. Please re-evaluate with mean AUC, per-class AUC, and patient-level splitting, and compare against published ChestX-ray8 baselines.
  2. [Section 2.4, CAMELYON16] The SAM baseline for the CAMELYON16 segmentation comparison is applied as a zero-shot promptable segmenter without histology-specific fine-tuning, so an IoU of 0.63 is not a state-of-the-art pathology segmentation result. Moreover, the meta-ONN IoU of 0.60 is derived from patch-level probability maps rather than direct mask prediction, which is not an apples-to-apples comparison. Please benchmark against a supervised segmentation model trained on CAMELYON16 (e.g., a U-Net) or against published baselines, and report the standard tumor-detection AUC on the official test set.
  3. [Section 2.1, system description] The claim of '41 million photonic neurons' overstates the effective dimensionality of the feature map. The detected field has at most M = 480,000 camera pixels, and after the described sum pooling the feature dimension is at most M (then further downsampled). The rank of the random projection is bounded by the number of input modes and camera pixels, not by the number of meta-atoms, so the phrases '480,000 x 41 million weights' and '41 million independent neurons' are misleading. Please quantify the effective rank or revise these claims.
  4. [Section 2.1, NTK justification] The NTK-based argument is incorrect as stated. Neural tangent kernel theory describes the training dynamics of infinitely wide networks and their equivalence to kernel regression; it does not imply that random Gaussian-initialized weights are already close to the global minimum. The empirical training-free MNIST result is interesting, but the sentence 'Gaussian-initialized weights, even without training, are already close to the global minimum' misrepresents the cited literature. Please replace this with a correct random-feature/kernel argument or explicitly label the NTK discussion as an analogy rather than a proof.
minor comments (4)
  1. [Figures and cross-references] Section 2.1 refers to 'Fig.2c', 'Fig.2d', and 'Fig.2e' for the NTK eigenvalue and accuracy plots, but those panels are actually in Fig.1c-1e; the Fig.2 caption labels panels (c)-(g) as microscope images and dataset illustrations. Please correct all figure callouts.
  2. [Typos] There are several typographical errors: 'serval' should be 'several' in the Discussion, 'Ostu' should be 'Otsu' in the Fig.4a caption, 'COMS' should be 'CMOS' in the Discussion, and 'deceleration' should be 'declaration' in the competing interest statement.
  3. [Code availability] The Code availability statement says 'Accession codes will be available before publication'; for a hardware paper, providing the evaluation scripts, digital-backend training details, and analysis code at review time would substantially strengthen reproducibility.
  4. [Time/energy comparisons] The time and energy comparisons in Section 2.2 compare training only the compact digital readout (192 or 9,600 weights) against full training of ResNet-50 or ViT; the claimed '1.2e5 times' training-time reduction therefore conflates model-size compression with hardware speed. Please state explicitly that this is a comparison of readout training against full-model training, not an end-to-end system training-time comparison.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the random metasurface is fixed, only the small digital readout is trained, and headline results are external benchmark measurements.

full rationale

The claimed derivation chain is not circular. The metasurface transmission matrix is fabricated once with Gaussian-random meta-atoms and is not trained or fitted to any benchmark; the same physical chip is reused across tasks, and only the compact digital readout (fewer than 10,000 trained weights) is optimized on the camera-captured outputs. The reported accuracies are therefore genuine measurements against external datasets (COVID-19 Radiography, NIH ChestX-ray8, RSNA ICH, KTH, CAMELYON16), not quantities that are equal to their inputs by construction. The NTK discussion in Section 2.1 motivates the Gaussian design and provides a spectral analysis of the resulting random feature map, but the headline parity claims do not follow by construction from that simulation; the experimental benchmarks are independent evidence. Concerns about in-house trained ResNet/ViT/SAM baselines, the non-standard 'average accuracy' metric on the multi-label ChestX-ray8 dataset, and SAM's zero-shot application to CAMELYON16 are benchmark-validity issues rather than circular reductions. There is no load-bearing self-citation chain, no imported uniqueness theorem, and no fitted parameter renamed as a prediction. Accordingly, no specific circular step can be quoted and exhibited, and the circularity score is 0.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The empirical parity claims rest on external benchmarks (MNIST, KTH, COVID-19 Radiography, ChestX-ray8, RSNA ICH, CAMELYON16), which keeps circularity low. The scaling and energy claims rest on a counting convention (41M independent neurons, scattering events as MACs) that is not justified, on task-tuned downsampling and weight counts, and on extrapolated component speeds. The NTK motivation in Section 2.1 misstates what NTK theorems prove. No new physical entities are invented; '41 million photonic neurons' is a re-labeling of physical meta-atoms.

