Pith. sign in

REVIEW 4 major objections 5 minor 45 references

Low-Rank Adaptation of Deep Prior Neural Networks For Room Impulse Response Reconstruction

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper shows that Low-Rank Adaptation fine-tunes a Deep Prior for RIR reconstruction to near full fine-tuning accuracy with roughly 70% fewer trainable parameters.

desk verdict Useful first empirical study of LoRA for Deep Prior RIR reconstruction, with honest results and a real parameter-count win, but the headline accuracy gap is undercut by test-set rank selection and missing error bars. read the letter →

arxiv 2507.09806 v1 pith:X57OTHT4 submitted 2025-07-13 eess.AS cs.SD

Mirco Pezzoli , Federico Miotello , Shoichi Koyama , Fabio Antonacci This is my paper · ORCID
classification eess.AScs.SD
keywords roomimpulseresponsereconstructiondeeppriorlow-rankadaptationLoRAtransferlearningsoundfieldsparsemicrophonemeasurements
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 asks whether a Deep Prior network for room impulse response (RIR) reconstruction can be reused across acoustic setups by adapting only a small set of low-rank parameters, using the LoRA technique. It claims that, in the single-room case where only the source position changes, LoRA fine-tuning of a pretrained MultiResUNet matches full fine-tuning within 1.6 dB while training roughly 30% of the parameters, and that with as few as four microphones it clearly outperforms training from scratch. In the multi-room case, LoRA performs comparably to full fine-tuning and to retraining, although pretraining on a room with different reverberation and direction of arrival can hurt. The paper thereby establishes that parameter-efficient transfer learning is a viable route for per-element Deep Prior sound-field reconstruction, with the caveat that the benefit depends on the acoustic similarity of the pretraining and target configurations.

What carries the argument

The machinery is the LoRA convolutional adapter: for a pretrained weight $W \in \mathbb{R}^{C_{\text{out}} \times C_{\text{in}} \times k \times k}$, the update is a low-rank tensor $\Delta W = \alpha \cdot BA$ with $A \in \mathbb{R}^{r \times C_{\text{in}} \times k}$ and $B \in \mathbb{R}^{C_{\text{out}} \times k \times r}$, contracted over $r$ and reshaped to the size of $W$. The adapted layer computes $Y(x) = (W + \Delta W) * x$, so only $A$, $B$, and the scalar $\alpha$ are learned. This carries the argument by letting the pretrained Deep Prior act as a fixed backbone while the adapter encodes the new acoustic configuration; because different adapters can be swapped in, one network can serve multiple setups. The Deep Prior itself supplies the implicit prior: a fixed random input tensor $z$ is mapped through the network to RIRs, fitting only the sparse measured positions with an $\ell^1$ loss.

What would settle it

In the single-room scenario, run the same LoRA adaptation protocol with the source moved by more than 90 cm in the same room; if the NMSE advantage of the pretrained initialization over training from scratch vanishes or reverses at larger displacements, the paper's single-room transfer claim would be falsified.

Watch

Extended reading notes

Core claim

The central claim is that Low-Rank Adaptation makes Deep Prior RIR reconstruction reusable across acoustic configurations. The authors embed LoRA into every convolutional layer of a MultiResUNet by writing each weight tensor as $W + \Delta W$, with $\Delta W = \alpha \cdot BA$ where $A$ and $B$ are low-rank factors and $W$ stays frozen; only the adapters are optimized. They compare this against full fine-tuning (FT) of all parameters and against the standard Deep Prior (DP) trained from scratch. With a single room and a changed source position, LoRA with rank $r=16$ (about 30% of the trainable parameters) achieves NMSE within 1.6 dB of FT for the full microphone array, and for $\tilde{M}=4$ it is clearly better than DP from scratch. In the multi-room scenario, LoRA and FT perform comparably to each other and to DP, with LoRA sometimes best, though pretraining on the Balder room degrades adaptation because of its lower reverberation time and different source direction. The paper also reports that rank-1 adapters are instead preferable for very short adaptation runs (under 100 iterations).

Load-bearing premise

The load-bearing premise is that a network pretrained in one acoustic configuration provides a useful starting point for another; if the pretraining and target rooms differ too much, as the Balder case shows, transfer can be worse than training from scratch.

