Pith. sign in

REVIEW 5 major objections 6 minor 30 references

The Computation of Generalized Embeddings for Underwater Acoustic Target Recognition using Contrastive Learning

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A self-supervised contrastive encoder trained on unlabeled single-hydrophone audio produces embeddings that transfer to ship-type and marine-mammal classification as well as supervised baselines.

desk verdict A solid proof of concept for self-supervised pretraining on unlabeled hydrophone data, but the 'generalized embeddings' claim would be stronger with a random-initialization baseline and repeated runs. read the letter →

arxiv 2505.12904 v1 pith:W2MAEP63 submitted 2025-05-19 cs.SD cs.AIeess.AS

Hilde I. Hummel , Arwin Gansekoele , Sandjai Bhulai , Rob van der Mei This is my paper · ORCID
classification cs.SDcs.AIeess.AS
keywords UnderwateracoustictargetrecognitionContrastivelearningVICRegSelf-supervisedShipradiatednoiseMarinemammalvocalizationsConformerTransfer
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

This paper tries to show that useful underwater acoustic classifiers can be built without labeled training data. It trains a Conformer encoder with the VICReg contrastive loss on unlabeled recordings collected at a single hydrophone near Vancouver, then uses the learned embeddings as features for simple logistic-regression classifiers on three labeled benchmarks: two ship-type datasets and one marine-mammal vocalization dataset. The reported accuracy is close to supervised contrastive baselines, and the unsupervised method degrades less when labeled data are reduced. If correct, passive acoustic monitoring could be automated from the large publicly available stock of unlabeled ocean recordings.

What carries the argument

The central object is the unsupervised contrastive-learning pipeline built on the VICReg loss, Variance-Invariance-Covariance Regularization. VICReg needs no negative samples: it trains the encoder by pulling augmented views of the same clip together in embedding space, pushing the variance of each embedding dimension toward a target, and minimizing off-diagonal covariances to prevent dimensional collapse. Four Conformer blocks serve as the encoder, and a two-layer MLP projects the embeddings for optimization. The augmentation family is the domain-knowledge injection: low-pass filtering at 1 kHz targets the band where ship-type discriminative information concentrates, and a mixup augmentation combines low- and high-frequency content from nearby time windows. This machinery lets the model learn from raw unlabeled audio and still produce features that separate ship types and whale calls in other recordings.

What would settle it

Train the identical VICReg pipeline on unlabeled recordings from a hydrophone far from Vancouver, ideally near the ShipsEar recording region, then attach the same linear classifiers. If accuracy on ShipsEar and Watkins drops substantially compared with the Vancouver-trained model, the claim that a single representative hydrophone yields generalized embeddings is falsified.

Watch

Extended reading notes

Core claim

The central claim is that an unsupervised contrastive-learning pipeline, trained without labels on raw audio from one hydrophone, yields generalized embeddings for underwater acoustic target recognition. The authors define positive pairs by augmenting two-second audio windows with domain-specific functions, then optimize a four-block Conformer encoder with the VICReg loss, which enforces invariance between augmented views, keeps feature variance alive, and decorrelates embedding dimensions. Linear classifiers built on these embeddings reach 54.87% accuracy on Deepship, 57.42% on ShipsEar, and 86.10% on Watkins, close to the supervised Conformer baseline, and they hold up better than the supervised baselines when the labeled fraction of Deepship falls to 25% and below. The paper offers this as a proof-of-concept that unlabeled ocean audio can replace labeled corpora as the training signal for UATR backbones.

Load-bearing premise

The load-bearing premise is that recordings from one hydrophone near Vancouver are representative enough of the acoustic environments in all three benchmark datasets, including ShipsEar off Spain and the decades-spanning Watkins collection, and this premise is never tested.

Editorial extensions

If this is right

  • Passive acoustic monitoring pipelines could be initialized from unlabeled ocean recordings, removing the dependence on large labeled corpora for underwater acoustic target recognition.
  • The same backbone may transfer to other underwater acoustic analysis tasks, such as climate-change monitoring or detection of human-made sounds, because the embeddings are not tied to one dataset's label set.
  • Time-wise evaluation of Deepship suggests that models trained on a random split overstate real-world performance; time-wise generalization is the harder and more deployment-relevant test.
  • Unsupervised models are more robust than supervised contrastive baselines to reduced label availability on Deepship, so label collection effort could be shifted to downstream evaluation.
  • Expanding the augmentation family with speech-based augmentations helps supervised models but can hurt unsupervised transfer, making augmentation choice a key design lever for self-supervised UATR.