free parameters (4)
  • Gaussian phase distribution width (sigma) = 0.4π
    The metasurface phase modulation is designed with standard deviation 0.4π to 'ensure a higher degree of randomness and thus support high performance' (Methods, Section 4). It is a hand-chosen design constant that sets the projection statistics and was not derived from first principles.
  • Per-task downsampling ratio of the 600x800 camera output = 12x16 (COVID), 15x20 (ICH), 30x40 (WSI), 34x45 (KTH)
    Downsampling ratios are 'task-dependent' (Section 2.2) and, together with the digital weight counts, set the feature dimension fed to the readout; these choices are tuned per benchmark in SI Notes 11-17 and directly affect the reported accuracies.
  • Per-task digital backend size = 192, 1,200/140, 3,000, 9,600, 9,180 weights
    The backend network is sized per task (Sections 2.2-2.4). The paper's claim of 'fewer than 10,000 parameters' holds by construction, but the specific sizes (140 to 9,600) are selected per benchmark rather than set by a general rule.
  • Input standardization to a fixed optical footprint = 512x512 SLM pixels over 4.1x4.1 mm²
    All inputs are rescaled (upsampled or minified) to a 512x512 pixel footprint (Methods). Because the metasurface is 3.2x3.2 mm², roughly 160k-262k independent input modes actually illuminate the chip; this normalization determines the effective input dimensionality of the optical projection.
assumptions (6)
  • standard math Random Gaussian projection yields linearly separable features for the target tasks (kernel-machine / ELM theory)
    Invoked in Section 2.1 ('random projection operates as a universal kernel machine'); the practical performance on each dataset is empirical.
  • domain assumption Square-law (intensity) detection at the camera provides the nonlinearity needed for the network
    Section 2.1: 'The image sensor array then applies a square function to the pooled signal, providing the nonlinearity of the NN.' If the nonlinearity were insufficient, the single-layer system would be linear and could not perform the reported classifications; this is asserted, not derived.
  • domain assumption The fabricated 41M meta-atoms realize the designed Gaussian transmission matrix
    Section 4: QPI measurement is shown only for a 36x36 µm² region (~72x72 meta-atoms); the statistics of the full 3.2x3.2 mm² chip are assumed. The design's tolerance of fabrication error is one of the paper's premises.
  • ad hoc to paper NTK convergence results imply that untrained Gaussian-initialized wide layers are near-optimal (as paraphrased in the paper)
    Section 2.1 states 'Gaussian-initialized weights, even without training, are already close to the global minimum.' The cited NTK literature (Jacot et al., Arora et al.) proves the kernel freezes during training, not that an untrained random network is at a loss minimum; the paper's conclusion is a stronger, unsupported reading.
  • standard math Scalar diffraction / Huygens-Fresnel propagation models the metasurface-to-sensor mapping
    Section 2.1 and Methods; the output equation uses the Fresnel diffraction operator H_d and a Fourier transform; standard physical optics.
  • standard math Fourier lens feature mapping improves high-frequency learnability (Tancik et al.)
    Section 2.1 cites Fourier-feature theory (Tancik 2020) and shows eigenvalue increases in simulations (Fig 2d); this is a borrowed, not re-derived, result.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Large-scale artificial intelligence with 41 million nanophotonic neurons on a metasurface." pith.science (2026). https://pith.science/paper/RAYLWTHR

@misc{pith2026250420416,
  author       = {Pith},
  title        = {Pith review of: Large-scale artificial intelligence with 41 million nanophotonic neurons on a metasurface},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RAYLWTHR}},
  note         = {Machine review of arXiv:2504.20416}
}
abstract

Conventional integrated circuits (ICs) struggle to meet the escalating demands of artificial intelligence (AI). This has sparked a renewed interest in an unconventional computing paradigm: neuromorphic (brain-inspired) computing. However, current neuromorphic systems face significant challenges in delivering a large number of parameters (i.e., weights) required for large-scale AI models. As a result, most neuromorphic hardware is limited to basic benchmark demonstrations, hindering its application to real-world AI challenges. Here, we present a large-scale optical neural network (ONN) for machine learning acceleration, featuring over 41 million photonic neurons. This system not only surpasses digital electronics in speed and energy efficiency but more importantly, closes the performance gap with large-scale AI models. Our ONN leverages an innovative optical metasurface device featuring numerous spatial modes. This device integrates over 41 million meta-atoms on a 10 mm$^2$ metasurface chip, enabling the processing of tens of millions of weights in a single operation. For the first time, we demonstrate that an ONN, utilizing a single-layer metasurface, can match the performance of deep and large-scale deep learning models, such as ResNet and Vision Transformer, across various benchmark tasks. Additionally, we show that our system can deliver high-performance solutions to real-world AI challenges through its unprecedented scale, such as accelerating the analysis of multi-gigapixel whole slide images (WSIs) for cancer detection by processing the million-pixel sub-image in a single shot. Our system reduces computing time and energy consumption by over 1,000 times compared to state-of-the-art graphic processing units (GPUs). This work presents a large-scale, low-power, and high-performance neuromorphic computing system, paving the way for future disruptive AI technologies.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

105 extracted references · 77 canonical work pages

  1. [1]

    nature 521(7553), 436–444 (2015)

    LeCun, Y., Bengio, Y., Hinton, G.: Deep learning. nature 521(7553), 436–444 (2015)

    2015

  2. [2]

    arXiv preprint arXiv:2303.08774 (2023)

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: Gpt-4 technical report. arXiv preprint arXiv:2303.08774 (2023)

    arXiv 2023

  3. [3]

    IEEE Journal of Selected Topics in Quantum Electronics 22(6), 194–203 (2016)

    Ahmed, A.H., Sharkia, A., Casper, B., Mirabbasi, S., Shekhar, S.: Siliconphoton- ics microring links for datacenters—challenges and opportunities. IEEE Journal of Selected Topics in Quantum Electronics 22(6), 194–203 (2016)

    2016

  4. [4]

    Proceedings of the IEEE 88(6), 728–749 (2000)

    Miller, D.A.: Rationale and challenges for optical interconnects to electronic chips. Proceedings of the IEEE 88(6), 728–749 (2000)

    2000

  5. [5]

    IEEE Journal of Selected Topics in Quantum Electronics 26(1), 1–18 (2019)

