Integrated Spectropolarimeter by Metasurface-Based Diffractive Optical Networks

Chenxuan Xiang; Jumin Qiu; Qiegen Liu; Shuyuan Xiao; Tianbao Yu; Tingting Liu

arxiv: 2507.20229 · v1 · submitted 2025-07-27 · ⚛️ physics.optics · physics.app-ph

Integrated Spectropolarimeter by Metasurface-Based Diffractive Optical Networks

Jumin Qiu , Tingting Liu , Chenxuan Xiang , Tianbao Yu , Qiegen Liu , Shuyuan Xiao This is my paper

Pith reviewed 2026-05-19 02:46 UTC · model grok-4.3

classification ⚛️ physics.optics physics.app-ph

keywords metasurfacespectropolarimeterdiffractive optical networkdeep neural networkon-chip integrationCMOS sensorStokes parameterssingle-shot measurement

0 comments

The pith

A metasurface diffractive network encodes spectral and polarization information of light into spatial intensity patterns that a neural network decodes in one shot.

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

The paper shows that metasurfaces can be arranged into diffractive networks whose wavelength- and polarization-dependent phase shifts turn incoming light into unique spatial intensity maps on a sensor. A trained deep neural network then inverts those maps to recover both the full spectrum and the Stokes parameters at once. This replaces separate bulky spectrometer and polarimeter optics with a single compact layer. Experiments confirm accurate reconstruction over a wide wavelength band and demonstrate imaging as well. A working prototype mounts the entire system directly onto a commercial CMOS image sensor to create an integrated chip-scale device.

Core claim

By designing metasurfaces to apply wavelength- and polarization-specific phase modulation, the diffractive network maps any incident spectral composition and Stokes vector onto a spatially resolved intensity distribution that a deep neural network can invert to reconstruct both quantities from a single exposure, with experimental validation across broad wavelengths and a demonstrated CMOS-integrated prototype.

What carries the argument

Metasurface-based diffractive optical networks that convert wavelength and polarization into spatially resolved intensity distributions decoded by a trained deep neural network.

Load-bearing premise

Metasurface phase shifts are assumed to generate distinct enough intensity patterns for every possible combination of wavelengths and polarization states so that the neural network can invert them reliably.

What would settle it

Reconstruction error would rise sharply when the network is tested on light containing spectral lines and polarization states deliberately chosen to produce overlapping or ambiguous intensity patterns on the sensor.

read the original abstract

Conventional spectrometer and polarimeter systems rely on bulky optics, fundamentally limiting compact integration and hindering multi-dimensional optical sensing capabilities. Here, we propose a spectropolarimeter enabled by metasurface-based diffractive optical networks that simultaneously performs spectrometric and polarimetric measurements in a compact device. By leveraging the wavelength- and polarization-dependent phase modulation of metasurfaces, our system encodes the spectral and polarization information of incident light into spatially resolved intensity distributions, which are subsequently decoded by a trained deep neural network, enabling simultaneous high-accuracy reconstruction of both spectral compositions and Stokes parameters through a single-shot measurement. Experiments validate the proposed network's accurate reconstruction of the spectral and polarization information across a broad wavelength range, and further confirm its imaging capability. Notably, we demonstrate a chip-integrated sensor prototype combing both measurement functionalities into a commercial CMOS image sensor. This integrated platform provides a compact solution for on-chip multi-dimensional optical sensing, holding significant potential for versatile sensing, biomedical diagnosis, and industrial metrology.

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.

