pith. sign in

arxiv: 2604.02624 · v1 · submitted 2026-04-03 · ⚛️ physics.optics · cs.CV· cs.NE· physics.app-ph

Wavelength-multiplexed massively parallel diffractive optical information storage and image projection

Che-Yung Shen , Yuhang Li , Cagatay Isil , Jingxi Li , Leon Lenk , Tianyi Gan , Guangdong Ma , Fazil Onuralp Ardic
show 2 more authors
Mona Jarrahi Aydogan Ozcan
This is my paper

Pith reviewed 2026-05-13 19:03 UTC · model grok-4.3

classification ⚛️ physics.optics cs.CVcs.NEphysics.app-ph
keywords diffractive opticswavelength multiplexingoptical information storageimage projectiondeep learning optimizationdielectric surfacesvisible spectrummassively parallel readout
0
0 comments X

The pith

Deep learning designs diffractive surfaces that store and project over 4000 images each tied to a unique wavelength.

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

The paper presents a diffractive optical platform made of dielectric surfaces that are optimized at the wavelength scale using deep learning. Each surface set encodes thousands of distinct image patterns, with every pattern assigned to its own input wavelength so that illumination at that color projects only the corresponding image. Numerical simulations across the visible spectrum show the system can handle more than 4000 independent patterns inside one shared output field of view while keeping image quality high and crosstalk low. A two-layer experimental prototype confirms the principle by storing and projecting six separate patterns at six discrete wavelengths from 500 nm to 740 nm.

Core claim

Wavelength-multiplexed diffractive surfaces, structurally optimized by deep learning, can store and project thousands of independent images or patterns within the same output field of view, each pattern activated by its own illumination wavelength, achieving high fidelity and minimal spectral crosstalk as shown in visible-spectrum simulations and a six-channel proof-of-concept experiment.

What carries the argument

Wavelength-multiplexed diffractive dielectric surfaces optimized by deep learning, which encode multiple independent patterns into a single physical structure for selective readout by spectral illumination.

If this is right

  • The same optimized layers can be reused across different parts of the electromagnetic spectrum without redesign or material changes.
  • Storage capacity scales directly with the number of available wavelengths while keeping the physical device compact and static.
  • The architecture supports fast, parallel optical readout of large image sets without mechanical scanning.
  • High image fidelity at each channel makes the approach suitable for both data storage and projection applications.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Compact optical memories using this principle could replace moving-part systems in archival or security applications.
  • Integration with broadband light sources might enable wavelength-addressed multi-image projectors for displays or sensing.
  • The lack of required material dispersion engineering suggests the method can be adapted quickly to infrared or ultraviolet ranges using existing fabrication tools.

Load-bearing premise

The deep learning optimization of the diffractive surfaces will translate accurately to physical fabrication without significant degradation from manufacturing tolerances or unmodeled optical effects.

What would settle it

Fabricate the two-layer diffractive design and illuminate it sequentially at 500, 548, 596, 644, 692, and 740 nm; if the projected images fail to match the six target patterns with low crosstalk, the central claim is falsified.

read the original abstract

We introduce a wavelength-multiplexed massively parallel diffractive information storage platform composed of dielectric surfaces that are structurally optimized at the wavelength scale using deep learning to store and project thousands of distinct image patterns, each assigned to a unique wavelength. Through numerical simulations in the visible spectrum, we demonstrated that our wavelength-multiplexed diffractive system can store and project over 4,000 independent desired images/patterns within its output field-of-view, with high image quality and minimal crosstalk between spectral channels. Furthermore, in a proof-of-concept experiment, we demonstrated a two-layer diffractive design that stored six distinct patterns and projected them onto the same output field of view at six different wavelengths (500, 548, 596, 644, 692, and 740 nm). This diffractive architecture is scalable and can operate at various parts of the electromagnetic spectrum without the need for material dispersion engineering or redesigning its optimized diffractive layers. The demonstrated storage capacity, reconstruction image fidelity, and wavelength-encoded massively parallel read-out of our diffractive platform offer a compact and fast-access solution for large-scale optical information storage, image projection applications.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces a wavelength-multiplexed diffractive optical platform using deep learning to optimize dielectric surfaces for storing and projecting thousands of distinct image patterns, each tied to a unique wavelength. Numerical simulations in the visible spectrum claim storage and projection of over 4,000 independent images with high fidelity and minimal crosstalk. A proof-of-concept two-layer experiment demonstrates six patterns projected at wavelengths 500, 548, 596, 644, 692, and 740 nm onto the same field of view. The architecture is described as scalable across the electromagnetic spectrum without material dispersion engineering.