Editorial extensions

If this is right

  • In the single-room scenario, a pretrained Deep Prior adapted with LoRA uses about 70% fewer trainable parameters than full fine-tuning with NMSE degradation of at most 1.6 dB, so hardware-constrained systems can reuse prior knowledge at a fraction of the update cost.
  • With very sparse measurements ($\tilde{M}=4$), transferring from a pretrained network clearly outperforms a randomly initialized Deep Prior, suggesting that adaptation is most valuable exactly when data are scarce.
  • Rank-1 LoRA adapters are the best choice for very short adaptation runs (fewer than 100 iterations), implying a compute-versus-accuracy trade-off that can be tuned per deployment.
  • Across different rooms, LoRA, full fine-tuning, and training from scratch land within about 1.8 dB of each other, so adaptation does not consistently beat retraining; in fact, pretraining on a dissimilar room can be harmful.

Reading between the lines

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

  • Because the adapter $\Delta W$ is small and swappable, the same pretrained Deep Prior could serve a library of acoustic configurations, and the set of adapters might be interpolated or averaged to approximate unseen configurations without any optimization.
  • The harmful case (pretraining on Balder) suggests a practical selection rule: compare coarse room descriptors such as reverberation time and source direction of arrival before deciding whether to adapt from a given pretrained model or retrain from scratch.
  • The optimum at $r=16$ indicates an intrinsic dimension of the adaptation task; measuring how this rank scales with room complexity or source displacement could guide adapter sizing for larger problems.
  • Since LoRA performs best with very few microphones, a natural extension is to combine it with active sensor selection or sequential measurement acquisition, where the adapter can be refined as new microphones arrive.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes integrating Low-Rank Adaptation (LoRA) into a MultiResUNet-based Deep Prior model for room impulse response (RIR) reconstruction. The method freezes pretrained convolutional weights and learns low-rank adapter tensors, aiming to adapt a pretrained model to new source positions or rooms with few trainable parameters. The authors evaluate the approach in two scenarios: same-room source relocation (single room, M=32 microphones) and cross-room adaptation (three rooms with M=100 microphones). The main reported results are that LoRA achieves NMSE close to full fine-tuning (within 1.6 dB in the single-room case) while training roughly 30% of the parameters, and that fine-tuning is especially beneficial when only four microphones are available. In the multi-room setting, the results are mixed: LoRA and full fine-tuning sometimes underperform training from scratch, depending on the pretraining room. The paper concludes that adaptation strategies are promising for RIR reconstruction, particularly for source-move scenarios.

Significance. If the claims hold, this is a useful empirical contribution to parameter-efficient transfer learning for acoustic inverse problems, an area with little prior work. The LoRA formulation for convolutional layers is clearly presented and the experimental setup uses real measured RIR datasets, which strengthens the practical relevance. The parameter-count reduction (about 70% fewer trainable parameters for r=16) is a robust and practically meaningful achievement that is independent of the absolute NMSE ranking. The observation that low-rank adapters behave differently at short training budgets (Fig. 2a) could inform future work on very fast adaptation. The main limitations are the apparent selection of the rank hyperparameter on the evaluation scenario and the lack of uncertainty estimates, both of which affect the strength of the central accuracy-comparability claim. The paper also does not offer a criterion for when transfer will help, which is important given the mixed multi-room results.