Desk Editor's Note private letter to a colleague

Metasurface-DNN spectropolarimeter integrates on CMOS with single-shot encoding but lacks the quantitative metrics needed to judge reconstruction accuracy or generalization.

read the letter

The main takeaway is a compact device that uses metasurface diffractive networks to map spectral and polarization content into spatial intensity patterns on a commercial CMOS sensor, then relies on a trained DNN to recover the spectrum and Stokes parameters from one shot. They also demonstrate basic imaging with the same setup. The on-chip prototype is the clearest practical step here, moving beyond benchtop optics toward something that could fit into real sensing hardware for biomedical or industrial uses. The encoding idea itself follows from known wavelength- and polarization-dependent metasurface behavior, paired with standard DNN decoding, so the joint functionality in one small package is the incremental advance. The paper does a reasonable job framing the size problem with conventional instruments and showing a hardware path around it. The soft spot is the evidence for performance. The abstract states that experiments confirm accurate reconstruction across a broad range, yet no error figures, dataset sizes, wavelength sampling details, or noise tests appear in the summary. That leaves the central claim—that the intensity patterns are distinct enough for reliable inversion—hard to evaluate. The stress-test point about training coverage and robustness is on target; without those specifics, degeneracies or overfitting remain possible. Readers working on compact multi-dimensional sensors or metasurface-ML combinations would find the integration concept useful as an example, even if they need the full methods to assess reliability. It deserves peer review so referees can check the actual data and experimental controls.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a compact spectropolarimeter based on metasurface diffractive optical networks that encode incident light's spectral composition and Stokes parameters into spatially resolved intensity patterns via wavelength- and polarization-dependent phase modulation. A trained deep neural network then decodes these patterns to reconstruct the spectral and polarization information from single-shot measurements. The work claims experimental validation of accurate reconstruction over a broad wavelength range, imaging capability, and a chip-integrated prototype combining both functions on a commercial CMOS image sensor.

Significance. If the performance claims hold with quantitative support, the approach could advance compact multi-dimensional optical sensing by replacing bulky conventional spectrometers and polarimeters with an integrated metasurface-plus-DNN platform, with potential applications in biomedical diagnosis and industrial metrology.

major comments (2)

[Abstract] Abstract: The central claims of 'high-accuracy reconstruction' and 'accurate reconstruction of the spectral and polarization information across a broad wavelength range' are asserted without any quantitative metrics, error bars, reconstruction accuracy figures, or dataset details; this absence directly limits evaluation of the performance claims that the experiments are said to validate.
[Abstract] The assumption that metasurface phase modulation produces sufficiently unique and invertible spatial intensity distributions for arbitrary spectral compositions and Stokes parameters is load-bearing for the single-shot reconstruction claim, yet the manuscript provides no specification of training-set diversity (e.g., number of wavelength samples or polarization combinations) or tests of robustness to realistic sensor noise.

minor comments (1)

[Abstract] The abstract contains a typographical error ('combing' instead of 'combining').

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and have revised the manuscript accordingly where the suggestions strengthen the presentation of our results.

read point-by-point responses

Referee: [Abstract] Abstract: The central claims of 'high-accuracy reconstruction' and 'accurate reconstruction of the spectral and polarization information across a broad wavelength range' are asserted without any quantitative metrics, error bars, reconstruction accuracy figures, or dataset details; this absence directly limits evaluation of the performance claims that the experiments are said to validate.

Authors: We agree that the abstract benefits from explicit quantitative support. In the revised manuscript we have updated the abstract to report key performance metrics drawn directly from the experimental results, including average spectral reconstruction error with standard deviation, Stokes parameter reconstruction accuracy, and a brief statement of the validation dataset size and composition. revision: yes
Referee: [Abstract] The assumption that metasurface phase modulation produces sufficiently unique and invertible spatial intensity distributions for arbitrary spectral compositions and Stokes parameters is load-bearing for the single-shot reconstruction claim, yet the manuscript provides no specification of training-set diversity (e.g., number of wavelength samples or polarization combinations) or tests of robustness to realistic sensor noise.