Significance. If the simulation results hold under rigorous validation, this diffractive approach could enable compact, passive, high-capacity optical information storage and massively parallel image projection. The reported capacity of over 4,000 channels and the experimental demonstration of wavelength-encoded readout represent potential advances for applications in data storage and display technologies. Credit is due for the scale of the numerical demonstration and the experimental proof-of-concept, though the absence of bounds comparisons and ablations limits assessment of generality.

major comments (3)
  1. Numerical Simulations (as described in the abstract): The claim of over 4,000 independent patterns with minimal crosstalk lacks any comparison to information-theoretic bounds based on the total number of phase pixels across layers and wavelengths, or ablation studies on training hyperparameters and random seeds. Without these, it is unclear if the headline capacity is a robust property of the architecture or an artifact of a specific optimization run.
  2. Experimental section: The two-layer proof-of-concept for six wavelengths provides only limited support for the scalability claim, as the manuscript lacks detailed methods, quantitative error analysis, full data on image fidelity, or crosstalk measurements.
  3. Abstract and discussion of fabrication: The assumption that deep learning-optimized designs will translate to physical fabrication without significant degradation from manufacturing tolerances or unmodeled optical effects is load-bearing for the scalability claims but is not addressed by any tolerance analysis or sensitivity study.
minor comments (2)
  1. The description of the deep learning optimization procedure (loss functions, network architecture, and training details) could be expanded for reproducibility.
  2. Figure captions and labels in the simulation results section would benefit from additional quantitative metrics (e.g., PSNR or SSIM values) to support the 'high image quality' claim.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed review. The comments highlight important aspects for strengthening the manuscript, particularly regarding robustness, experimental detail, and fabrication considerations. We address each major comment below and describe the revisions we will implement.

read point-by-point responses

  1. Referee: Numerical Simulations (as described in the abstract): The claim of over 4,000 independent patterns with minimal crosstalk lacks any comparison to information-theoretic bounds based on the total number of phase pixels across layers and wavelengths, or ablation studies on training hyperparameters and random seeds. Without these, it is unclear if the headline capacity is a robust property of the architecture or an artifact of a specific optimization run.

    Authors: We appreciate this point on assessing robustness. Precise information-theoretic bounds are nontrivial for diffractive systems because capacity emerges from continuous wave propagation and spectral multiplexing rather than discrete pixel counting alone. In the revised manuscript we will add an estimate of available degrees of freedom derived from the total number of phase pixels and spectral channels. We will also include ablation studies on key training hyperparameters together with results from multiple independent optimization runs using different random seeds, demonstrating that the reported capacity and low crosstalk are reproducible. These additions will appear in the main text and supplementary information. revision: yes

  2. Referee: Experimental section: The two-layer proof-of-concept for six wavelengths provides only limited support for the scalability claim, as the manuscript lacks detailed methods, quantitative error analysis, full data on image fidelity, or crosstalk measurements.

    Authors: We agree that additional experimental detail is required. The revised manuscript will expand the experimental section with complete fabrication and optical characterization methods, including the precise layer thicknesses, illumination conditions, and imaging setup. We will report quantitative fidelity metrics (PSNR and SSIM) for each projected pattern, include error analysis from repeated measurements, and provide explicit crosstalk values between the six channels. All raw data and analysis scripts will be supplied in the supplementary materials. revision: yes

  3. Referee: Abstract and discussion of fabrication: The assumption that deep learning-optimized designs will translate to physical fabrication without significant degradation from manufacturing tolerances or unmodeled optical effects is load-bearing for the scalability claims but is not addressed by any tolerance analysis or sensitivity study.