major comments (4)
  1. [Section 3.1, Fig. 2] The LoRA rank r=16 is selected by sweeping r on the same single-room source-move scenario that is subsequently used in Fig. 2(b) to report the 'within 1.6 dB' comparison between LoRA and full fine-tuning. Because no validation split or separate development scenario is used, this is effectively a selection-on-the-test-set protocol; the reported gap may be optimistic relative to a blind rank choice. Section 3.2 then reuses r=16 in the multi-room case without reporting a separate validation for that scenario. This concern directly affects the central accuracy-comparability claim, although the parameter-count advantage is not affected.
  2. [Section 3.1 and Table 1] No error bars or repeated runs are reported in Fig. 2 or Table 1. The Deep Prior optimization involves random initialization, a randomly sampled fixed input tensor z, and randomly selected microphone subsets during adaptation, so run-to-run variability could be comparable in magnitude to the 1.6 dB difference observed at M=32. Without multiple seeds or some form of uncertainty quantification, the reader cannot assess whether the reported differences among DP, FT, and LoRA are statistically meaningful.
  3. [Section 3.1, Fig. 2(a)] The comparisons are made at a fixed training budget of 500 iterations for all methods and ranks, yet Fig. 2(a) itself shows that convergence behavior is rank-dependent: rank-1 performs best under 100 iterations, and the text suggests that r=64 might improve with a larger iteration span. Thus the 'comparable to full fine-tuning' conclusion is conditional on a single training budget, and the paper does not show that the ranking is stable across budgets. A convergence-matched or multi-budget comparison would substantially strengthen the efficiency claim.
  4. [Section 3.2, Table 1] The multi-room results show that pretraining can hurt adaptation relative to training from scratch; for example, when pretraining on Balder and adapting to Freja with M=33, DP achieves -7.2 dB while FT and LoRA achieve -6.3 dB and -6.9 dB, respectively. The paper acknowledges this but provides no diagnostic or criterion for when transfer is expected to help. Since the paper's motivation is the reuse of pretrained models, this omission limits the practical guidance; a simple proxy (e.g., T30 mismatch or direction-of-arrival mismatch) or an explicit discussion of failure modes would be needed.
minor comments (5)
  1. [Section 3.1] The rank set is stated as r=(1,2,4,16,32,64), but Fig. 2(a) also includes r=8 and the parameter-percentage list contains seven values (0.2,4,7,14,30,60,122)%. Please reconcile the enumeration.
  2. [Sections 3.1 and 3.2] The reverberation time is given as 'T60 ≈ 0.4 s' in the single-room setup and 'T30' values in the multi-room setup; please use consistent terminology and specify which reverberation metric is reported.
  3. [Section 3.2] The sentence 'the differences in NMSE between LoRA and the best results are limited in all the cases below −1.81 dB' is ambiguous; presumably the magnitude of the difference is at most 1.81 dB, but the phrasing should be clarified.
  4. [Table 1] The caption states that best values are in bold and italic indicates when LoRA outperforms FT, but the formatted table as rendered does not visibly distinguish these cases; please ensure the typesetting is clear.
  5. [Equation (7)] The NMSE definition would benefit from an explicit note that the norm is the Euclidean norm over time samples; the notation ∥·∥ is used without definition.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the LoRA adaptation results are empirical comparisons built on standard external formulations; rank selection is a validation concern, not a derivation-level reduction.

full rationale

The paper's claimed contribution is an empirical study applying LoRA to a Deep Prior MultiResUNet for RIR reconstruction. The core update equations (5)-(6) are the standard LoRA formulation [35,41], the DP optimization (3) is the standard Deep Prior objective [28], and the NMSE metric (7) is a standard evaluation measure. None of these defines the target outcome in terms of itself. The fine-tuning and from-scratch comparisons in Fig. 2(b) and Table 1 are measured, not derived, and the qualitative conclusions (LoRA within 1.6 dB, ~70% parameter reduction, stronger benefit with few microphones) are not forced by construction. The main methodological caveat is that the LoRA rank r=16 is selected in Fig. 2(a) on the same single-room source-move scenario used to report the headline comparison, and then transferred to the multi-room experiments by stating it 'has been experimentally found to be effective also in this scenario' without a separate validation split; this is a test-set selection / overfitting concern that may inflate the reported accuracy, but it does not make the result equivalent to its inputs by definition. Self-citations ([33] for the MultiResUNet-based DP architecture, [44] for a dataset) are used as prior components, but the central comparison does not depend on an unverified claim from those citations, so they are not load-bearing circularity. Therefore, under a strict definition of circularity, the paper is self-contained and receives score 0.