Authors: The Methods section of the original manuscript already specifies the training-set composition (wavelength sampling across the target band and the six polarization states used to generate the Stokes-parameter training data). To directly address the concern about noise robustness, we have added a new paragraph and supplementary figure that quantify reconstruction fidelity under additive Gaussian noise levels matching the measured CMOS sensor characteristics. revision: partial

Circularity Check

0 steps flagged

No circularity in derivation chain

full rationale

The paper encodes spectral and polarization information via wavelength- and polarization-dependent metasurface phase modulation into spatial intensity patterns, then decodes via a separately trained DNN for single-shot reconstruction. This chain relies on physical metasurface properties for encoding and supervised learning for decoding, with experimental validation on a CMOS sensor prototype. No equations, predictions, or central claims reduce by construction to fitted inputs or self-citations; the uniqueness of encodings is an empirical assumption rather than a definitional tautology, and no load-bearing uniqueness theorem or ansatz is imported from prior author work. The derivation remains self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The proposal rests on established metasurface physics and standard DNN training practices; no new free parameters, axioms beyond domain assumptions, or invented entities are introduced in the abstract description.

axioms (1)

domain assumption Metasurfaces provide wavelength- and polarization-dependent phase modulation sufficient for encoding spectral and polarization information into spatially resolved intensities.
Invoked as the basis for the encoding step in the proposed system.

pith-pipeline@v0.9.0 · 5715 in / 1167 out tokens · 52130 ms · 2026-05-19T02:46:59.105610+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

By leveraging the wavelength- and polarization-dependent phase modulation of metasurfaces, our system encodes the spectral and polarization information of incident light into spatially resolved intensity distributions, which are subsequently decoded by a trained deep neural network
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The metasurface imposes independent phase profiles on each circular polarization component... angular spectrum propagation

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages

[1]

Yuan, S. et al. Geometric deep optical sensing. Science 379, eade1220 (2023)

work page 2023
[2]

He, C. et al. Polarisation optics for biomedical and clinical applications: a review. Light Sci. Appl. 10, 194 (2021)

work page 2021
[3]

& Hasan, T

Yang, Z., Albrow-Owen, T., Cai, W. & Hasan, T. Miniaturization of optical spectrometers. Science 371, eabe0722 (2021)

work page 2021
[4]

Li, A. et al. Advances in cost-effective integrated spectrometers. Light Sci. Appl. 11, 174 (2022)

work page 2022
[5]

Xue, Q. et al. Advances in miniaturized computational spectrometers. Adv. Sci. 11, 2404448 (2024)

work page 2024
[6]

Dorrah, A. H. & Capasso, F. Tunable structured light with flat optics. Science 376, eabi6860 (2022)

work page 2022
[7]

Zheng, P. et al. Metasurface-based key for computational imaging encryption. Sci. Adv. 7, eabg0363 (2021)

work page 2021
[8]

Deng, Z.-L. et al. Poincar´ e sphere trajectory encoding metasurfaces based on generalized malus’ law. Nat. Commun. 15, 2380 (2024)

work page 2024
[9]

& Xiao, S

Liu, T., Li, J. & Xiao, S. Single-sized phase-change metasurfaces for dynamic information multiplexing and encryption. Phys. Rev. Research 6, 023258 (2024)

work page 2024
[10]

Deng, Z.-L. et al. Diatomic metasurface for vectorial holography. Nano Lett. 18, 2885–2892 (2018)

work page 2018
[11]

Li, J. et al. Racemic dielectric metasurfaces for arbitrary terahertz polarization rotation and wavefront manipulation. Opto-Electron. Adv. 7, 240075 (2024)

work page 2024
[12]

Ding, F. et al. Versatile polarization generation and manipulation using dielectric metasurfaces. Laser Photonics Rev. 14, 2000116 (2020)

work page 2020
[13]

& Yang, Y

Wang, S., Wen, S., Deng, Z.-L., Li, X. & Yang, Y. Metasurface-based solid poincar´ e sphere polarizer.Phys. Rev. Lett. 130, 123801 (2023). 13

work page 2023
[14]