    Nahmias, M.A., De Lima, T.F., Tait, A.N., Peng, H.-T., Shastri, B.J., Prucnal, P.R.: Photonic multiply-accumulate operations for neural networks. IEEE Journal of Selected Topics in Quantum Electronics 26(1), 1–18 (2019)

    2019

  6. [6]

    In: 2017 IEEE Custom Integrated Circuits Conference (CICC), pp

    Sze, V., Chen, Y.-H., Emer, J., Suleiman, A., Zhang, Z.: Hardware for machine learning: Challenges and opportunities. In: 2017 IEEE Custom Integrated Circuits Conference (CICC), pp. 1–8 (2017)

    2017

  7. [7]

    Nature Photonics 15(2), 102–114 (2021)

    Shastri, B.J., Tait, A.N., Lima, T., Pernice, W.H., Bhaskaran, H., Wright, C.D., Prucnal, P.R.: Photonics for artificial intelligence and neuromorphic computing. Nature Photonics 15(2), 102–114 (2021)

    2021

  8. [8]

    Nature Reviews Physics 2(9), 499–510 (2020)

    Markovi´c, D., Mizrahi, A., Querlioz, D., Grollier, J.: Physics for neuromorphic computing. Nature Reviews Physics 2(9), 499–510 (2020)

    2020

Show all 105 references
  1. [9]

    Advances in Physics: X 7(1), 1981155 (2022)

    Huang, C., Sorger, V.J., Miscuglio, M., Al-Qadasi, M., Mukherjee, A., Lampe, L., Nichols, M., Tait, A.N., Lima, T., Marquez, B.A., et al.: Prospects and applications of photonic neural networks. Advances in Physics: X 7(1), 1981155 (2022)

    2022

  2. [10]

    Nature 588(7836), 39–47 (2020)

    Wetzstein, G., Ozcan, A., Gigan, S., Fan, S., Englund, D., Solja ˇci´c, M., Denz, C., Miller, D.A., Psaltis, D.: Inference in artificial intelligence with deep optics and photonics. Nature 588(7836), 39–47 (2020)

    2020

  3. [11]

    Nature Reviews Physics 5(12), 717–734 (2023)

    McMahon, P.L.: The physics of optical computing. Nature Reviews Physics 5(12), 717–734 (2023)

    2023

  4. [12]

    Nanotechnology 32(1), 012002 (2020) 22

    Berggren, K., Xia, Q., Likharev, K.K., Strukov, D.B., Jiang, H., Mikolajick, T., Querlioz, D., Salinga, M., Erickson, J.R., Pi, S., et al.: Roadmap on emerging hardware and technology for machine learning. Nanotechnology 32(1), 012002 (2020) 22

    2020

  5. [13]

    Light: Science & Applications 11(1), 30 (2022)

    Zhou, H., Dong, J., Cheng, J., Dong, W., Huang, C., Shen, Y., Zhang, Q., Gu,M., Qian, C., Chen, H., et al.: Photonic matrix multiplication lights up photonicaccelera- tor and beyond. Light: Science & Applications 11(1), 30 (2022)

    2022

  6. [14]

    Applied optics 24(10), 1469–1475 (1985)

    Farhat, N.H., Psaltis, D., Prata, A., Paek, E.: Optical implementation of thehop- field model. Applied optics 24(10), 1469–1475 (1985)

    1985

  7. [15]

    Nature 586(7828),207–216 (2020)

    Bogaerts, W., P ´erez, D., Capmany, J., Miller, D.A., Poon, J., Englund, D.,Morichetti, F., Melloni, A.: Programmable photonic circuits. Nature 586(7828),207–216 (2020)

    2020

  8. [16]

    Scientific reports 7(1), 7430 (2017)

    Tait, A.N., De Lima, T.F., Zhou, E., Wu, A.X., Nahmias, M.A., Shastri, B.J.,Prucnal, P.R.: Neuromorphic photonic networks using silicon photonic weight- banks. Scientific reports 7(1), 7430 (2017)

    2017

  9. [17]

    Nature Electronics 4(11), 837–844(2021)

    Huang, C., Fujisawa, S., Lima, T.F., Tait, A.N., Blow, E.C., Tian, Y., Bilodeau,S., Jha, A., Yaman, F., Peng, H.-T., et al.: A silicon photonic–electronic neuralnetwork for fibre nonlinearity compensation. Nature Electronics 4(11), 837–844(2021)

    2021

  10. [18]

    Nature photonics 11(7), 441–446 (2017)

    Shen, Y., Harris, N.C., Skirlo, S., Prabhu, M., Baehr-Jones, T., Hochberg, M.,Sun, X., Zhao, S., Larochelle, H., Englund, D., et al.: Deep learning with coher- entnanophotonic circuits. Nature photonics 11(7), 441–446 (2017)

    2017

  11. [19]

    Nanophotonics 9(13), 4221–4232 (2020)

    Stark, P., Horst, F., Dangel, R., Weiss, J., Offrein, B.J.: Opportunities forinte- grated photonic neural networks. Nanophotonics 9(13), 4221–4232 (2020)

    2020

  12. [20]

    Pappas, C., Moschos, T., Moralis-Pegios, M., Giamougiannis, G., Tsakyridis,A., Kirtas, M., Passalis, N., Tefas, A., Pleros, N.: A teraflop photonic matrixmultiplier using time-space-wavelength multiplexed awgr-based architectures.In: 2024 Optical Fiber Communications Conferenc...