    Authors: This concern is well taken for the scalability discussion. Although the two-layer experiment provides initial physical validation, we will add a dedicated sensitivity analysis. The revised manuscript will include Monte Carlo simulations that introduce realistic fabrication perturbations (e.g., etch-depth and lateral-alignment errors) to the optimized phase profiles and quantify the resulting degradation in image fidelity and crosstalk. We will also briefly discuss the influence of unmodeled effects such as residual material dispersion. These results will be presented in a new subsection. revision: yes

Circularity Check

0 steps flagged

No circularity; claims rest on independent simulation and experiment

full rationale

The paper reports storage capacity and crosstalk performance from numerical simulations of a deep-learning-optimized diffractive structure and a separate two-layer experimental proof-of-concept. No load-bearing step equates the reported 4000-pattern capacity to a parameter fitted from the same target images, nor does any uniqueness theorem or ansatz reduce to a self-citation. The optimization objective and evaluation metrics are defined externally to the final performance numbers, so the derivation chain remains self-contained.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that deep learning can reliably optimize diffractive surfaces for independent wavelength-specific image projection without crosstalk, with free parameters including the number of layers and the specific wavelength assignments.

free parameters (2)
  • Number of diffractive layers
    Set to two for the proof-of-concept experiment; affects storage capacity and crosstalk.
  • Wavelength channel count and spacing
    Six wavelengths chosen for experiment and thousands for simulation; directly determines multiplexing capacity.
axioms (1)
  • domain assumption Deep learning optimization can produce dielectric surface patterns that achieve wavelength-specific image projection with minimal crosstalk
    Invoked throughout the description of the design process and results.

pith-pipeline@v0.9.0 · 5547 in / 1239 out tokens · 31718 ms · 2026-05-13T19:03:41.673163+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages

  1. [1]

    D. A. Parthenopoulos, P. M. Rentzepis, Three -Dimensional Optical Storage Memory. Science 245, 843–845 (1989)

    work page 1989

  2. [2]

    F. H. Mok, Angle-multiplexed storage of 5000 holograms in lithium niobate. Opt. Lett. 18, 915–917 (1993)

    work page 1993

  3. [3]

    J. F. Heanue, M. C. Bashaw, L. Hesselink, Volume Holographic Storage and Retrieval of Digital Data. Science 265, 749–752 (1994)

    work page 1994

  4. [4]

    Hesselink, S

    L. Hesselink, S. S. Orlov, M. C. Bashaw, Holographic data storage systems. Proc. IEEE 92, 1231– 1280 (2004)

    work page 2004

  5. [5]

    Tanaka, M

    K. Tanaka, M. Hara, K. Tokuyama, K. Hirooka, K. Ishioka, A. Fukumoto, K. Watanabe, Improved performance in coaxial holographic data recording. Opt. Express 15, 16196–16209 (2007)

    work page 2007

  6. [6]

    Project Silica: Towards Sustainable Cloud Archival Storage in Glass

    P. Anderson, E. B. Aranas, Y. Assaf, R. Behrendt, R. Black, M. Caballero, P. Cameron, B. Canakci, T. De Carvalho, A. Chatzieleftheriou, R. Storan Clarke, J. Clegg, D. Cletheroe, B. Cooper, T. Deegan, A. Donnelly, R. Drevinskas, A. Gaunt, C. Gkantsidis, A. Gomez Diaz, I. Haller, F. Hong, T. Ilieva, S. Joshi, R. Joyce, M. Kunkel, D. Lara, S. Legtchenko, F. ...

    work page doi:10.1145/3600006.3613208 2023

  7. [7]

    X. Ni, A. V. Kildishev, V. M. Shalaev, Metasurface holograms for visible light. Nat. Commun. 4, 2807 (2013)

    work page 2013

  8. [8]

    Huang, S

    L. Huang, S. Zhang, T. Zentgraf, Metasurface holography: from fundamentals to applications. Nanophotonics 7, 1169–1190 (2018)

    work page 2018

  9. [9]

    R. Zhao, L. Huang, Y. Wang, Recent advances in multi -dimensional metasurfaces holographic technologies. PhotoniX 1, 20 (2020)

    work page 2020

  10. [10]

    Xiong, Y

    B. Xiong, Y. Liu, Y. Xu, L. Deng, C. -W. Chen, J. -N. Wang, R. Peng, Y. Lai, Y. Liu, M. Wang, Breaking the limitation of polarization multiplexing in optical metasurfaces with engineered noise. Science 379, 294–299 (2023)

    work page 2023

  11. [11]

    Márquez, C

    A. Márquez, C. Li, A. Beléndez, S. A. Maier, H. Ren, Information multiplexing from optical holography to multi-channel metaholography. Nanophotonics 12, 4415–4440 (2023)

    work page 2023

  12. [12]