Li, S. et al. Metasurface polarization optics: Phase manipulation for arbitrary polarization conversion condition. Phys. Rev. Lett. 134, 023803 (2025)

work page 2025
[15]

Ren, H. et al. Complex-amplitude metasurface-based orbital angular momentum holography in momentum space. Nat. Nanotechnol. 15, 948–955 (2020)

work page 2020
[16]

Meng, W. et al. Ultranarrow-linewidth wavelength-vortex metasurface hologra- phy. Sci. Adv. 11, eadt9159 (2025)

work page 2025
[17]

Wang, R. et al. Compact multi-foci metalens spectrometer. Light Sci. Appl. 12, 103 (2023)

work page 2023
[18]

Cai, G. et al. Compact angle-resolved metasurface spectrometer. Nat. Mater. 23, 71–78 (2024)

work page 2024
[19]

Zhang, Z., Xiao, S., Song, Q. & Xu, K. Scalable on-chip diffractive speckle spectrometer with high spectral channel density. Light Sci. Appl. 14, 130 (2025)

work page 2025
[20]

& Jang, M

Lee, D.-g., Song, G., Lee, C., Lee, C. & Jang, M. Reconstructive spectrometer using double-layer disordered metasurfaces. Sci. Adv. 11, eadv2376 (2025)

work page 2025
[21]

Lin, C.-H., Huang, S.-H., Lin, T.-H. & Wu, P. C. Metasurface-empowered snap- shot hyperspectral imaging with convex/deep (code) small-data learning theory. Nat. Commun. 14, 6979 (2023)

work page 2023
[22]

Kim, I. et al. Metasurfaces-driven hyperspectral imaging via multiplexed plasmonic resonance energy transfer. Adv. Mater. 35, 2300229 (2023)

work page 2023
[23]

Audhkhasi, R., Xie, N., Fr¨ och, J. E. & Majumdar, A. Single-shot multispectral imaging via a chromatic metalens array. ACS Photonics 12, 2761–2766 (2025)

work page 2025
[24]

Tittl, A. et al. Imaging-based molecular barcoding with pixelated dielectric metasurfaces. Science 360, 1105–1109 (2018)

work page 2018
[25]

Tang, F. et al. Metasurface spectrometers beyond resolution-sensitivity con- straints. Sci. Adv. 10, eadr7155 (2024)

work page 2024
[26]

Li, Z. et al. Three-channel metasurfaces for simultaneous meta-holography and meta-nanoprinting: A single-cell design approach. Laser Photonics Rev. 14, 2000032 (2020)

work page 2020
[27]

& Xiao, S

Liu, T., Han, Z., Duan, J. & Xiao, S. Phase-change metasurfaces for dynamic image display and information encryption. Phys. Rev. Applied 18, 044078 (2022)

work page 2022
[28]

Ou, X. et al. Tunable polarization-multiplexed achromatic dielectric metalens. Nano Lett. 22, 10049–10056 (2022). 14

work page 2022
[29]

Wang, J. et al. Unlocking ultra-high holographic information capacity through nonorthogonal polarization multiplexing. Nat. Commun. 15, 6284 (2024)

work page 2024
[30]

Hu, Y. et al. Achromatic full stokes polarimetry metasurface for full-color polarization imaging in the visible range. Nano Lett. 24, 13018–13026 (2024)

work page 2024
[31]

Chen, C. et al. Neural network assisted high-spatial-resolution polarimetry with non-interleaved chiral metasurfaces. Light Sci. Appl. 12, 288 (2023)

work page 2023
[32]

Yang, H. et al. Metasurface higher-order poincar´ e sphere polarization detection clock. Light Sci. Appl. 14, 63 (2025)

work page 2025
[33]

Zuo, J. et al. Chip-integrated metasurface full-stokes polarimetric imaging sensor. Light Sci. Appl. 12, 218 (2023)

work page 2023
[34]