    2024

  13. [21]

    Applied Sciences 11(13), 6232 (2021)

    De Marinis, L., Cococcioni, M., Liboiron-Ladouceur, O., Contestabile, G., Cas- toldi, P., Andriolli, N.: Photonic integrated reconfigurable linear processors asneural network accelerators. Applied Sciences 11(13), 6232 (2021)

    2021

  14. [22]

    Donati, G., Mirasso, C.R., Mancinelli, M., Pavesi, L., Argyris, A.: Microringres- onators with external optical feedback for time delay reservoir computing.Optics Express 30(1), 522–537 (2022)

    2022

  15. [23]

    Optical Materials Express 12(6), 2395–2414 (2022)

    Skalli, A., Robertson, J., Owen-Newns, D., Hejda, M., Porte, X., Reitzenstein,S., Hurtado, A., Brunner, D.: Photonic neuromorphic computing using verticalcavity semiconductor lasers. Optical Materials Express 12(6), 2395–2414 (2022)

    2022

  16. [24]

    Communications Engineering 1(1), 24 (2022)

    Sozos, K., Bogris, A., Bienstman, P., Sarantoglou, G., Deligiannidis, S., Mesar- itakis, C.: High-speed photonic neuromorphic computing using recurrent optical 23 spectrum slicing neural networks. Communications Engineering 1(1), 24 (2022)

    2022

  17. [25]

    Apl Photonics 6(12)(2021)

    Amin, R., George, J.K., Wang, H., Maiti, R., Ma, Z., Dalir, H., Khurgin, J.B.,Sorger, V.J.: An ito–graphene heterojunction integrated absorption modulator on si-photonics for neuromorphic nonlinear activation. Apl Photonics 6(12)(2021)

    2021

  18. [26]

    Nature 589(7840), 44–51 (2021)

    Xu, X., Tan, M., Corcoran, B., Wu, J., Boes, A., Nguyen, T.G., Chu, S.T.,Little, B.E., Hicks, D.G., Morandotti, R., et al.: 11 tops photonic convolutionalaccelerator for optical neural networks. Nature 589(7840), 44–51 (2021)

    2021

  19. [27]

    APL Photonics 7(1) (2022)

    Shi, B., Calabretta, N., Stabile, R.: Inp photonic integrated multi-layer neural- networks: Architecture and performance analysis. APL Photonics 7(1) (2022)

    2022

  20. [28]

    Nature communications 12(1), 96 (2021)

    Wu, C., Yu, H., Lee, S., Peng, R., Takeuchi, I., Li, M.: Programmable phasechange metasurfaces on waveguides for multimode photonic convolutional neuralnetwork. Nature communications 12(1), 96 (2021)

    2021

  21. [29]

    Nature Photonics 17(12),1080–1088 (2023)

    Dong, B., Aggarwal, S., Zhou, W., Ali, U.E., Farmakidis, N., Lee, J.S., He, Y.,Li, X., Kwong, D.-L., Wright, C., et al.: Higher-dimensional processing usinga photonic tensor core with continuous-time data. Nature Photonics 17(12),1080–1088 (2023)

    2023

  22. [30]

    Physical Review Applied 7(3),034013 (2017)

    Shainline, J.M., Buckley, S.M., Mirin, R.P., Nam, S.W.: Superconducting opto- electronic circuits for neuromorphic computing. Physical Review Applied 7(3),034013 (2017)

    2017

  23. [31]

    Apl Photonics 6(12) (2021)

    Xiao, X., On, M.B., Van Vaerenbergh, T., Liang, D., Beausoleil, R.G., Yoo, S.:Large-scale and energy-efficient tensorized optical neural networks on iii–v-onsilicon moscap platform. Apl Photonics 6(12) (2021)

    2021

  24. [32]

    NatureCommunications 14(1), 70 (2023)

    Fu, T., Zang, Y., Huang, Y., Du, Z., Huang, H., Hu, C., Chen, M., Yang, S.,Chen, H.: Photonic machine learning with on-chip diffractive optics. NatureCommunications 14(1), 70 (2023)

    2023

  25. [33]

    Frontiers of Optoelectronics 15(1), 15 (2022)

    Cheng, J., Zhao, Y., Zhang, W., Zhou, H., Huang, D., Zhu, Q., Guo, Y., Xu, B.,Dong, J., Zhang, X.: A small microring array that performs large complex- valuedmatrix-vector multiplication. Frontiers of Optoelectronics 15(1), 15 (2022)

    2022

  26. [34]

    Nature communications 13(1), 7970 (2022) 24

    Bai, B., Yang, Q., Shu, H., Chang, L., Yang, F., Shen, B., Tao, Z., Wang, J.,Xu, S., Xie, W., et al.: Microcomb-based integrated photonic processing unit.Nature Communications 14(1), 66 (2023) [35] Xu, S., Wang, J., Yi, S., Zou, W.: High-order tensor flow processing usinginteg...