    J. Hu, D. Mengu, D. C. Tzarouchis, B. Edwards, N. Engheta, A. Ozcan, Diffractive optical computing in free space. Nat. Commun. 15, 1525 (2024)

    work page 2024

  13. [13]

    K.-S. Lee, D. -Y. Yang, S. H. Park, R. H. Kim, Recent developments in the use of two -photon polymerization in precise 2D and 3D microfabrications. Polym. Adv. Technol. 17, 72–82 (2006). 14

    work page 2006

  14. [14]

    Q. Geng, D. Wang, P. Chen, S. -C. Chen, Ultrafast multi -focus 3-D nano-fabrication based on two - photon polymerization. Nat. Commun. 10, 2179 (2019)

    work page 2019

  15. [15]

    Pires, J

    D. Pires, J. L. Hedrick, A. De Silva, J. Frommer, B. Gotsmann, H. Wolf, M. Despont, U. Duerig, A. W. Knoll, Nanoscale Three -Dimensional Patterning of Molecular Resists by Scanning Probes. Science 328, 732–735 (2010)

    work page 2010

  16. [16]

    Lassaline, R

    N. Lassaline, R. Brechbühler, S. J. W. Vonk, K. Ridderbeek, M. Spieser, S. Bisig, B. le Feber, F. T. Rabouw, D. J. Norris, Optical Fourier surfaces. Nature 582, 506–510 (2020)

    work page 2020

  17. [17]

    P. Chen, X. Xu, T. Wang, C. Zhou, D. Wei, J. Ma, J. Guo, X. Cui, X. Cheng, C. Xie, S. Zhang, S. Zhu, M. Xiao, Y. Zhang, Laser nanoprinting of 3D nonlinear holograms beyond 25000 pixels -per- inch for inter-wavelength-band information processing. Nat. Commun. 14, 5523 (2023)

    work page 2023

  18. [18]

    N. U. Dinc, C. Moser, D. Psaltis, Volume holograms with linear diffraction efficiency relation by (3 + 1)D printing. Opt. Lett. 49, 322–325 (2024)

    work page 2024

  19. [19]

    Curtis, D

    K. Curtis, D. Psaltis, Cross talk in phase-coded holographic memories. JOSA A 10, 2547–2550 (1993)

    work page 1993

  20. [20]

    Curtis, D

    K. Curtis, D. Psaltis, Cross talk for angle- and wavelength-multiplexed image plane holograms. Opt. Lett. 19, 1774–1776 (1994)

    work page 1994

  21. [21]

    X. Lin, Y. Rivenson, N. T. Yardimci, M. Veli, Y. Luo, M. Jarrahi, A. Ozcan, All -optical machine learning using diffractive deep neural networks. Science 361, 1004–1008 (2018)

    work page 2018

  22. [22]

    J. Li, D. Mengu, N. T. Yardimci, Y. Luo, X. Li, M. Veli, Y. Rivenson, M. Jarrahi, A. Ozcan, Spectrally encoded single-pixel machine vision using diffractive networks. Sci. Adv. 7, eabd7690 (2021)

    work page 2021

  23. [23]

    B. Bai, Y. Luo, T. Gan, J. Hu, Y. Li, Y. Zhao, D. Mengu, M. Jarrahi, A. Ozcan, To image, or not to image: class-specific diffractive cameras with all-optical erasure of undesired objects. eLight 2, 1–20 (2022)

    work page 2022

  24. [24]

    C.-Y. Shen, J. Li, T. Gan, Y. Li, M. Jarrahi, A. Ozcan, All-optical phase conjugation using diffractive wavefront processing. Nat. Commun. 15, 4989 (2024)

    work page 2024

  25. [25]

    J. Hu, K. Liao, N. U. Dinç, C. Gigli, B. Bai, T. Gan, X. Li, H. Chen, X. Yang, Y. Li, Ç. Işıl, M. S. S. Rahman, J. Li, X. Hu, M. Jarrahi, D. Psaltis, A. Ozcan, Subwavelength imaging using a solid - immersion diffractive optical processor. eLight 4, 8 (2024)

    work page 2024

  26. [26]

    Chuang, D

    E. Chuang, D. Psaltis, Storage of 1000 holograms with use of a dual-wavelength method. Appl. Opt. 36, 8445–8454 (1997)

    work page 1997

  27. [27]