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

The paper introduces no new physical or mathematical entities. Its central claims rest on the Deep Prior assumption that network architecture acts as a regularizer, on the transferability of pretrained weights across acoustic configurations, and on several hyperparameters (LoRA rank, scaling factor, training iterations, learning rate) that are fixed without a validation protocol.

free parameters (4)
  • LoRA rank r = 16
    Chosen as the best rank on the single-room source-position-change scenario (Fig. 2a), then used for all other experiments. This is a hyperparameter fitted to the evaluation setup, and no separate validation split or error bars are reported.
  • LoRA scaling factor alpha = alpha = 2r
    Set as twice the rank following common LoRA practice; no ablation for alpha is provided.
  • Training iterations = 500
    Fixed number of AdamW iterations for all experiments; no convergence analysis is reported.
  • Learning rate = 0.05
    AdamW learning rate, fixed across all runs; no tuning or schedule is described.
assumptions (4)
  • domain assumption The CNN structure of the Deep Prior acts as an implicit regularizer sufficient to solve the RIR reconstruction problem (from Ulyanov et al., 2018, and Pezzoli et al., 2022)
    The whole method relies on the DIP prior property; this is assumed without re-validation in this paper.
  • domain assumption A network pretrained on one acoustic configuration provides a useful initialization for another configuration in the same or a different room
    This is the transfer-learning premise underlying the adaptation experiments; the paper acknowledges it fails in some room-to-room cases but keeps it as the foundation of the method.
  • domain assumption The random fixed input tensor z remains the same across pretraining and adaptation
    The DP paradigm keeps z fixed; the paper does not discuss the effect of changing z on transfer.
  • domain assumption The adopted LoRA parameterization from LoRA-C (Ding et al., 2024) is applicable to the MultiResUNet convolutional layers without loss of generality
    The adaptation is based on a reference implementation, and the paper does not validate the specific tensor contraction formulation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Low-Rank Adaptation of Deep Prior Neural Networks For Room Impulse Response Reconstruction." pith.science (2026). https://pith.science/paper/X57OTHT4

@misc{pith2026250709806,
  author       = {Pith},
  title        = {Pith review of: Low-Rank Adaptation of Deep Prior Neural Networks For Room Impulse Response Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X57OTHT4}},
  note         = {Machine review of arXiv:2507.09806}
}
read the original abstract

The Deep Prior framework has emerged as a powerful generative tool which can be used for reconstructing sound fields in an environment from few sparse pressure measurements. It employs a neural network that is trained solely on a limited set of available data and acts as an implicit prior which guides the solution of the underlying optimization problem. However, a significant limitation of the Deep Prior approach is its inability to generalize to new acoustic configurations, such as changes in the position of a sound source. As a consequence, the network must be retrained from scratch for every new setup, which is both computationally intensive and time-consuming. To address this, we investigate transfer learning in Deep Prior via Low-Rank Adaptation (LoRA), which enables efficient fine-tuning of a pre-trained neural network by introducing a low-rank decomposition of trainable parameters, thus allowing the network to adapt to new measurement sets with minimal computational overhead. We embed LoRA into a MultiResUNet-based Deep Prior model and compare its adaptation performance against full fine-tuning of all parameters as well as classical retraining, particularly in scenarios where only a limited number of microphones are used. The results indicate that fine-tuning, whether done completely or via LoRA, is especially advantageous when the source location is the sole changing parameter, preserving high physical fidelity, and highlighting the value of transfer learning for acoustics applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 39 canonical work pages

  1. [1]

    Virtual reality for architectural acoustics,

    M. V orl¨ander, D. Schr ¨oder, S. Pelzer, and F. Wefers, “Virtual reality for architectural acoustics,” Journal of Building Performance Simulation , vol. 8, no. 1, pp. 15–25, 2015

    work page 2015

  2. [2]

    A survey of sound source localization methods in wireless acoustic sensor networks,

    M. Cobos, F. Antonacci, A. Alexandridis, A. Mouchtaris, B. Lee et al. , “A survey of sound source localization methods in wireless acoustic sensor networks,” Wireless Communications and Mobile Computing , vol. 2017, 2017

    work page 2017

  3. [3]

    Acoustic source localization in the spherical harmonics domain exploiting low-rank approximations,

    M. Cobos, M. Pezzoli, F. Antonacci, and A. Sarti, “Acoustic source localization in the spherical harmonics domain exploiting low-rank approximations,” in Proc. ICASSP. IEEE, 2023, pp. 1–5

    work page 2023

  4. [4]

    A consoli- dated perspective on multimicrophone speech enhancement and source separation,

    S. Gannot, E. Vincent, S. Markovich-Golan, and A. Ozerov, “A consoli- dated perspective on multimicrophone speech enhancement and source separation,” IEEE/ACM Trans. Audio, Speech, Lang. Process. , vol. 25, no. 4, pp. 692–730, 2017

    work page 2017

  5. [5]

