pith. sign in

arxiv: 2605.16031 · v1 · pith:JQQQ5HJKnew · submitted 2026-05-15 · ⚛️ physics.app-ph

Physics-Aware Machine-Learning-Driven Inverse Design of Broadband Ultra-Open Acoustic Metamaterials

Zhiwei Yang , Mengyu Li , Xiaohang Xie , Ao Chen , Thomas G. Bifano , Xin Zhang This is my paper

Pith reviewed 2026-05-19 17:20 UTC · model grok-4.3

classification ⚛️ physics.app-ph
keywords acoustic metamaterialsinverse designmachine learningultra-open silencersbroadband attenuationventilated acousticsGreen's function parameterization
0
0 comments X

The pith

A physics-aware machine learning framework designs ultra-open acoustic silencers achieving over 830 Hz broadband bandwidth with 80 percent ventilation in ultra-thin profiles.

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

The paper develops a machine-learning-driven inverse design method for ultra-open acoustic silencers that combine sound attenuation with high air flow. By using Green's function parameterization to separate spectral and radial design aspects, it reduces complexity while keeping physical meaning. A two-stage prediction model handles broad frequency responses and sharp resonances, paired with a parallel inverse optimizer to quickly find many good designs. This approach reveals simple linear geometric rules for best performance in single structures and uses composite arrangements to push beyond previous limits, as shown in experiments with prototypes.

Core claim

The framework uncovers hidden linear design rules that govern high-performance monolithic designs, acting as geometric proxies for optimal impedance-matching, and demonstrates through a parallel-composite architecture that broadband bandwidth exceeding 830 Hz can be achieved with an ultra-thin profile of 0.1-0.2 lambda and 80% ventilation.

What carries the argument

Green's function-based parameterization that physically decouples the design space into spectral and radial parameters to ensure interpretability and reduce complexity.

If this is right

  • Optimized designs can be identified in seconds using the population-based hybrid-objective parallel inverse strategy.
  • Monolithic ultra-open silencers follow linear design rules that serve as proxies for impedance matching.
  • Parallel-composite architectures overcome trade-offs in single-unit resonators through spatial interference.
  • Experimental prototypes validate high ventilation and broadband attenuation in ultra-thin formats.

Where Pith is reading between the lines

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

  • The decoupling approach could inspire similar parameterizations in electromagnetic or elastic metamaterial design.
  • Rapid inverse design might enable on-demand customization for specific noise environments.
  • The linear design rules might extend to other functional metamaterials where impedance matching governs performance.

Load-bearing premise

Green's function-based parameterization accurately decouples the design space into independent spectral and radial parameters without missing key physical couplings.

What would settle it

Measurements on the fabricated UAS-4 prototype showing less than 830 Hz bandwidth or below 80% effective ventilation under standard acoustic testing conditions would disprove the performance claims.

read the original abstract

Ventilated acoustic silencers combing sound attenuation with high ventilation are pivotal for advanced noise control. However, balancing attenuation, bandwidth, openness, and thickness remains a high-dimensional challenge. Here, we report a physics-aware machine-learning-driven inverse design framework for ultra-open acoustic silencers (UAS). By leveraging Green's function-based parameterization, we physically decouple the design space into spectral and radial parameters, ensuring physical interpretability while reducing complexity. We introduce a two-stage forward prediction architecture that captures broadband envelopes and sharp resonant features via a coarse-to-fine strategy. Coupled with a population-based, hybrid-objective parallel (PHP) inverse strategy, our framework enables rapid exploration of non-convex landscapes, identifying hundreds of optimized candidates within seconds. Crucially, this framework uncovers hidden linear design rules that govern high-performance monolithic designs, acting as geometric proxies for optimal impedance-matching. We experimentally validate a family of prototypes: UAS-2 demonstrates the monolithic limit with high ventilation ratio, while UAS-3 demonstrates versatility in multi-mode interactions. To circumvent the trade-off ceiling of single-unit resonators, a parallel-composite architecture (UAS-4) is introduced to enhance performance through spatial interference distribution. Results confirm a broadband bandwidth exceeding 830 Hz achieved with an ultra-thin profile (0.1-0.2{\lambda}) and 80% ventilation. This work establishes a data-driven paradigm for discovering design principles in functional metamaterials.