Shen, Z. et al. Monocular metasurface camera for passive single-shot 4d imaging. Nat. Commun. 14, 1035 (2023)

work page 2023
[35]

Zaidi, A. et al. Metasurface-enabled single-shot and complete mueller matrix imaging. Nat. Photonics 18, 704–712 (2024)

work page 2024
[36]

He, H. et al. Meta-attention network based spectral reconstruction with snapshot near-infrared metasurface. Adv. Mater. 36, 2313357 (2024)

work page 2024
[37]

W., Oh, J., Miller, H., Capasso, F

Li, L. W., Oh, J., Miller, H., Capasso, F. & Rubin, N. A. Flat, wide field-of-view imaging polarimeter. Optica 12, 799 (2025)

work page 2025
[38]

Ni, Y. et al. Computational spectropolarimetry with a tunable liquid crystal metasurface. eLight 2, 23 (2022)

work page 2022
[39]

& Zhang, X

Chen, L., Yu, Y. & Zhang, X. Imaging spectropolarimeter using a multifunctional metasurface. Nano Lett. 24, 12634–12641 (2024)

work page 2024
[40]

Zhang, L. et al. Real-time machine learning–enhanced hyperspectro-polarimetric imaging via an encoding metasurface. Sci. Adv. 10, eadp5192 (2024)

work page 2024
[41]

Lin, X. et al. All-optical machine learning using diffractive deep neural networks. Science 361, 1004–1008 (2018)

work page 2018
[42]

Yan, T. et al. Fourier-space diffractive deep neural network. Phys. Rev. Lett. 123, 023901 (2019)

work page 2019
[43]

Liu, C. et al. A programmable diffractive deep neural network based on a digital- coding metasurface array. Nat. Electron. 5, 113–122 (2022)

work page 2022
[44]

Luo, X. et al. Metasurface-enabled on-chip multiplexed diffractive neural networks in the visible. Light Sci. Appl. 11, 158 (2022). 15

work page 2022
[45]

Zhan, Z., Wang, H., Liu, Q. & Fu, X. Photonic diffractive generators through sampling noises from scattering media. Nat. Commun. 15, 10643 (2024)

work page 2024
[46]

Qiu, J. et al. Optoelectronic generative adversarial networks. Commun. Phys. 8, 162 (2025)

work page 2025
[47]

Bai, B. et al. Information-hiding cameras: Optical concealment of object information into ordinary images. Sci. Adv. 10, eadn9420 (2024)

work page 2024
[48]

Guo, Z. et al. Polarization-selective unidirectional and bidirectional diffractive neural networks for information security and sharing. Nat. Commun. 16, 4492 (2025)

work page 2025
[49]

Li, J. et al. Unidirectional imaging using deep learning–designed materials. Sci. Adv. 9, eadg1505 (2023)

work page 2023
[50]

& Ozcan, A

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

work page 2023
[51]

Xu, Z. et al. Large-scale photonic chiplet taichi empowers 160-tops/w artificial general intelligence. Science 384, 202–209 (2024)

work page 2024
[52]

Hu, J. et al. Subwavelength imaging using a solid-immersion diffractive optical processor. eLight 4, 8 (2024)

work page 2024
[53]

Qiu, J. et al. Decision-making and control with diffractive optical networks. Adv. Photonics Nexus 3, 046003 (2024)

work page 2024
[54]

Fan, Y. et al. Dispersion-assisted high-dimensional photodetector. Nature 630, 77–83 (2024)

work page 2024
[55]

Zhou, Q. et al. Generation of perfect vortex beams by dielectric geometric metasurface for visible light. Laser Photonics Rev. 15, 2100390 (2021)

work page 2021
[56]

Gu, M. et al. Dielectric supercell metasurfaces for generating focused higher-order poincar´ e beams with the residual copolarization component eliminated. ACS Photonics 11, 204–217 (2024)

work page 2024
[57]