    Acoustic imaging with circular microphone array: A new approach for sound field analysis,

    M. Olivieri, A. Bastine, M. Pezzoli, F. Antonacci, T. Abhayapala, and A. Sarti, “Acoustic imaging with circular microphone array: A new approach for sound field analysis,” IEEE/ACM Trans. Audio, Speech, Lang. Process., vol. 32, pp. 1750–1761, 2024

    work page 2024

  6. [6]

    Para- metric ambisonic encoding of arbitrary microphone arrays,

    L. McCormack, A. Politis, R. Gonzalez, T. Lokki, and V . Pulkki, “Para- metric ambisonic encoding of arbitrary microphone arrays,” IEEE/ACM Trans. Audio, Speech, Lang. Process. , vol. 30, pp. 2062–2075, 2022

    work page 2022

  7. [7]

    Deep, data-driven modeling of room acoustics: literature review and research perspectives,

    T. van Waterschoot, “Deep, data-driven modeling of room acoustics: literature review and research perspectives,” in F orum Acusticum. EAA, 2025

    work page 2025

  8. [8]

    Sound field reconstruction in rooms: Inpainting meets super-resolution,

    F. Lluis, P. Martinez-Nuevo, M. Bo Møller, and S. Ewan Shepstone, “Sound field reconstruction in rooms: Inpainting meets super-resolution,” J. Acoust. Soc. Am. , vol. 148, no. 2, pp. 649–659, 2020

    work page 2020

Show all 45 references
  1. [9]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,” J. Comput. phys., vol. 378, pp. 686–707, 2019

    2019

  2. [10]

    Scientific machine learning through physics–informed neural networks: where we are and what’s next,

    S. Cuomo, V . S. Di Cola, F. Giampaolo, G. Rozza, M. Raissi, and F. Piccialli, “Scientific machine learning through physics–informed neural networks: where we are and what’s next,” J. Sci. Comput. , vol. 92, no. 3, p. 88, 2022

    2022

  3. [11]

    Koyama, J

    S. Koyama, J. G. Ribeiro, T. Nakamura, N. Ueno, and M. Pezzoli, “Physics-informed machine learning for sound field estimation: Funda- mentals, state of the art, and challenges [special issue on model-based and data-driven audio signal processing],” IEEE Signal Process. Mag. , ...

    2025

  4. [12]

    Physics- informed convolutional neural network with bicubic spline interpolation for sound field estimation,

    K. Shigemi, S. Koyama, T. Nakamura, and H. Saruwatari, “Physics- informed convolutional neural network with bicubic spline interpolation for sound field estimation,” in Proc. IWAENC. IEEE, 2022, pp. 1–5

    2022

  5. [13]

    Physics-informed neural network for volumetric sound field reconstruction of speech signals,

    M. Olivieri, X. Karakonstantis, M. Pezzoli, F. Antonacci, A. Sarti, and E. Fernandez-Grande, “Physics-informed neural network for volumetric sound field reconstruction of speech signals,” Eurasip Journal on Audio, Speech, and Music Processing , vol. 2024, no. 1, 2024

    2024

  6. [14]

    Point neuron learning: a new physics- informed neural network architecture,

    H. Bi and T. D. Abhayapala, “Point neuron learning: a new physics- informed neural network architecture,” J. Audio, Speech, Music Process. , vol. 2024, no. 1, p. 56, 2024

    2024

  7. [15]

    Room impulse response reconstruction with physics-informed deep learning,

    X. Karakonstantis, D. Caviedes-Nozal, A. Richard, and E. Fernandez- Grande, “Room impulse response reconstruction with physics-informed deep learning,” arXiv preprint arXiv:2401.01206 , 2024

    2024 arXiv

  8. [16]

    Physics-informed neural network assisted spherical microphone array signal processing,

    F. Ma, S. Zhao, and T. Abhayapala, “Physics-informed neural network assisted spherical microphone array signal processing,” J. Acoust. Soc. Am., vol. 154, no. 4 supplement, pp. A182–A182, 2023

    2023

  9. [17]

    Generative models for sound field reconstruction,

    E. Fernandez-Grande, X. Karakonstantis, D. Caviedes-Nozal, and P. Ger- stoft, “Generative models for sound field reconstruction,” The Journal of the Acoustical Society of America , vol. 153, no. 2, pp. 1179–1190, 2023