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 manuscript proposes a physics-aware machine-learning-driven inverse design framework for ultra-open acoustic silencers (UAS). It employs Green's function-based parameterization to decouple spectral and radial design parameters, a two-stage forward prediction architecture for broadband and resonant features, and a population-based hybrid-objective parallel (PHP) inverse strategy to explore non-convex landscapes. The framework is said to uncover hidden linear design rules acting as geometric proxies for impedance matching. Experimental validation is claimed for monolithic prototypes UAS-2 and UAS-3 as well as a parallel-composite UAS-4 architecture, achieving broadband bandwidth exceeding 830 Hz at 0.1-0.2λ thickness with 80% ventilation.

Significance. If the central claims hold, this work is significant for advancing inverse design methods in acoustic metamaterials by combining physical parameterization with ML to efficiently handle high-dimensional trade-offs between attenuation, bandwidth, openness, and thickness. The two-stage prediction and PHP strategy offer methodological tools that may generalize beyond acoustics. The experimental demonstration of high-performance thin, highly ventilated designs would support practical applications in noise control, while the reported linear design rules could provide interpretable guidelines if independently validated.

major comments (3)
  1. Abstract and framework description: The claim that Green's function-based parameterization physically decouples the design space into spectral and radial parameters may not fully hold for the parallel-composite UAS-4 architecture, which explicitly relies on spatial interference distribution. If the parameterization implicitly assumes isolated or weakly coupled scatterers, residual coupling would undermine the reduced-complexity inverse search and the reported performance numbers.
  2. Experimental validation paragraph: The abstract states experimental validation of a broadband bandwidth exceeding 830 Hz with ultra-thin profile (0.1-0.2λ) and 80% ventilation. However, the manuscript provides no measurement details, error bars, baseline comparisons, or description of how post-optimization designs were selected, leaving the central performance claim only partially supported.
  3. Section on uncovering hidden linear design rules: The claim that the framework uncovers hidden linear design rules governing high-performance monolithic designs risks circularity if these rules are data-driven fits extracted from the same ML-generated candidates rather than independently derived or externally validated relations.
minor comments (2)
  1. Notation: Define the UAS-2, UAS-3, and UAS-4 architectures more explicitly early in the main text to distinguish monolithic from composite designs.
  2. Figures: Ensure experimental plots include error bars and direct comparisons to simulations or non-optimized baselines for improved clarity and support of the performance claims.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive review of our manuscript. We have carefully considered each comment and provide point-by-point responses below. We believe the revisions will strengthen the paper.

read point-by-point responses

  1. Referee: Abstract and framework description: The claim that Green's function-based parameterization physically decouples the design space into spectral and radial parameters may not fully hold for the parallel-composite UAS-4 architecture, which explicitly relies on spatial interference distribution. If the parameterization implicitly assumes isolated or weakly coupled scatterers, residual coupling would undermine the reduced-complexity inverse search and the reported performance numbers.

    Authors: We appreciate this observation. The Green's function parameterization is applied to model the acoustic response of individual scatterers, decoupling spectral (frequency-dependent) and radial (geometric) parameters for each unit. In the UAS-4 parallel-composite design, the units are arranged to exploit spatial interference, but the inverse design still operates on the decoupled parameters per unit, with the composite effects incorporated via the two-stage forward model that accounts for interactions. This does not assume complete isolation but rather uses the parameterization to reduce the search space while the PHP strategy explores the coupled landscape. We will revise the abstract and framework section to clarify this distinction and emphasize that the decoupling facilitates the search even in the presence of controlled interference. revision: yes

  2. Referee: Experimental validation paragraph: The abstract states experimental validation of a broadband bandwidth exceeding 830 Hz with ultra-thin profile (0.1-0.2λ) and 80% ventilation. However, the manuscript provides no measurement details, error bars, baseline comparisons, or description of how post-optimization designs were selected, leaving the central performance claim only partially supported.

    Authors: We agree that additional details are necessary to fully support the experimental claims. In the revised manuscript, we will expand the experimental section to include: (i) a description of the measurement setup and protocol, (ii) error bars derived from repeated measurements, (iii) comparisons with baseline designs and simulations, and (iv) the selection criteria for the fabricated prototypes from the optimized candidates. This will provide stronger evidence for the reported performance. revision: yes

  3. Referee: Section on uncovering hidden linear design rules: The claim that the framework uncovers hidden linear design rules governing high-performance monolithic designs risks circularity if these rules are data-driven fits extracted from the same ML-generated candidates rather than independently derived or externally validated relations.