Reading between the lines

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

  • A stronger version of the claim would hold if training on a hydrophone far from all three benchmarks still transferred; that test would distinguish general acoustic-structure learning from site-specific similarity to Deepship.
  • The ShipsEar result, where the unsupervised Conformer exceeds the supervised Conformer baseline, is consistent with supervised training on Deepship overfitting site-specific class boundaries, but the paper does not identify the exact spectral or temporal features driving the transfer.
  • A direct extension would measure whether linear probes trained on Deepship embeddings transfer to ShipsEar and Watkins labels in a zero-shot or few-shot protocol, and compare against spectrogram-statistics baselines.
  • A systematic sweep over hydrophone location, depth, season, and recording hardware would define the practical envelope of the single-hydrophone claim.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes an unsupervised contrastive learning pipeline for underwater acoustic target recognition (UATR). A Conformer encoder is trained with the VICReg loss on unlabeled recordings from a single ONC hydrophone near Vancouver, and the learned embeddings are evaluated by training a logistic regression classifier on three labeled benchmarks: Deepship (ship type), ShipsEar (ship type), and The Best of Watkin's (marine mammal vocalizations). The method is compared against supervised contrastive learning baselines (ResNet18 and Conformer) and is additionally analyzed with respect to VICReg loss weights, augmentation functions, reduced labeled data, embedding size, and computational cost. The central claim is that the unsupervised contrastive approach produces robust and generalized embeddings that transfer across geographically and temporally diverse datasets, with accuracy comparable to supervised baselines and greater robustness when labeled data are scarce.

Significance. If the claims hold, the paper offers a useful proof-of-concept that publicly available unlabeled hydrophone data can serve as pretraining data for multiple underwater acoustic classification tasks, reducing reliance on scarce labeled data. The paper is honest about using held-out test sets and a time-wise split for Deepship, and it makes the code available. The study addresses a relevant problem for passive acoustic monitoring and is likely to interest the underwater acoustics and applied self-supervised learning communities. However, the strength of the central claim depends on the missing controls and evaluation issues discussed in the major comments. The manuscript is a solid empirical contribution in need of revision rather than a breakthrough result.