Guo, Y. et al. Spin-decoupled metasurface for simultaneous detection of spin and orbital angular momenta via momentum transformation. Light Sci. Appl. 10, 63 (2021)

work page 2021
[58]

Zhang, Y. et al. Ultracompact metasurface for simultaneous detection of polar- ization state and orbital angular momentum. Laser Photonics Rev. 18, 2301012 (2023). 16 Acknowledgments This work was supported by the National Natural Science Foundation of China (Nos. 12064025, 12264028, 12364045, 12304420, and U24A20304), the Natural Sci- ence Foundation of ...

work page 2023

[1] [1]

Yuan, S. et al. Geometric deep optical sensing. Science 379, eade1220 (2023)

work page 2023

[2] [2]

He, C. et al. Polarisation optics for biomedical and clinical applications: a review. Light Sci. Appl. 10, 194 (2021)

work page 2021

[3] [3]

& Hasan, T

Yang, Z., Albrow-Owen, T., Cai, W. & Hasan, T. Miniaturization of optical spectrometers. Science 371, eabe0722 (2021)

work page 2021

[4] [4]

Li, A. et al. Advances in cost-effective integrated spectrometers. Light Sci. Appl. 11, 174 (2022)

work page 2022

[5] [5]

Xue, Q. et al. Advances in miniaturized computational spectrometers. Adv. Sci. 11, 2404448 (2024)

work page 2024

[6] [6]

Dorrah, A. H. & Capasso, F. Tunable structured light with flat optics. Science 376, eabi6860 (2022)

work page 2022

[7] [7]

Zheng, P. et al. Metasurface-based key for computational imaging encryption. Sci. Adv. 7, eabg0363 (2021)

work page 2021

[8] [8]

Deng, Z.-L. et al. Poincar´ e sphere trajectory encoding metasurfaces based on generalized malus’ law. Nat. Commun. 15, 2380 (2024)

work page 2024

[9] [9]

& Xiao, S

Liu, T., Li, J. & Xiao, S. Single-sized phase-change metasurfaces for dynamic information multiplexing and encryption. Phys. Rev. Research 6, 023258 (2024)

work page 2024

[10] [10]

Deng, Z.-L. et al. Diatomic metasurface for vectorial holography. Nano Lett. 18, 2885–2892 (2018)

work page 2018

[11] [11]

Li, J. et al. Racemic dielectric metasurfaces for arbitrary terahertz polarization rotation and wavefront manipulation. Opto-Electron. Adv. 7, 240075 (2024)

work page 2024

[12] [12]

Ding, F. et al. Versatile polarization generation and manipulation using dielectric metasurfaces. Laser Photonics Rev. 14, 2000116 (2020)

work page 2020

[13] [13]

& Yang, Y

Wang, S., Wen, S., Deng, Z.-L., Li, X. & Yang, Y. Metasurface-based solid poincar´ e sphere polarizer.Phys. Rev. Lett. 130, 123801 (2023). 13

work page 2023

[14] [14]

Li, S. et al. Metasurface polarization optics: Phase manipulation for arbitrary polarization conversion condition. Phys. Rev. Lett. 134, 023803 (2025)

work page 2025

[15] [15]

Ren, H. et al. Complex-amplitude metasurface-based orbital angular momentum holography in momentum space. Nat. Nanotechnol. 15, 948–955 (2020)

work page 2020

[16] [16]

Meng, W. et al. Ultranarrow-linewidth wavelength-vortex metasurface hologra- phy. Sci. Adv. 11, eadt9159 (2025)

work page 2025

[17] [17]

Wang, R. et al. Compact multi-foci metalens spectrometer. Light Sci. Appl. 12, 103 (2023)

work page 2023

[18] [18]

Cai, G. et al. Compact angle-resolved metasurface spectrometer. Nat. Mater. 23, 71–78 (2024)

work page 2024

[19] [19]