    Authors: We acknowledge the potential concern regarding circularity. The linear design rules were initially observed from the ML-optimized designs but were then analytically derived from the impedance matching condition using the Green's function model, independent of the specific ML outputs. We will include in the revision an explicit step-by-step derivation of these rules from first principles, along with their application to predict performance outside the training set, to demonstrate their general validity. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected; derivation remains self-contained

full rationale

The paper introduces a Green's function parameterization to decouple spectral and radial parameters, employs a two-stage ML forward model, and uses a hybrid inverse optimizer to generate candidates. It then extracts linear design rules by post-processing those candidates and validates performance via physical prototypes. No quoted equation or step reduces a claimed prediction or discovery to a fitted input by construction, nor does any load-bearing premise collapse to a self-citation chain. Experimental results on bandwidth and ventilation supply an external benchmark independent of the internal fitting process. The framework is therefore a standard data-driven pipeline rather than a tautological restatement of its inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Ledger entries are inferred strictly from claims in the abstract because the full manuscript text was not available for detailed inspection.

axioms (1)
  • domain assumption Green's function-based parameterization physically decouples the design space into spectral and radial parameters.
    Invoked in the framework description to ensure physical interpretability and complexity reduction.

pith-pipeline@v0.9.0 · 5803 in / 1126 out tokens · 62869 ms · 2026-05-19T17:20:43.049706+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

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

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. [1]

    Noise pollution

    Slabbekoorn, H. Noise pollution. Current Biology 29(19), R957–R960 (2019)

    work page 2019

  2. [2]

    B., Karim, Y

    Thompson, R., Smith, R. B., Karim, Y. B., Shen, C., Drummond, K., Teng, C. & Toledano, M. B. Noise pollution and human cognition: An updated systematic review and meta- analysis of recent evidence. Environment International 158, 106905 (2022). 3 Gao, N., Zhang, Z., Deng, J., Guo, X., Cheng, B. & Hou, H. Acoustic metamaterials for noise reduction: a review...

    work page 2022

  3. [3]

    & Wang, Y.- S

    Zhen, N., Huang, R.- R., Fan, S.- W., Wang, Y.- F. & Wang, Y.- S. Resonance- based acoustic ventilated metamaterials for sound insulation. npj Acoustics 1, 7 (2025)

    work page 2025

  4. [4]

    & Sheng, P

    Ma, G. & Sheng, P. Acoustic metamaterials: From local resonances to broad horizons. Science Advances 2(2), e1501595 (2016)

    work page 2016

  5. [5]

    A., Christensen, J

    Cummer, S. A., Christensen, J. & Alù, A. Controlling sound with acoustic metamaterials. Nature Reviews Materials 1(3), 1–13 (2016)

    work page 2016

  6. [6]

    & Jing, Y

    Assouar, B., Liang, B., Wu, Y., Li, Y., Cheng, J.-C. & Jing, Y. Acoustic metasurfaces. Nature Reviews Materials 3(12), 460–472 (2018)

    work page 2018

  7. [7]

    & Kim, Y

    Seo, S.-H. & Kim, Y. -H. Silencer design by using array resonators for low -frequency band noise reduction. Journal of the Acoustical Society of America 118(4), 2332–2338 (2005)

    work page 2005

  8. [8]

    & Assouar, B

    Zhu, Y. & Assouar, B. Multifunctional acoustic metasurface based on an array of helmholtz resonators. Physical Review B 99(17), 174109 (2019)

    work page 2019

  9. [9]

    Liang, Z. & Li, J. Extreme acoustic metamaterial by coiling up space. Physical Review Letters 108(11), 114301 (2012)

    work page 2012

  10. [10]

    Zhang, C. & Hu, X. Three-dimensional single-port labyrinthine acoustic metamaterial: Perfect absorption with large bandwidth and tunability. Physical Review Applied 6(6), 064025 (2016)

    work page 2016

  11. [11]

    & Assouar, B

    Zhu, Y., Merkel, A., Donda, K., Fan, S., Cao, L. & Assouar, B. Nonlocal acoustic metasurface for ultrabroadband sound absorption. Physical Review B 103(6), 064102 (2021)

    work page 2021

  12. [12]

    Liao, G., Luan, C., Wang, Z., Liu, J., Yao, X. & Fu, J. Acoustic metamaterials: A review of theories, structures, fabrication approaches, and applications. Advanced Materials Technologies 6(5), 2000787 (2021)

    work page 2021

  13. [13]