major comments (5)
  1. [Section 3, Table 1] The central claim that VICReg training on unlabeled ONC data produces generalized embeddings is not supported by the current baseline set. Tables 1, 3, and 4 compare only VICReg-trained encoders with supervised-CL-trained encoders; there is no control using the same encoder with random weights or with VICReg trained on a non-representative unlabeled corpus. This control is essential because Table 3 shows that the LowPass augmentation alone matches the full augmentation family (87.00 vs 87.00 on Watkin's, 55.93 vs 55.61 on Deepship), suggesting that a large part of the class signal may be present in the input spectrogram itself. Please add (a) linear-probe accuracy on randomly initialized Conformer and ResNet18 encoders and (b) a VICReg training run on a mismatched unlabeled corpus (e.g., white noise or a different hydrophone) to attribute the observed accuracy to the contrastive objective and the specific ONC data.
  2. [Sections 3.2-3.3, Tables 2-4] The VICReg loss weights (Table 2), the augmentation family choice (Table 3), and the decision to expand the augmentation family (Table 4) are selected using accuracy on the Deepship test set, which is also used in the final evaluation in Table 1. This is test-set selection and inflates the reported generalization numbers, undermining the claim that the embeddings generalize without dataset-specific tuning. Use a held-out validation split (separate from the test sets) or nested cross-validation to select hyperparameters, and report the corresponding final accuracies. This is load-bearing because the selected configuration is then presented as the proposed method's performance.
  3. [Equation (2)] The variance term in the VICReg loss is written as abs(γ − S(z^j, ε)), but the original VICReg formulation (Bardes et al., 2021) uses max(0, γ − S(z^j, ε)), a hinge function that only penalizes standard deviations below the target γ. The absolute value also penalizes standard deviations above γ, which is not part of VICReg and can change the learned representation. If the implementation matches the equation, this is a substantive deviation; if not, the equation misleads. Please correct the formula and, if the implementation used the absolute value, report whether the results change under the correct VICReg variance term.
  4. [All result tables (Tables 1, 3, 4, Figures 5, 7, 8)] All reported accuracy and F1 scores are point estimates from a single training run. Several key comparisons involve small differences, e.g., Deepship 54.87 vs 55.16 and Watkin's 86.10 vs 87.10 between unsupervised and supervised Conformer in Table 1. Without multiple random seeds and error bars or a significance test, the claims of 'comparable performance' and 'more robust' are not statistically supported. Please report mean ± standard deviation over at least three independent runs, or provide a paired significance test for the main comparisons.
  5. [Section 2.1.1] The load-bearing premise that unlabeled recordings from a single ONC hydrophone near Vancouver are sufficiently representative of the acoustic conditions in all three labeled benchmarks—including ShipsEar off Spain and the decades-spanning Watkins database—is never tested empirically. The paper provides no spectral or distributional analysis comparing the unlabeled data to the evaluation datasets, and no experiment with an alternative unlabeled corpus. This makes it difficult to rule out the possibility that the cross-dataset accuracy reflects generic spectrogram statistics rather than genuinely transferable embeddings learned from the ONC data. Please add an analysis of the acoustic feature distributions or a sensitivity experiment using a different unlabeled corpus.
minor comments (6)
  1. [Equation (8)] The symbol v is used both for the variance component function v(Z) and as the scalar weight for the covariance term in Eq. (8). Rename the scalar weight ν (or similar) to avoid confusing the variance function with the covariance weight.
  2. [Table 3 caption] The caption says 'The influence of the augmentation functions on the classification of Ship types,' but the table also reports results on the Watkin's marine mammal dataset. Adjust the caption to reflect that both ship type and marine mammal classification are evaluated.
  3. [Section 2.2, item 2] The Gaussian noise augmentation constrains the output SNR between 0.3 dB and 0.5 dB, which would make the noise overwhelmingly stronger than the signal. Please confirm that this range is intentional or whether the SNR values are defined differently (e.g., SNR in dB of the noise added relative to the signal).
  4. [Abstract and Section 1] The sentence 'This shows to potential of unsupervised methods' contains a typo; it should be 'This shows the potential of unsupervised methods.'
  5. [Figure 2b caption] The caption reads 'Visualization fo the proposed augmentation functions'; 'fo' should be 'of.'
  6. [Section 1, contribution 1] The claim 'The first implementation of unsupervised CL on a separate unlabeled underwater acoustic dataset' is strong given that the reference list includes self-supervised work on underwater acoustics (e.g., Xu et al., 2024). Suggest softening to 'to the best of our knowledge' and clarifying the precise distinction from prior self-supervised underwater acoustic studies.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: embeddings are trained without labels on separate ONC data and evaluated on held-out labeled benchmarks; the only self-citation (augmentation design from authors' prior survey) is empirically validated, not load-bearing.

full rationale

The central derivation chain is self-contained against external benchmarks. Unlabeled audio from a single ONC hydrophone is used to train a Conformer encoder with the VICReg loss (no labels, Sections 2.1.1, 2.5); the frozen embeddings are then used to fit a logistic regression on labeled training splits of Deepship, ShipsEar, and Watkins, and accuracy is reported on genuinely held-out test splits (Deepship time-wise future data; the other two 80/20 splits, Section 2.1.2). No fitted quantity is renamed as a prediction: the test accuracies in Table 1 are out-of-sample measurements, and the label information from the benchmarks never enters the unsupervised training stage. The only self-citation is the authors' own prior survey (Hummel et al., 2024) used to justify the 50 Hz-1 kHz low-pass augmentation (Section 2.2). This is not load-bearing: the claim is externally falsifiable and is in fact tested against the external benchmarks in the Table 3 ablation, where LowPass alone roughly matches the full family. The choice of the ONC hydrophone to match the Deepship site (Section 2.1.1) shapes representativeness, but it does not force any result by construction: the ShipsEar and Watkins results are cross-site, and the Deepship test set covers ships not seen in the pretraining period under a time-wise split. Concerns about missing controls (random-initialized Conformer or a non-representative unlabeled corpus) bear on causal attribution of the accuracy numbers to the contrastive objective, which is an experimental-design risk rather than a reduction of the claimed result to its own inputs. The paper therefore does not exhibit the specific reductions required to establish circularity.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