    Psaltis, M

    D. Psaltis, M. Levene, A. Pu, G. Barbastathis, K. Curtis, Holographic storage using shift multiplexing. Opt. Lett. 20, 782–784 (1995)

    work page 1995

  28. [28]

    Y. Li, J. Li, Y. Zhao, T. Gan, J. Hu, M. Jarrahi, A. Ozcan, Universal Polarization Transformations: Spatial Programming of Polarization Scattering Matrices Using a Deep Learning -Designed Diffractive Polarization Transformer. Adv. Mater. 35, 2303395 (2023)

    work page 2023

  29. [29]

    D. E. Rumelhart, G. E. Hinton, R. J. Williams, Learning representations by back-propagating errors. nature 323, 533–536 (1986). 15

    work page 1986

  30. [30]

    Y. Luo, D. Mengu, N. T. Yardimci, Y. Rivenson, M. Veli, M. Jarrahi, A. Ozcan, Design of task - specific optical systems using broadband diffractive neural networks. Light Sci. Appl. 8, 112 (2019)

    work page 2019

  31. [31]

    Mengu, A

    D. Mengu, A. Ozcan, All -optical phase recovery: diffractive computing for quantitative phase imaging. Adv. Opt. Mater. 10, 2200281 (2022)

    work page 2022

  32. [32]

    B. Bai, X. Yang, T. Gan, J. Li, D. Mengu, M. Jarrahi, A. Ozcan, Pyramid diffractive optical networks for unidirectional magnification and demagnification. arXiv arXiv:2308.15019 [Preprint] (2023). http://arxiv.org/abs/2308.15019

    work page arXiv 2023

  33. [33]

    C.-Y. Shen, J. Li, Y. Li, T. Gan, L. Bai, M. Jarrahi, A. Ozcan, Multiplane quantitative phase imaging using a wavelength-multiplexed diffractive optical processor. Adv. Photonics 6, 056003 (2024)

    work page 2024

  34. [34]

    Kulce, D

    O. Kulce, D. Mengu, Y. Rivenson, A. Ozcan, All -optical synthesis of an arbitrary linear transformation using diffractive surfaces. Light Sci. Appl. 10, 196 (2021)

    work page 2021

  35. [35]

    Kulce, D

    O. Kulce, D. Mengu, Y. Rivenson, A. Ozcan, All -optical information -processing capacity of diffractive surfaces. Light Sci. Appl. 10, 25 (2021)

    work page 2021

  36. [36]

    Y. Li, Y. Luo, D. Mengu, B. Bai, A. Ozcan, Quantitative phase imaging (QPI) through random diffusers using a diffractive optical network. Light Adv. Manuf. 4, 1–16 (2023)

    work page 2023

  37. [37]

    Y. Li, S. Chen, T. Gong, A. Ozcan, Model -free Optical Processors using In Situ Reinforcement Learning with Proximal Policy Optimization. arXiv arXiv:2507.05583 [Preprint] (2025). https://doi.org/10.48550/arXiv.2507.05583

    work page doi:10.48550/arxiv.2507.05583 2025

  38. [38]

    Mengu, Y

    D. Mengu, Y. Zhao, N. T. Yardimci, Y. Rivenson, M. Jarrahi, A. Ozcan, Misalignment resilient diffractive optical networks. Nanophotonics 9, 4207–4219 (2020)

    work page 2020

  39. [39]

    G. Zhao, X. Shu, R. Zhou, High -Performance Real-World Optical Computing Trained by in Situ Gradient-Based Model-Free Optimization. IEEE Trans. Pattern Anal. Mach. Intell. 47, 7194–7205 (2025)

    work page 2025

  40. [40]

    Skalli, M

    A. Skalli, M. Goldmann, N. Haghighi, S. Reitzenstein, J. A. Lott, D. Brunner, Annealing -inspired training of an optical neural network with ternary weights. Commun. Phys. 8, 68 (2025)

    work page 2025

  41. [41]

    arXiv preprint arXiv:2503.16943 , year=

    A. Skalli, S. Sunada, M. Goldmann, M. Gebski, S. Reitzenstein, J. A. Lott, T. Czyszanowski, D. Brunner, Model-free front-to-end training of a large high performance laser neural network. arXiv arXiv:2503.16943 [Preprint] (2025). https://doi.org/10.48550/arXiv.2503.16943

    work page doi:10.48550/arxiv.2503.16943 2025

  42. [42]