    & Sheng, P

    Mei, J., Ma, G., Yang, M., Yang, Z., Wen, W. & Sheng, P. Dark acoustic metamaterials as super absorbers for low-frequency sound. Nature Communications 3(1), 756 (2012)

    work page 2012

  14. [14]

    & Huang, Z

    Wang, X., Zhao, H., Luo, X. & Huang, Z. Membrane -constrained acoustic metamaterials for low frequency sound insulation. Applied Physics Letters 108(4), 041905 (2016). 31

    work page 2016

  15. [15]

    Ding, H., Wang, N., Wu, S., Wang, Q., Cheng, Y. & Li, Y. Compact customizable acoustic metamaterial liner with robust broadband performance under grazing flow. Physical Review Applied 25(3), 034024 (2026)

    work page 2026

  16. [16]

    & Huang, G

    Nguyen, H., Wu, Q., Xu, X., Chen, H., Tracy, S. & Huang, G. Broadband acoustic silencer with ventilation based on slit- type helmholtz resonators. Applied Physics Letters 117(13), 134103 (2020)

    work page 2020

  17. [17]

    & Lee, H

    Kumar, S. & Lee, H. P. Labyrinthine acoustic metastructures enabling broadband sound absorption and ventilation. Applied Physics Letters 116(13), 134103 (2020)

    work page 2020

  18. [18]

    & Hang, Z

    Fu, C. & Hang, Z. H. Ultrabroadband chamber silencer using metamaterials. Physical Review Applied 24(4), 044019 (2025)

    work page 2025

  19. [19]

    Dong, R., Mao, D., Wang, X. & Li, Y. Ultrabroadband acoustic ventilation barriers via hybrid-functional metasurfaces. Physical Review Applied 15(2), 024044 (2021)

    work page 2021

  20. [20]

    & Liu, X

    Shao, C., Liu, C., Ma, C., Long, H., Chen, K., Cheng, Y. & Liu, X. Multiband asymmetric sound absorber enabled by ultrasparse Mie resonators. Journal of the Acoustical Society of America 149(3), 2072–2080 (2021)

    work page 2072

  21. [21]

    Yang, Z., Chen, A., Xie, X., Anderson, S. W. & Zhang, X. Phase gradient ultra open metamaterials for broadband acoustic silencing. Scientific Reports 15, 21434 (2025)

    work page 2025

  22. [22]

    & Cheng, J.-C

    Xu, Z.-X., Gao, H., Ding, Y.-J., Yang, J., Liang, B. & Cheng, J.-C. Topology-optimized omnidirectional broadband acoustic ventilation barrier. Physical Review Applied 14(5), 054016 (2020)

    work page 2020

  23. [23]

    H., Liu, X., Christensen, J., Fang, N

    Liu, C., Shi, J., Zhao, W., Zhou, X., Ma, C., Peng, R., Wang, M., Hang, Z. H., Liu, X., Christensen, J., Fang, N. X. & Lai, Y. Three -dimensional soundproof acoustic metacage. Physical Review Letters 127(8), 084301 (2021)

    work page 2021

  24. [24]

    & Sheng, Z.- Q

    Yin, Y.-Q., Wu, H.- W., Cheng, S.- L., Sun, W.- J. & Sheng, Z.- Q. Acoustic metacage with arbitrary shape for broadband and ventilated sound insulation. Journal of Applied Physics 132(14) (2022)

    work page 2022

  25. [25]

    & Zhang, X

    Ghaffarivardavagh, R., Nikolajczyk, J., Anderson, S. & Zhang, X. Ultra -open acoustic metamaterial silencer based on Fano-like interference. Physical Review B 99(2), 024302 (2019)

    work page 2019

  26. [26]

    & Zhang, X

    Chen, A., Zhao, X., Yang, Z., Anderson, S. & Zhang, X. Broadband labyrinthine acoustic insulator. Physical Review Applied 18(6), 064057 (2022)

    work page 2022

  27. [27]

    & Chen, G

    Shao, H., He, H., Chen, Y., Tan, X. & Chen, G. A tunable metamaterial muffler with a membrane structure based on Helmholtz cavities. Applied Acoustics 157, 107022 (2020)

    work page 2020

  28. [28]

    & Huang, Z

    Wang, X., Luo, X., Yang, B. & Huang, Z. Ultrathin and durable open metamaterials for simultaneous ventilation and sound reduction. Applied Physics Letters 115(17) (2019). 32

    work page 2019

  29. [29]