No new entities are introduced. The model is built from existing components, and the central assumptions are about data relevance, windowing, augmentation validity, and the adequacy of the evaluation probe.

free parameters (7)
  • VICReg loss weights (lambda, mu, v) = lambda=5, mu=5, v=1
    Selected by comparing three combinations from Bardes et al. (2021) on Deepship accuracy in Table 2; the 25/25/1 combination performs much worse, so the choice is data-dependent.
  • Low-pass filter cutoff frequency = 1 kHz
    Chosen based on the authors' own survey claim that discriminative ship information lies between 50 Hz and 1 kHz; this drives the main augmentation and encodes prior knowledge into training.
  • Gaussian noise SNR range = 0.3-0.5 dB
    Hand-selected range for the noise augmentation; no ablation across SNR levels is reported.
  • Mixup temporal sampling sigma = 50 seconds
    Hand-selected width of the temporal neighborhood for drawing the second mixup sample; no sensitivity analysis is reported.
  • Number of Mel filters = 128
    Standard choice for spectrogram input; not ablated in the paper.
  • Embedding dimension = 2048
    Ablated in Section 3.5; accuracy saturates around 512 dimensions, so 2048 is conservative but not independently justified.
  • Batch size and LARS learning rate = batch=2048, LR=0.01
    Chosen to stabilize large-batch training with the LARS optimizer; no systematic tuning is reported.
assumptions (5)
  • domain assumption Unlabeled recordings from the single ONC hydrophone near Vancouver are representative for the acoustic conditions in all three labeled benchmarks (Deepship, ShipsEar, Watkins).
    Invoked in Section 2.1.1: the site is chosen to align with the Deepship benchmark, but no distributional analysis supports transfer to ShipsEar or Watkins.
  • domain assumption Two-second non-overlapping windows preserve the class-discriminative information for ship type and marine mammal recognition.
    Adopted in Section 2.1.3; no ablation over window length is reported.
  • domain assumption The four augmentations (identity, Gaussian noise, low-pass filter, temporal mixup) produce valid positive pairs that do not change the latent class identity.
    Defined in Section 2.2; the mixup strategy in particular mixes a low-pass filtered anchor with a high-pass filtered neighboring sample, and the paper does not verify that this preserves class identity.
  • domain assumption Logistic regression on frozen embeddings is a sufficient probe for measuring embedding quality and generalization.
    Section 2.6 uses logistic regression as the only downstream classifier; results may depend on this probe choice.
  • domain assumption VICReg's variance and covariance regularization prevent representation collapse for spectrogram inputs as they do for images.
    Section 3.2 tunes loss weights from Bardes et al. (2021); the paper assumes the collapse-prevention mechanism carries over to audio.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Computation of Generalized Embeddings for Underwater Acoustic Target Recognition using Contrastive Learning." pith.science (2026). https://pith.science/paper/W2MAEP63

@misc{pith2026250512904,
  author       = {Pith},
  title        = {Pith review of: The Computation of Generalized Embeddings for Underwater Acoustic Target Recognition using Contrastive Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W2MAEP63}},
  note         = {Machine review of arXiv:2505.12904}
}
read the original abstract

The increasing level of sound pollution in marine environments poses an increased threat to ocean health, making it crucial to monitor underwater noise. By monitoring this noise, the sources responsible for this pollution can be mapped. Monitoring is performed by passively listening to these sounds. This generates a large amount of data records, capturing a mix of sound sources such as ship activities and marine mammal vocalizations. Although machine learning offers a promising solution for automatic sound classification, current state-of-the-art methods implement supervised learning. This requires a large amount of high-quality labeled data that is not publicly available. In contrast, a massive amount of lower-quality unlabeled data is publicly available, offering the opportunity to explore unsupervised learning techniques. This research explores this possibility by implementing an unsupervised Contrastive Learning approach. Here, a Conformer-based encoder is optimized by the so-called Variance-Invariance-Covariance Regularization loss function on these lower-quality unlabeled data and the translation to the labeled data is made. Through classification tasks involving recognizing ship types and marine mammal vocalizations, our method demonstrates to produce robust and generalized embeddings. This shows to potential of unsupervised methods for various automatic underwater acoustic analysis tasks.

Figures

Figures reproduced from arXiv: 2505.12904 by the authors.