    2023

  27. [36]

    Nature Communications 14(1), 3000 (2023)

    Meng, X., Zhang, G., Shi, N., Li, G., Aza˜na, J., Capmany, J., Yao, J., Shen, Y., Li, W., Zhu, N., et al.: Compact optical convolution processing unit based onmulti- mode interference. Nature Communications 14(1), 3000 (2023)

    2023

  28. [37]

    Nature 606(7914), 501–506 (2022)

    Ashtiani, F., Geers, A.J., Aflatouni, F.: An on-chip photonic deep neural net- workfor image classification. Nature 606(7914), 501–506 (2022)

    2022

  29. [38]

    Nature 589(7840), 52–58(2021)

    Feldmann, J., Youngblood, N., Karpov, M., Gehring, H., Li, X., Stappers, M.,Le Gallo, M., Fu, X., Lukashchuk, A., Raja, A.S., et al.: Parallel convolutionalprocessing using an integrated photonic tensor core. Nature 589(7840), 52–58(2021)

    2021

  30. [39]

    Nature 493(7431), 195–199 (2013)

    Sun, J., Timurdogan, E., Yaacobi, A., Hosseini, E.S., Watts, M.R.: Large- scalenanophotonic phased array. Nature 493(7431), 195–199 (2013)

    2013

  31. [40]

    Nature603(7900), 253–258 (2022)

    Zhang, X., Kwon, K., Henriksson, J., Luo, J., Wu, M.C.: A largescale microelectromechanical-systems-based silicon photonics lidar. Nature603(7900), 253–258 (2022)

    2022

  32. [41]

    Science361(6406), 1004–1008 (2018)

    Lin, X., Rivenson, Y., Yardimci, N.T., Veli, M., Luo, Y., Jarrahi, M., Ozcan,A.: All-optical machine learning using diffractive deep neural networks. Science361(6406), 1004–1008 (2018)

    2018

  33. [42]

    Nanophotonics 12(5), 777–793 (2023)

    Dinc, N.U., Saba, A., Madrid-Wolff, J., Gigli, C., Boniface, A., Moser, C., Psaltis,D.: From 3d to 2d and back again. Nanophotonics 12(5), 777–793 (2023)

    2023

  34. [43]

    arXiv preprint arXiv:2307.08533 (2023)

    Yildirim, M., Dinc, N.U., Oguz, I., Psaltis, D., Moser, C.: Nonlinear processing- with linear optics. arXiv preprint arXiv:2307.08533 (2023)

    2023 arXiv

  35. [44]

    Nature Photonics 17(5), 408–415 (2023)

    Wang, T., Sohoni, M.M., Wright, L.G., Stein, M.M., Ma, S.-Y., Onodera,T., Anderson, M.G., McMahon, P.L.: Image sensing with multilayer nonlinearoptical neural networks. Nature Photonics 17(5), 408–415 (2023)

    2023

  36. [45]

    Science Advances 9(25), 7904(2023)

    Bernstein, L., Sludds, A., Panuski, C., Trajtenberg-Mills, S., Hamerly, R.,Englund, D.: Single-shot optical neural network. Science Advances 9(25), 7904(2023)

    2023

  37. [46]

    Nature Photonics 15(5), 367–373 (2021)

    Zhou, T., Lin, X., Wu, J., Chen, Y., Xie, H., Li, Y., Fan, J., Wu, H., Fang, L., Dai,Q.: Large-scale neuromorphic optoelectronic computing with a reconfigurabled- iffractive processing unit. Nature Photonics 15(5), 367–373 (2021)

    2021

  38. [47]

    Light: Science & Applications 10(1), 40(2021)

    Goi, E., Chen, X., Zhang, Q., Cumming, B.P., Schoenhardt, S., Luan, H.,Gu, M.: Nanoprinted high-neuron-density optical linear perceptrons performingnear-infrared inference on a cmos chip. Light: Science & Applications 10(1), 40(2021)

    2021

  39. [48]

    25 Light: Science & Applications 11(1), 158 (2022)

    Luo, X., Hu, Y., Ou, X., Li, X., Lai, J., Liu, N., Cheng, X., Pan, A., Duan,H.: Metasurface-enabled on-chip multiplexed diffractive neural networks in thevisible. 25 Light: Science & Applications 11(1), 158 (2022)

    2022

  40. [49]

    Science Advances 8(30), 6410 (2022)

    Zheng, H., Liu, Q., Zhou, Y., Kravchenko, I.I., Huo, Y., Valentine, J.: Meta- opticaccelerators for object classifiers. Science Advances 8(30), 6410 (2022)

    2022

  41. [50]

    Nature Nanotechnology,1–8 (2024)

    Zheng, H., Liu, Q., Kravchenko, I.I., Zhang, X., Huo, Y., Valentine, J.G.: Mul- tichannel meta-imagers for accelerating machine vision. Nature Nanotechnology,1–8 (2024)

    2024

  42. [51]

    Advances in neuralinforma- tion processing systems 30 (2017)

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N.,Kaiser, L., Polosukhin, I.: Attention is all you need. Advances in neuralinforma- tion processing systems 30 (2017)

    2017

  43. [52]

    arXivpreprint arXiv:2010.11929 (2020)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X.,Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: Animage is worth 16x16 words: Transformers for image recognition at scale. arXivpreprint arXiv:2010.11929 (2020)

    2020 arXiv

  44. [53]

    Optics express 26(10), 13148–13182 (2018)

    Su, V.-C., Chu, C.H., Sun, G., Tsai, D.P.: Advances in optical metasur- faces:fabrication and applications. Optics express 26(10), 13148–13182 (2018)

    2018

  45. [54]

    Nature Reviews Materials 5(8), 604–620 (2020)

    Chen, W.T., Zhu, A.Y., Capasso, F.: Flat optics with dispersion- engineeredmetasurfaces. Nature Reviews Materials 5(8), 604–620 (2020)

    2020

  46. [55]

    Light: Science & Applications 7(1), 58 (2018)

    Neshev, D., Aharonovich, I.: Optical metasurfaces: new generation buildingblocks for multi-functional optics. Light: Science & Applications 7(1), 58 (2018)