    & Zhang, X

    Chen, A., Yang, Z., Zhao, X., Anderson, S. & Zhang, X. Composite acoustic metamaterial for broadband low -frequency acoustic attenuation. Physical Review Applied 20(1), 014011 (2023)

    work page 2023

  30. [30]

    & Zhang, H

    Gao, S., Hu, X., Mo, Y., Zeng, H., Mao, F., Zhu, Y. & Zhang, H. Tunable acoustic transmission control and dual -mode ventilated sound insulation by a coupled acoustic metasurface. Physical Review Applied 21(4), 044045 (2024)

    work page 2024

  31. [31]

    Zhu, Y., Dong, R., Mao, D., Wang, X. & Li, Y. Nonlocal ventilating metasurfaces. Physical Review Applied 19(1), 014067 (2023)

    work page 2023

  32. [32]

    & Zhang, H

    Gao, S., Zhu, Y., Su, Z., Zeng, H. & Zhang, H. Broadband ventilated sound insulation in a highly sparse acoustic meta-insulator array. Physical Review B 106(18), 184107 (2022)

    work page 2022

  33. [33]

    Sun, M., Fang, X., Mao, D., Wang, X. & Li, Y. Broadband acoustic ventilation barriers. Physical Review Applied 13(4), 044028 (2020)

    work page 2020

  34. [34]

    & Lin, S

    Tang, Y., Liang, B. & Lin, S. Broadband ventilated meta -barrier based on the synergy of mode superposition and consecutive F ano resonances. Journal of the Acoustical Society of America 152(4), 2412–2418 (2022)

    work page 2022

  35. [35]

    & Cheng, J.- C

    Xu, Z.-X., Qiu, W.- L., Cheng, Z.- Q., Yang, J., Liang, B. & Cheng, J.- C. Broadband ventilated sound insulation based on acoustic consecutive multiple F ano resonances. Physical Review Applied 21(4), 044049 (2024)

    work page 2024

  36. [36]

    E., Likas, A

    Lagaris, I. E., Likas, A. & Fotiadis, D. I. Artificial neural networks for solving ordinary and partial differential equations. IEEE Transactions on Neural Networks 9(5), 987–1000 (1998)

    work page 1998

  37. [37]

    E., Kevrekidis, I

    Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S. & Yang, L. Physics-informed machine learning. Nature Reviews Physics 3(6), 422–440 (2021)

    work page 2021

  38. [38]

    Cheng, M., Fu, C.-L., Okabe, R., Chotrattanapituk, A., Boonkird, A., Hung, N. T. & Li, M. Artificial intelligence -driven approaches for materials design and discovery. Nature Materials 25, 174-190 (2026)

    work page 2026

  39. [39]

    A., Boltasseva, A., Cai, W

    Ma, W., Liu, Z., Kudyshev, Z. A., Boltasseva, A., Cai, W. & Liu, Y. Deep learning for the design of photonic structures. Nature Photonics 15(2), 77–90 (2021)

    work page 2021

  40. [40]

    & Liu, Y

    Ma, W., Xu, Y., Xiong, B., Deng, L., Peng, R.- W., Wang, M. & Liu, Y. Pushing the limits of functionality -multiplexing capability in metasurface design based on statistical machine learning. Advanced Materials 34(16), 2110022 (2022)

    work page 2022

  41. [41]

    Deep learning in neural networks: An overview

    Schmidhuber, J. Deep learning in neural networks: An overview. Neural Networks 61, 85–117 (2015). 33

    work page 2015

  42. [42]

    & Zhou, J

    Li, Z., Liu, F., Yang, W., Peng, S. & Zhou, J. A survey of convolutional neural networks: analysis, applications, and prospects. IEEE Transactions on Neural Networks and Learning Systems 33(12), 6999–7019 (2021)

    work page 2021

  43. [43]

    T., Veli, M., Luo, Y., Jarrahi, M

    Lin, X., Rivenson, Y., Yardimci, N. T., Veli, M., Luo, Y., Jarrahi, M. & Ozcan, A. All- optical machine learning using diffractive deep neural networks. Science 361(6406), 1004– 1008 (2018)

    work page 2018

  44. [44]

    & Grossman, J

    Xie, T. & Grossman, J. C. Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties. Physical Review Letters 120(14), 145301 (2018)

    work page 2018

  45. [45]