Figure 1
Figure 1. The proposed architecture for the unsupervised framework. The framework starts [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Visualization of the augmentation family proposed for defining positive samples [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. A single Conformer Block. 7 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: 3.1. Multi-Purpose Classification [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 4
Figure 4. Figure 4: Overview of the experiments. working with very large batch sizes, all further backbones were optimized using the LARS optimizer. Further comparison is made between the ResNet18 model and the Conformer model. As the Conformer model has more parameters and is built speci…
Figure 5
Figure 5. Figure 5: Accuracy of supervised baseline models trained on either time-wise split or random [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Temporal Variation of Deepship and ShipsEar, showing the mean and the [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: The accuracy on Deepship when trained on reduced amount of labeled data. [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]
Figure 8
Figure 8. Figure 8: The accuracy of the unsupervised conformer model on Deepship over the embedding [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 20 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    , author Ponce, J

    author Bardes, A. , author Ponce, J. , author LeCun, Y. , year 2021 . title Vicreg: Variance-invariance-covariance regularization for self-supervised learning . journal arXiv preprint arXiv:2105.04906

    arXiv 2021

  3. [3]

    , year 2007

    author Canada, O.N. , year 2007 . title Dataportal . howpublished https://data.oceannetworks.ca/home . note Accessed: 1 May 2023

    work page 2007

  4. [4]

    , author Kornblith, S

    author Chen, T. , author Kornblith, S. , author Norouzi, M. , author Hinton, G. , year 2020 . title A simple framework for contrastive learning of visual representations , in: booktitle International conference on machine learning , organization PMLR . pp. pages 1597--1607

    2020

  5. [5]

    , author Ortego, D

    author Fonseca, E. , author Ortego, D. , author McGuinness, K. , author O’Connor, N.E. , author Serra, X. , year 2021 . title Unsupervised contrastive learning of sound event representations , in: booktitle ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , organization IEEE . pp. pages 371--375

    work page 2021

  6. [6]

    , author Strub, F

    author Grill, J.B. , author Strub, F. , author Altch \'e , F. , author Tallec, C. , author Richemond, P. , author Buchatskaya, E. , author Doersch, C. , author Avila Pires, B. , author Guo, Z. , author Gheshlaghi Azar, M. , et al., year 2020 . title Bootstrap your own latent-a new approach to self-supervised learning . journal Advances in neural informati...

    2020

  7. [7]

    , author Qin, J

    author Gulati, A. , author Qin, J. , author Chiu, C.C. , author Parmar, N. , author Zhang, Y. , author Yu, J. , author Han, W. , author Wang, S. , author Zhang, Z. , author Wu, Y. , et al., year 2020 . title Conformer: Convolution-augmented transformer for speech recognition . journal arXiv preprint arXiv:2005.08100

    arXiv 2020

  8. [8]

    , author Pham, M.T

    author Hamard, Q. , author Pham, M.T. , author Cazau, D. , author Heerah, K. , year 2024 . title A deep learning model for detecting and classifying multiple marine mammal species from passive acoustic data . journal Ecological Informatics , pages 102906

    work page 2024

Show all 30 references
  1. [9]

    , author Fan, H

    author He, K. , author Fan, H. , author Wu, Y. , author Xie, S. , author Girshick, R. , year 2020 . title Momentum contrast for unsupervised visual representation learning , in: booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. p...

    2020

  2. [10]

    , author van der Mei, R.D

    author Hummel, H.I. , author van der Mei, R.D. , author Bhulai, S. , year 2024 . title A survey on machine learning in ship radiated noise . journal Ocean Engineering volume 298 , pages 117252

    2024

  3. [11]

    , author Jiangbin, Z

    author Irfan, M. , author Jiangbin, Z. , author Ali, S. , author Iqbal, M. , author Masood, Z. , author Hamid, U. , year 2021 . title Deepship: An underwater acoustic benchmark dataset and a separable convolution based autoencoder for classification . journal Expert Systems wi...

    2021

  4. [12]

    , author Wood, C.M

    author Kahl, S. , author Wood, C.M. , author Eibl, M. , author Klinck, H. , year 2021 . title Birdnet: A deep learning solution for avian diversity monitoring . journal Ecological Informatics volume 61 , pages 101236

    2021

  5. [13]

    , author Teterwak, P

    author Khosla, P. , author Teterwak, P. , author Wang, C. , author Sarna, A. , author Tian, Y. , author Isola, P. , author Maschinot, A. , author Liu, C. , author Krishnan, D. , year 2020 . title Supervised contrastive learning . journal Advances in neural information processi...