    2018

  47. [56]

    Jour- nalof Optics 16(12), 123001 (2014)

    Zhao, Y., Liu, X.-X., Al‘u, A.: Recent advances on optical metasurfaces. Jour- nalof Optics 16(12), 123001 (2014)

    2014

  48. [57]

    Advanced Materials 32(3), 1805912 (2020)

    Chen, S., Liu, W., Li, Z., Cheng, H., Tian, J.: Metasurface-empowered optical- multiplexing and multifunction. Advanced Materials 32(3), 1805912 (2020)

    2020

  49. [58]

    IEEE Transactions onPattern Analysis and Machine Intelligence 44(10), 7128–7148 (2021)

    Liu, F., Huang, X., Chen, Y., Suykens, J.A.: Random features for kernel approx- imation: A survey on algorithms, theory, and beyond. IEEE Transactions onPattern Analysis and Machine Intelligence 44(10), 7128–7148 (2021)

    2021

  50. [59]

    Rahimi, A., Recht, B.: Random features for large-scale kernel machines.Advances in neural information processing systems 20 (2007)

    2007

  51. [60]

    Engelberg, J., Zhou, C., Mazurski, N., Bar-David, J., Kristensen, A., Levy, U.:Near-ir wide-field-of-view huygens metalens for outdoor imaging applica- tions.Nanophotonics 9(2), 361–370 (2020) 26

    2020

  52. [61]

    Nano letters 19(12), 8673–8682 (2019)

    Park, J.-S., Zhang, S., She, A., Chen, W.T., Lin, P., Yousef, K.M., Cheng, J.-X.,Capasso, F.: All-glass, large metalens at visible wavelength using deep- ultravioletprojection lithography. Nano letters 19(12), 8673–8682 (2019)

    2019

  53. [62]

    Acs Photonics 5(11), 4497–4503(2018)

    Afridi, A., Canet-Ferrer, J., Philippet, L., Osmond, J., Berto, P., Quidant, R.:Electrically driven varifocal silicon metalens. Acs Photonics 5(11), 4497–4503(2018)

    2018

  54. [63]

    AcsPhotonics 4(3), 544–551 (2017)

    Zhou, Z., Li, J., Su, R., Yao, B., Fang, H., Li, K., Zhou, L., Liu, J., Stellinga,D., Reardon, C.P., et al.: Efficient silicon metasurfaces for visible light. AcsPhotonics 4(3), 544–551 (2017)

    2017

  55. [64]

    Advances in neural information processingsystems 31 (2018)

    Jacot, A., Gabriel, F., Hongler, C.: Neural tangent kernel: Convergence andgen- eralization in neural networks. Advances in neural information processingsystems 31 (2018)

    2018

  56. [65]

    Advances in neural information- processing systems 9 (1996)

    Williams, C.: Computing with infinite networks. Advances in neural information- processing systems 9 (1996)

    1996

  57. [66]

    arX- ivpreprint arXiv:2011.14522 (2020)

    Yang, G., Hu, E.J.: Feature learning in infinite-width neural networks. arX- ivpreprint arXiv:2011.14522 (2020)

    2020 arXiv

  58. [67]

    Advances in neural informationprocessing systems 32 (2019)

    Arora, S., Du, S.S., Hu, W., Li, Z., Salakhutdinov, R.R., Wang, R.: On exactcom- putation with an infinitely wide neural net. Advances in neural informationprocessing systems 32 (2019)

    2019

  59. [68]

    In:International Conference on Machine Learning, pp

    Arora, S., Du, S., Hu, W., Li, Z., Wang, R.: Fine-grained analysis of optimization and generalization for overparameterized two-layer neural networks. In:International Conference on Machine Learning, pp. 322–332 (2019). PMLR

    2019

  60. [69]

    Advancesin Neural Information Processing Systems 32 (2019)

    Bietti, A., Mairal, J.: On the inductive bias of neural tangent kernels. Advancesin Neural Information Processing Systems 32 (2019)

    2019

  61. [70]

    Advances in neuralinformation processing systems 33, 7537–7547 (2020)

    Tancik, M., Srinivasan, P., Mildenhall, B., Fridovich-Keil, S., Raghavan, N.,Singhal, U., Ramamoorthi, R., Barron, J., Ng, R.: Fourier features let net- workslearn high frequency functions in low dimensional domains. Advances in neuralinformation processing systems 33, 7537–75...

    2020

  62. [71]

    Nanophotonics 9(14), 4327–4335(2020)

    Zhu, W., Yang, R., Geng, G., Fan, Y., Guo, X., Li, P., Fu, Q., Zhang, F., Gu,C., Li, J.: Titanium dioxide metasurface manipulating high-efficiency and broadband photonic spin hall effect in visible regime. Nanophotonics 9(14), 4327–4335(2020)

    2020

  63. [72]

    Naturecommunica- tions 11(1), 1864 (2020) 27

    Yang, W., Xiao, S., Song, Q., Liu, Y., Wu, Y., Wang, S., Yu, J., Han, J., Tsai,D.-P.: All-dielectric metasurface for high-performance structural color. Naturecommunica- tions 11(1), 1864 (2020) 27

    2020

  64. [73]

    Proceed- ings ofthe National Academy of Sciences 113(38), 10473–10478 (2016)

    Devlin, R.C., Khorasaninejad, M., Chen, W.T., Oh, J., Capasso, F.: Broadbandhigh-efficiency dielectric metasurfaces for the visible spectrum. Proceed- ings ofthe National Academy of Sciences 113(38), 10473–10478 (2016)

    2016

  65. [74]

    In: 2016 IEEE International Conferenceon Acoustics, Speech and Signal Processing (ICASSP), pp

    Saade, A., Caltagirone, F., Carron, I., Daudet, L., Dr ´emeau, A., Gigan, S.,Krzakala, F.: Random projections through multiple optical scattering: Approx- imating kernels at the speed of light. In: 2016 IEEE International Conferenceon Acoustics, Speech and Signal Processing (I...