Zhang, Z., Xiao, S., Song, Q. & Xu, K. Scalable on-chip diffractive speckle spectrometer with high spectral channel density. Light Sci. Appl. 14, 130 (2025)

work page 2025

[20] [20]

& Jang, M

Lee, D.-g., Song, G., Lee, C., Lee, C. & Jang, M. Reconstructive spectrometer using double-layer disordered metasurfaces. Sci. Adv. 11, eadv2376 (2025)

work page 2025

[21] [21]

Lin, C.-H., Huang, S.-H., Lin, T.-H. & Wu, P. C. Metasurface-empowered snap- shot hyperspectral imaging with convex/deep (code) small-data learning theory. Nat. Commun. 14, 6979 (2023)

work page 2023

[22] [22]

Kim, I. et al. Metasurfaces-driven hyperspectral imaging via multiplexed plasmonic resonance energy transfer. Adv. Mater. 35, 2300229 (2023)

work page 2023

[23] [23]

Audhkhasi, R., Xie, N., Fr¨ och, J. E. & Majumdar, A. Single-shot multispectral imaging via a chromatic metalens array. ACS Photonics 12, 2761–2766 (2025)

work page 2025

[24] [24]

Tittl, A. et al. Imaging-based molecular barcoding with pixelated dielectric metasurfaces. Science 360, 1105–1109 (2018)

work page 2018

[25] [25]

Tang, F. et al. Metasurface spectrometers beyond resolution-sensitivity con- straints. Sci. Adv. 10, eadr7155 (2024)

work page 2024

[26] [26]

Li, Z. et al. Three-channel metasurfaces for simultaneous meta-holography and meta-nanoprinting: A single-cell design approach. Laser Photonics Rev. 14, 2000032 (2020)

work page 2020

[27] [27]

& Xiao, S

Liu, T., Han, Z., Duan, J. & Xiao, S. Phase-change metasurfaces for dynamic image display and information encryption. Phys. Rev. Applied 18, 044078 (2022)

work page 2022

[28] [28]

Ou, X. et al. Tunable polarization-multiplexed achromatic dielectric metalens. Nano Lett. 22, 10049–10056 (2022). 14

work page 2022

[29] [29]

Wang, J. et al. Unlocking ultra-high holographic information capacity through nonorthogonal polarization multiplexing. Nat. Commun. 15, 6284 (2024)

work page 2024

[30] [30]

Hu, Y. et al. Achromatic full stokes polarimetry metasurface for full-color polarization imaging in the visible range. Nano Lett. 24, 13018–13026 (2024)

work page 2024

[31] [31]

Chen, C. et al. Neural network assisted high-spatial-resolution polarimetry with non-interleaved chiral metasurfaces. Light Sci. Appl. 12, 288 (2023)

work page 2023

[32] [32]

Yang, H. et al. Metasurface higher-order poincar´ e sphere polarization detection clock. Light Sci. Appl. 14, 63 (2025)

work page 2025

[33] [33]

Zuo, J. et al. Chip-integrated metasurface full-stokes polarimetric imaging sensor. Light Sci. Appl. 12, 218 (2023)

work page 2023

[34] [34]

Shen, Z. et al. Monocular metasurface camera for passive single-shot 4d imaging. Nat. Commun. 14, 1035 (2023)

work page 2023

[35] [35]

Zaidi, A. et al. Metasurface-enabled single-shot and complete mueller matrix imaging. Nat. Photonics 18, 704–712 (2024)

work page 2024

[36] [36]

He, H. et al. Meta-attention network based spectral reconstruction with snapshot near-infrared metasurface. Adv. Mater. 36, 2313357 (2024)

work page 2024

[37] [37]

W., Oh, J., Miller, H., Capasso, F

Li, L. W., Oh, J., Miller, H., Capasso, F. & Rubin, N. A. Flat, wide field-of-view imaging polarimeter. Optica 12, 799 (2025)

work page 2025

[38] [38]