    Mengu, A

    D. Mengu, A. Tabassum, M. Jarrahi, A. Ozcan, Snapshot multispectral imaging using a diffractive optical network. Light Sci. Appl. 12, 86 (2023)

    work page 2023

  43. [43]

    C.-Y. Shen, J. Li, D. Mengu, A. Ozcan, Multispectral Quantitative Phase Imaging Using a Diffractive Optical Network. Adv. Intell. Syst., 2300300 (2023)

    work page 2023

  44. [44]

    B. Bai, H. Wei, X. Yang, T. Gan, D. Mengu, M. Jarrahi, A. Ozcan, Data‐Class‐Specific All‐Optical Transformations and Encryption. Adv. Mater. 35, 2212091 (2023)

    work page 2023

  45. [45]

    S. Chen, Y. Li, H. Chen, A. Ozcan, Optical Generative Models. arXiv arXiv:2410.17970 [Preprint] (2024). https://doi.org/10.48550/arXiv.2410.17970. 16

    work page doi:10.48550/arxiv.2410.17970 2024

  46. [46]

    E. Goi, X. Chen, Q. Zhang, B. P. Cumming, S. Schoenhardt, H. Luan, M. Gu, Nanoprinted high - neuron-density optical linear perceptrons performing near-infrared inference on a CMOS chip. Light Sci. Appl. 10, 40 (2021)

    work page 2021

  47. [47]

    H. Chen, J. Feng, M. Jiang, Y. Wang, J. Lin, J. Tan, P. Jin, Diffractive deep neural networks at visible wavelengths. Engineering 7, 1483–1491 (2021)

    work page 2021

  48. [48]

    B. Bai, H. Wei, X. Yang, T. Gan, D. Mengu, M. Jarrahi, A. Ozcan, Data Class -Specific All-Optical Transformations and Encryption. Adv. Mater. n/a, 2212091

    work page

  49. [49]

    Schreiber, X

    H. Schreiber, X. G. LaFosse, J. J. Zhang, P. Batoni, J. D. Stack, J. Gardner, Diffractive optical elements for calibration of LIDAR systems: materials and fabrication. Opt. Eng. 62, 031206 (2022)

    work page 2022

  50. [50]

    C.-Y. Shen, P. Batoni, X. Yang, J. Li, K. Liao, J. Stack, J. Gardner, K. Welch, A. Ozcan, Broadband unidirectional visible imaging using wafer -scale nano-fabrication of multi -layer diffractive optical processors. Light Sci. Appl. 14, 267 (2025)

    work page 2025

  51. [51]

    Khorasaninejad, W

    M. Khorasaninejad, W. T. Chen, R. C. Devlin, J. Oh, A. Y. Zhu, F. Capasso, Metalenses at visible wavelengths: Diffraction -limited focusing and subwavelength resolution imaging. Science 352, 1190–1194 (2016)

    work page 2016

  52. [52]

    Z. Fan, C. Qian, Y. Jia, Y. Feng, H. Qian, E. -P. Li, R. Fleury, H. Chen, Holographic multiplexing metasurface with twisted diffractive neural network. Nat. Commun. 15, 9416 (2024)

    work page 2024

  53. [53]

    Zheng, Q

    H. Zheng, Q. Liu, I. I. Kravchenko, X. Zhang, Y. Huo, J. G. Valentine, Multichannel meta -imagers for accelerating machine vision. Nat. Nanotechnol. 19, 471–478 (2024)

    work page 2024

  54. [54]

    https://www.schott.com/shop/advanced -optics/en/Optical- Glass/N-BK7/c/glass-N-BK7

    N-BK7 | SCHOTT Advanced Optics. https://www.schott.com/shop/advanced -optics/en/Optical- Glass/N-BK7/c/glass-N-BK7

    work page

  55. [55]

    Gedanken

    CIFAR-10 and CIFAR-100 datasets. https://www.cs.toronto.edu/~kriz/cifar.html. 17 Figures Figure 1. Schematics of diffractive optical data storage and image projection. Illustration of diffractive storage composed of K diffractive layers, each comprising sub-wavelength phase elements jointly optimized through deep learning. The system is capable of storing...

    work page