    2023

  10. [18]

    Generative adversarial networks with physical sound field priors,

    X. Karakonstantis and E. Fernandez-Grande, “Generative adversarial networks with physical sound field priors,” J. Acoust. Soc. Am. , vol. 154, no. 2, pp. 1226–1238, 2023

    2023

  11. [19]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in neural information processing systems , vol. 33, pp. 6840– 6851, 2020

    2020

  12. [20]

    Diffusion models for audio restoration: A review,

    J.-M. Lemercier, J. Richter, S. Welker, E. Moliner, V . V ¨alim¨aki, and T. Gerkmann, “Diffusion models for audio restoration: A review,” IEEE Signal Process. Mag. , vol. 41, no. 6, pp. 72–84, 2025

    2025

  13. [21]

    Buddy: Single-channel blind unsupervised dereverberation with diffusion models,

    E. Moliner, J.-M. Lemercier, S. Welker, T. Gerkmann, and V . V ¨alim¨aki, “Buddy: Single-channel blind unsupervised dereverberation with diffusion models,” in Proc. IWAENC. IEEE, 2024, pp. 120–124

    2024

  14. [22]

    Towards hrtf personalization using denoising diffusion models,

    J. C. Albarrac ´ın S ´anchez, L. Comanducci, M. Pezzoli, and F. Antonacci, “Towards hrtf personalization using denoising diffusion models,” in Proc. ICASSP, 2025, pp. 1–5

    2025

  15. [23]

    Hrtf estimation using a score-based prior,

    E. Thuillier, J.-M. Lemercier, E. Moliner, T. Gerkmann, and V . V ¨alim¨aki, “Hrtf estimation using a score-based prior,” in Proc. ICASSP. IEEE, 2025, pp. 1–5

    2025

  16. [24]

    Diffusion-based audio inpainting,

    E. M. Juanpere and V . V ¨alim¨aki, “Diffusion-based audio inpainting,” J. Audio Eng. Soc. , vol. 72, no. 3, pp. 100–113, 2024

    2024

  17. [25]

    Reconstruction of sound field through diffusion models,

    F. Miotello, L. Comanducci, M. Pezzoli, A. Bernardini, F. Antonacci, and A. Sarti, “Reconstruction of sound field through diffusion models,” in Proc. ICASSP. IEEE, 2024, pp. 1476–1480

    2024

  18. [26]

    Blind spatial impulse response generation from separate room-and scene-specific information,

    F. Llu´ıs and N. Meyer-Kahlen, “Blind spatial impulse response generation from separate room-and scene-specific information,” in Proc. ICASSP. IEEE, 2025, pp. 1–5

    2025

  19. [27]

    Diffusionrir: Room impulse response interpolation using diffusion models,

    S. Della Torre, M. Pezzoli, F. Antonacci, and S. Gannot, “Diffusionrir: Room impulse response interpolation using diffusion models,” in F orum Acusticum. EAA, 2025

    2025

  20. [28]

    Deep image prior,

    D. Ulyanov, A. Vedaldi, and V . Lempitsky, “Deep image prior,” in Proc. CVPR, 2018, pp. 9446–9454

    2018

  21. [29]

    Deep prior-based unsupervised reconstruction of irregularly sampled seismic data,

    F. Kong, F. Picetti, V . Lipari, P. Bestagini, X. Tang, and S. Tubaro, “Deep prior-based unsupervised reconstruction of irregularly sampled seismic data,” GRSL, 2020

    2020

  22. [30]

    Prediction of missing frequency response functions through deep image prior,

    R. Malvermi, F. Antonacci, A. Sarti, and R. Corradi, “Prediction of missing frequency response functions through deep image prior,” in Proc. WASPAA, 2021

    2021

  23. [31]

    Deep prior-based audio inpainting using multi-resolution harmonic convolutional neural networks,

    F. Miotello, M. Pezzoli, L. Comanducci, F. Antonacci, and A. Sarti, “Deep prior-based audio inpainting using multi-resolution harmonic convolutional neural networks,” IEEE/ACM Trans. Audio, Speech, Lang. Process., vol. 32, pp. 113–123, 2023

    2023

  24. [32]