    Deng, B., Zhong, P., Jun, K., Riebesell, J., Han, K., Bartel, C. J. & Ceder, G. CHGnet as a pretrained universal neural network potential for charge -informed atomistic modelling. Nature Machine Intelligence 5(9), 1031–1041 (2023)

    work page 2023

  46. [46]

    & Cui, T

    Li, L., Ruan, H., Liu, C., Li, Y., Shuang, Y., Alù, A., Qiu, C.-W. & Cui, T. J. Machine- learning reprogrammable metasurface imager. Nature Communications 10(1), 1082 (2019)

    work page 2019

  47. [47]

    & Zhou, H

    Xiao, C., Liu, M., Yao, K., Zhang, Y., Zhang, M., Yan, M., Sun, Y., Liu, X., Cui, X., Fan, T., Zhao, C., Hua, W., Ying, Y., Zheng, Y., Zhang, D., Qiu, C.- W. & Zhou, H. Ultrabroadband and band- selective thermal meta- emitters by machine learning. Nature 643(8070), 80–88 (2025)

    work page 2025

  48. [48]

    T., Abueidda, D

    Kollmann, H. T., Abueidda, D. W., Koric, S., Guleryuz, E. & Sobh, N. A. Deep learning for topology optimization of 2d metamaterials. Materials & Design 196, 109098 (2020)

    work page 2020

  49. [49]

    Amirul Islam, M., Rochan, M., Bruce, N. D. & Wang, Y. Gated feedback refinement network for dense image labeling. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 3751–3759 (2017)

    work page 2017

  50. [50]

    & Koltun, V

    Chen, Q. & Koltun, V. Photographic image synthesis with cascaded refinement networks. Proceedings of the IEEE International Conference on Computer Vision (ICCV), 1511–1520 (2017)

    work page 2017

  51. [51]

    & Shi, H

    Li, J., Hassani, A., Walton, S. & Shi, H. Convmlp: Hierarchical convolutional mlps for vision. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 6307–6316 (2023)

    work page 2023

  52. [52]

    & Zhou, T

    Liu, D., Liang, J., Geng, T., Loui, A. & Zhou, T. Tripartite feature enhanced pyramid network for dense prediction. IEEE Transactions on Image Processing 32, 2678–2692 (2023)

    work page 2023

  53. [53]

    S., Bucior, B

    Yao, Z., S´anchez-Lengeling, B., Bobbitt, N. S., Bucior, B. J., Kumar, S. G. H., Collins, S. P., Burns, T., Woo, T. K., Farha, O. K., Snurr, R. Q. & Aspuru- Guzik, A. Inverse design of 34 nanoporous crystalline reticular materials with deep generative models. Nature Machine Intelligence 3(1), 76–86 (2021)

    work page 2021

  54. [54]

    A., Liu, Z., Gupta, A., Shao, Z., Zhou, J., Khankhoje, U

    Karahan, E. A., Liu, Z., Gupta, A., Shao, Z., Zhou, J., Khankhoje, U. & Sengupta, K. Deep-learning enabled generalized inverse design of multi -port radio- frequency and sub- terahertz passives and integrated circuits. Nature Communications 15, 10734 (2024)

    work page 2024

  55. [55]

    G., Joannopoulos, J

    Peurifoy, J., Shen, Y., Jing, L., Yang, Y., Cano- Renteria, F., DeLacy, B. G., Joannopoulos, J. D., Tegmark, M. & Soljačić, M. Nanophotonic particle simulation and inverse design using artificial neural networks. Science Advances 4(6), eaar4206 (2018)

    work page 2018

  56. [56]

    Bastek, J. -H. & Kochmann, D. M. Inverse design of nonlinear mechanical metamaterials via video denoising diffusion models. Nature Machine Intelligence 5(12), 1466– 1475 (2023)

    work page 2023

  57. [57]

    & Huang, G

    Yu, Y., Chen, J., Tushar, N., Shou, W., Xu, X. & Huang, G. Compact multifunctional slit-type helmholtz silencer for broadband sound attenuation with ventilation. Applied Physics Letters 127(16), 161702 (2025)

    work page 2025

  58. [58]

    Dong, R., Mao, D., Zhu, Y., Mo, F., Wang, X. & Li, Y. A ventilating acoustic barrier for attenuating broadband diffuse sound. Applied Physics Letters 119(26), 263505 (2021). Supporting Information Supporting Information is available from the Online Library or from the author

    work page 2021