    2016

  66. [75]

    Nature Photon- ics18(10), 1067–1075 (2024)

    Xia, F., Kim, K., Eliezer, Y., Han, S., Shaughnessy, L., Gigan, S., Cao, H.: Nonlinear optical encoding enabled by recurrent linear scattering. Nature Photon- ics18(10), 1067–1075 (2024)

    2024

  67. [76]

    Oguz, I., Hsieh, J.-L., Dinc, N.U., Te ˘gin, U., Yildirim, M., Gigli, C., Moser, C.,Psaltis, D.: Programming Nonlinear Propagation for Efficient Optical Learning- Machines (2022)

    2022

  68. [77]

    Nature Machine Intelligence1(11), 530–537 (2019)

    Antonik, P., Marsal, N., Brunner, D., Rontani, D.: Human action recognitionwith a large-scale brain-inspired photonic computer. Nature Machine Intelligence1(11), 530–537 (2019)

    2019

  69. [78]

    Nature Computational Science 4(5), 11–21 (2024)

    Hao, W., Jiaqi, H., Andrea, M., Alfonso, N., Fei, X., Xuanchen, L., Romolo,S., Qiang, L., Rachel, G., Sylvain, G.: Large-scale photonic computing withnonlinear disordered media. Nature Computational Science 4(5), 11–21 (2024)

    2024

  70. [79]

    Nature Computational Science 1(8), 542–549 (2021)

    Te ˘gin, U., Yıldırım, M., O ˘guz, I., Moser, C., Psaltis, D.: Scalable optical learning ˙operator. Nature Computational Science 1(8), 542–549 (2021)

    2021

  71. [80]

    Photonics Research 9(8), 1446–1454 (2021)

    Pierangeli, D., Marcucci, G., Conti, C.: Photonic extreme learning machine byfree-space optical propagation. Photonics Research 9(8), 1446–1454 (2021)

    2021

  72. [81]

    Nanophotonics 6(3), 561–576 (2017)

    Sande, G., Brunner, D., Soriano, M.C.: Advances in photonic reservoir comput- ing. Nanophotonics 6(3), 561–576 (2017)

    2017

  73. [82]

    Naturecommunications 4(1), 1364 (2013)

    Brunner, D., Soriano, M.C., Mirasso, C.R., Fischer, I.: Parallel photonic information processing at gigabyte per second data rates using transient states. Naturecommunications 4(1), 1364 (2013)

    2013

  74. [83]

    Nature communications 2(1), 468 (2011)

    Appeltant, L., Soriano, M.C., Sande, G., Danckaert, J., Massar, S., Dambre, J.,Schrauwen, B., Mirasso, C.R., Fischer, I.: Information processing using a singledy- namical node as complex system. Nature communications 2(1), 468 (2011)

    2011

  75. [84]

    Naturecommunications 28 5(1), 3541 (2014)

    Vandoorne, K., Mechet, P., Van Vaerenbergh, T., Fiers, M., Morthier, G.,Verstraeten, D., Schrauwen, B., Dambre, J., Bienstman, P.: Experimentaldemon- stration of reservoir computing on a silicon photonics chip. Naturecommunications 28 5(1), 3541 (2014)

    2014

  76. [85]

    Optica 5(6), 756–760 (2018)

    Bueno, J., Maktoobi, S., Froehly, L., Fischer, I., Jacquot, M., Larger, L.,Brunner, D.: Reinforcement learning in a large-scale photonic recurrent neuralnetwork. Optica 5(6), 756–760 (2018)

    2018

  77. [86]

    Journal of Machine Learning Research25(184), 1–49 (2024)

    Belfer, Y., Geifman, A., Galun, M., Basri, R.: Spectral analysis of the neuraltan- gent kernel for deep residual networks. Journal of Machine Learning Research25(184), 1–49 (2024)

    2024

  78. [87]

    Muminov, B., Vuong, L.T.: Fourier optical preprocessing in lieu of deep learn- ing.Optica 7(9), 1079–1088 (2020)

    2020

  79. [88]

    In: 2015 Fourth Interna- tional Conference on Future Generation Communication Technology (FGCT),pp

    Mironovova, M., B ´ıla, J.: Fast fourier transform for feature extraction and neuralnetwork for classification of electrocardiogram signals. In: 2015 Fourth Interna- tional Conference on Future Generation Communication Technology (FGCT),pp. 1–6 (2015). IEEE

    2015

  80. [89]

    In: Proceedings of the IEEE Conference on Computer Vision and PatternRecognition, pp

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and PatternRecognition, pp. 770–778 (2016)

    2016

  81. [90]

    In: Proceedingsof the IEEE/CVF International Conference on Computer Vision, pp

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T.,Whitehead, S., Berg, A.C., Lo, W.-Y., et al.: Segment anything. In: Proceedingsof the IEEE/CVF International Conference on Computer Vision, pp. 4015–4026(2023)

    2023

  82. [91]

    Computers in biology and medicine 132, 104319 (2021)

    Rahman, T., Khandakar, A., Qiblawey, Y., Tahir, A., Kiranyaz, S., Kashem,S.B.A., Islam, M.T., Al Maadeed, S., Zughaier, S.M., Khan, M.S., et al.: Exploring the effect of image enhancement techniques on covid-19 detection using chestx-ray images. Computers in biology and medici...