    Noise2inverse: Self-supervised deep convolutional denoising for tomography,

    A. A. Hendriksen, D. M. Pelt, and K. J. Batenburg, “Noise2inverse: Self-supervised deep convolutional denoising for tomography,” IEEE Transactions on Computational Imaging , vol. 6, pp. 1320–1335, 2020

    2020

  25. [33]

    Deep prior approach for room impulse response reconstruction,

    M. Pezzoli, D. Perini, A. Bernardini, F. Borra, F. Antonacci, and A. Sarti, “Deep prior approach for room impulse response reconstruction,” Sensors, vol. 22, no. 7, p. 2710, 2022

    2022

  26. [34]

    Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities,

    W. Lu, R. K. Luu, and M. J. Buehler, “Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities,” npj Comput. Mat. , vol. 11, no. 1, p. 84, 2025

    2025

  27. [35]

    Lora: Low-rank adaptation of large language models

    E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, W. Chen et al. , “Lora: Low-rank adaptation of large language models.” ICLR, vol. 1, no. 2, p. 3, 2022

    2022

  28. [36]

    Niirf: Neural iir filter field for hrtf upsampling and personalization,

    Y . Masuyama, G. Wichern, F. G. Germain, Z. Pan, S. Khurana, C. Hori, and J. Le Roux, “Niirf: Neural iir filter field for hrtf upsampling and personalization,” in Proc. ICASSP. IEEE, 2024, pp. 1016–1020

    2024

  29. [37]

    Direction- aware neural acoustic fields for few-shot interpolation of ambisonic impulse responses,

    C. Ick, G. Wichern, Y . Masuyama, F. Germain, and J. L. Roux, “Direction- aware neural acoustic fields for few-shot interpolation of ambisonic impulse responses,” arXiv preprint arXiv:2505.13617 , 2025

    2025 arXiv

  30. [38]

    Sparse representation of a spatial sound field in a reverberant environment,

    S. Koyama and L. Daudet, “Sparse representation of a spatial sound field in a reverberant environment,” IEEE Journal of Selected Topics in Signal Processing, vol. 13, no. 1, pp. 172–184, 2019

    2019

  31. [39]

    Kernel interpolation of acoustic transfer functions with adaptive kernel for directed and residual reverberations,

    J. G. Ribeiro, S. Koyama, and H. Saruwatari, “Kernel interpolation of acoustic transfer functions with adaptive kernel for directed and residual reverberations,” in Proc. ICASSP. IEEE, 2023, pp. 1–5

    2023

  32. [40]

    A zero-shot physics-informed dictionary learning approach for sound field reconstruction,

    S. Damiano, F. Miotello, M. Pezzoli, A. Bernardini, F. Antonacci, A. Sarti, and T. Van Waterschoot, “A zero-shot physics-informed dictionary learning approach for sound field reconstruction,” in Proc. ICASSP , 2025, pp. 1–5

    2025

  33. [41]

    Lora-c: Parameter- efficient fine-tuning of robust cnn for iot devices,

    C. Ding, X. Cao, J. Xie, L. Fan, S. Wang, and Z. Lu, “Lora-c: Parameter- efficient fine-tuning of robust cnn for iot devices,” arXiv preprint arXiv:2410.16954, 2024

    2024 arXiv

  34. [42]

    Multiresunet : Rethinking the u-net architecture for multimodal biomedical image segmentation,

    N. Ibtehaz and M. S. Rahman, “Multiresunet : Rethinking the u-net architecture for multimodal biomedical image segmentation,” Neural Networks, vol. 121, pp. 74–87, 2020

    2020

  35. [43]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101 , 2017

    2017 arXiv

  36. [44]

    Ray-space-based multichannel nonnegative matrix factorization for audio source separation,

    M. Pezzoli, J. J. Carabias-Orti, M. Cobos, F. Antonacci, and A. Sarti, “Ray-space-based multichannel nonnegative matrix factorization for audio source separation,” IEEE Signal Process. Lett. , vol. 28, pp. 369–373, 2021

    2021

  37. [45]

    Compressed sensing of impulse responses in rooms of unknown properties and contents,

    E. Zea, “Compressed sensing of impulse responses in rooms of unknown properties and contents,” Journal of Sound and Vibration , vol. 459, p. 114871, 2019

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.