Ni, Y. et al. Computational spectropolarimetry with a tunable liquid crystal metasurface. eLight 2, 23 (2022)

work page 2022

[39] [39]

& Zhang, X

Chen, L., Yu, Y. & Zhang, X. Imaging spectropolarimeter using a multifunctional metasurface. Nano Lett. 24, 12634–12641 (2024)

work page 2024

[40] [40]

Zhang, L. et al. Real-time machine learning–enhanced hyperspectro-polarimetric imaging via an encoding metasurface. Sci. Adv. 10, eadp5192 (2024)

work page 2024

[41] [41]

Lin, X. et al. All-optical machine learning using diffractive deep neural networks. Science 361, 1004–1008 (2018)

work page 2018

[42] [42]

Yan, T. et al. Fourier-space diffractive deep neural network. Phys. Rev. Lett. 123, 023901 (2019)

work page 2019

[43] [43]

Liu, C. et al. A programmable diffractive deep neural network based on a digital- coding metasurface array. Nat. Electron. 5, 113–122 (2022)

work page 2022

[44] [44]

Luo, X. et al. Metasurface-enabled on-chip multiplexed diffractive neural networks in the visible. Light Sci. Appl. 11, 158 (2022). 15

work page 2022

[45] [45]

Zhan, Z., Wang, H., Liu, Q. & Fu, X. Photonic diffractive generators through sampling noises from scattering media. Nat. Commun. 15, 10643 (2024)

work page 2024

[46] [46]

Qiu, J. et al. Optoelectronic generative adversarial networks. Commun. Phys. 8, 162 (2025)

work page 2025

[47] [47]

Bai, B. et al. Information-hiding cameras: Optical concealment of object information into ordinary images. Sci. Adv. 10, eadn9420 (2024)

work page 2024

[48] [48]

Guo, Z. et al. Polarization-selective unidirectional and bidirectional diffractive neural networks for information security and sharing. Nat. Commun. 16, 4492 (2025)

work page 2025

[49] [49]

Li, J. et al. Unidirectional imaging using deep learning–designed materials. Sci. Adv. 9, eadg1505 (2023)

work page 2023

[50] [50]

& Ozcan, A

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

work page 2023

[51] [51]

Xu, Z. et al. Large-scale photonic chiplet taichi empowers 160-tops/w artificial general intelligence. Science 384, 202–209 (2024)

work page 2024

[52] [52]

Hu, J. et al. Subwavelength imaging using a solid-immersion diffractive optical processor. eLight 4, 8 (2024)

work page 2024

[53] [53]

Qiu, J. et al. Decision-making and control with diffractive optical networks. Adv. Photonics Nexus 3, 046003 (2024)

work page 2024

[54] [54]

Fan, Y. et al. Dispersion-assisted high-dimensional photodetector. Nature 630, 77–83 (2024)

work page 2024

[55] [55]

Zhou, Q. et al. Generation of perfect vortex beams by dielectric geometric metasurface for visible light. Laser Photonics Rev. 15, 2100390 (2021)

work page 2021

[56] [56]

Gu, M. et al. Dielectric supercell metasurfaces for generating focused higher-order poincar´ e beams with the residual copolarization component eliminated. ACS Photonics 11, 204–217 (2024)

work page 2024

[57] [57]

Guo, Y. et al. Spin-decoupled metasurface for simultaneous detection of spin and orbital angular momenta via momentum transformation. Light Sci. Appl. 10, 63 (2021)

work page 2021

[58] [58]

Zhang, Y. et al. Ultracompact metasurface for simultaneous detection of polar- ization state and orbital angular momentum. Laser Photonics Rev. 18, 2301012 (2023). 16 Acknowledgments This work was supported by the National Natural Science Foundation of China (Nos. 12064025, 12264028, 12364045, 12304420, and U24A20304), the Natural Sci- ence Foundation of ...

work page 2023