    2021

  83. [92]

    In: Proceedings of theIEEE Conference on Computer Vision and Pattern Recognition, pp

    Wang, X., Peng, Y., Lu, L., Lu, Z., Bagheri, M., Summers, R.M.: Chestx- ray8:Hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases. In: Proceedings of theIEEE Conference on Computer Vision and Pattern ...

    2017

  84. [93]

    Radiology: Artificial Intelligence 2(3), 190211(2020)

    Flanders, A.E., Prevedello, L.M., Shih, G., Halabi, S.S., Kalpathy-Cramer, J.,Ball, R., Mongan, J.T., Stein, A., Kitamura, F.C., Lungren, M.P., et al.: Con- struction of a machine learning dataset through collaboration: the rsna 2019brain ct hemorrhage challenge. Radiology: Ar...

    2020

  85. [94]

    ScienceAdvances 10(45), 0391 (2024) 29

    Wei, K., Li, X., Froech, J., Chakravarthula, P., Whitehead, J., Tseng, E., Majum- dar, A., Heide, F.: Spatially varying nanophotonic neural networks. ScienceAdvances 10(45), 0391 (2024) 29

    2024

  86. [95]

    Journal of Lightwave Technology 40(10), 3243–3254(2022)

    Moralis-Pegios, M., Mourgias-Alexandris, G., Tsakyridis, A., Giamougiannis,G., Totovic, A., Dabos, G., Passalis, N., Kirtas, M., Rutirawut, T., Gardes,F., et al.: Neuromorphic silicon photonics and hardware-aware deep learningfor high-speed inference. Journal of Lightwave Tech...

    2022

  87. [96]

    In: Proceedings of the 17th International Conference on PatternRecognition,

    Schuldt, C., Laptev, I., Caputo, B.: Recognizing human actions: a local svmap- proach. In: Proceedings of the 17th International Conference on PatternRecognition,

  88. [97]

    Gilbert, A., Illingworth, J., Bowden, R.: Action recognition using mined hierarchical compound features. IEEE Transactions on Pattern Analysis and MachineIntelligence 33(5), 883–897 (2010) [98] Grushin, A., Monner, D.D., Reggia, J.A., Mishra, A.: Robust human actionrecognition...

    2010

  89. [99]

    Nature medicine 25(8), 1301–1309 (2019)

    Campanella, G., Hanna, M.G., Geneslaw, L., Miraflor, A., Werneck Krauss Silva,V., Busam, K.J., Brogi, E., Reuter, V.E., Klimstra, D.S., Fuchs, T.J.: Clinical- grade computational pathology using weakly supervised deep learning on wholeslide images. Nature medicine 25(8), 1301–...

    2019

  90. [100]

    Jama 318(22), 2199–2210 (2017)

    Bejnordi, B.E., Veta, M., Van Diest, P.J., Van Ginneken, B., Karssemeijer, N.,Litjens, G., Van Der Laak, J.A., Hermsen, M., Manson, Q.F., Balkenhol, M.,et al.: Diagnostic assessment of deep learning algorithms for detection of lymphnode metastases in women with breast cancer. ...

    2017

  91. [101]

    Otsu, N., et al.: A threshold selection method from gray-level his- tograms.Automatica 11(285-296), 23–27 (1975)

    1975

  92. [102]

    Nature Photonics 17(8), 723–730 (2023)

    Chen, Z., Sludds, A., Davis III, R., Christen, I., Bernstein, L., Ateshian, L.,Heuser, T., Heermeier, N., Lott, J.A., Reitzenstein, S., et al.: Deep learning with- coherent vcsel neural networks. Nature Photonics 17(8), 723–730 (2023)

    2023

  93. [103]

    Nature Electron- ics3(11), 664–671 (2020)

    Zhou, F., Chai, Y.: Near-sensor and in-sensor computing. Nature Electron- ics3(11), 664–671 (2020)

    2020

  94. [104]

    Nature 623(7985), 48–57 (2023)

    Chen, Y., Nazhamaiti, M., Xu, H., Meng, Y., Zhou, T., Li, G., Fan, J., Wei, Q.,Wu, J., Qiao, F., et al.: All-analog photoelectronic chip for high-speed visiontasks. Nature 623(7985), 48–57 (2023)

    2023

  95. [105]

    Nano letters 18(3), 2124–2132 (2018) 30

    Paniagua-Dominguez, R., Yu, Y.F., Khaidarov, E., Choi, S., Leong, V., Bakker,R.M., Liang, X., Fu, Y.H., Valuckas, V., Krivitsky, L.A., et al.: A metalens witha near-unity numerical aperture. Nano letters 18(3), 2124–2132 (2018) 30

    2018

  96. [106]

    Laser & Photonics Reviews 16(10), 2100732 (2022) 31

    Qu, G., Cai, G., Sha, X., Chen, Q., Cheng, J., Zhang, Y., Han, J., Song, Q.,Xiao, S.: All-dielectric metasurface empowered optical-electronic hybrid neuralnetworks. Laser & Photonics Reviews 16(10), 2100732 (2022) 31

    2022

  97. [2004]

    ICPR 2004., vol. 3, pp. 32–36 (2004). IEEE

    2004

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.