    2020

  6. [14]

    , author Carbone, D

    author Licciardi, A. , author Carbone, D. , year 2024 . title Whalenet: A novel deep learning architecture for marine mammals vocalizations on watkins marine mammal sound database . journal IEEE Access volume 12 , pages 154182--154194

    2024

  7. [15]

    , author Chen, Y

    author Lin, L. , author Chen, Y. , author Wang, F. , year 2024 . title Underwater passive target recognition based on self-supervised contrastive learning , in: booktitle 2024 3rd International Conference on Artificial Intelligence, Internet of Things and Cloud Computing Techn...

    2024

  8. [16]

    , author Reermann, J

    author Müller, N. , author Reermann, J. , author Meisen, T. , year 2024 . title Navigating the depths: A comprehensive survey of deep learning for passive underwater acoustic target recognition . journal IEEE Access volume 12 , pages 154092--154118

    2024

  9. [17]

    , author Li, C

    author Nie, L. , author Li, C. , author Wang, H. , author Wang, J. , author Zhang, Y. , author Yin, F. , author Marzani, F. , author Bozorg Grayeli, A. , year 2023 . title A contrastive-learning-based method for the few-shot identification of ship-radiated noises . journal Jou...

    2023

  10. [18]

    , author Takeuchi, D

    author Niizumi, D. , author Takeuchi, D. , author Ohishi, Y. , author Harada, N. , author Kashino, K. , year 2022 . title Byol for audio: Exploring pre-trained general-purpose audio representations . journal IEEE/ACM Transactions on Audio, Speech, and Language Processing volum...

    2022

  11. [19]

    , author Grangier, D

    author Saeed, A. , author Grangier, D. , author Zeghidour, N. , year 2021 . title Contrastive learning of general-purpose audio representations , in: booktitle ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. pages 3875...

    2021

  12. [20]

    , author Torres-Guijarro, S

    author Santos-Dom \' nguez, D. , author Torres-Guijarro, S. , author Cardenal-L \'o pez, A. , author Pena-Gimenez, A. , year 2016 . title Shipsear: An underwater vessel noise database . journal Applied Acoustics volume 113 , pages 64--69

    2016

  13. [21]

    , author Daher, M.A

    author Sayigh, L. , author Daher, M.A. , author Allen, J. , author Gordon, H. , author Joyce, K. , author Stuhlmann, C. , author Tyack, P. , year 2016 . title The watkins marine mammal sound database: An online, freely accessible resource , in: booktitle Proceedings of Meeting...

    2016

  14. [22]

    , author Wang, Y

    author Srivastava, S. , author Wang, Y. , author Tjandra, A. , author Kumar, A. , author Liu, C. , author Singh, K. , author Saraf, Y. , year 2022 . title Conformer-based self-supervised learning for non-speech audio tasks , in: booktitle ICASSP 2022-2022 IEEE International Co...

    2022

  15. [23]

    , author Luo, X

    author Sun, B. , author Luo, X. , year 2023 . title Underwater acoustic target recognition based on automatic feature and contrastive coding . journal IET Radar, Sonar & Navigation volume 17 , pages 1277--1285

    2023

  16. [24]

    , author Martin, B

    author Thomas, M. , author Martin, B. , author Kowarski, K. , author Gaudet, B. , author Matwin, S. , year 2020 . title Marine mammal species classification using convolutional neural networks and a novel acoustic representation , in: booktitle Machine Learning and Knowledge D...

    2020

  17. [25]

    , author Bai, D

    author Tian, S. , author Bai, D. , author Zhou, J. , author Fu, Y. , author Chen, D. , year 2023 . title Few-shot learning for joint model in underwater acoustic target recognition . journal Scientific Reports volume 13 , pages 17502

    2023

  18. [26]

    , author Eytan, D

    author Tonekaboni, S. , author Eytan, D. , author Goldenberg, A. , year 2021 . title Unsupervised representation learning for time series with temporal neighborhood coding . journal arXiv preprint arXiv:2106.00750

    2021 arXiv

  19. [27]

    , author Hamer, J

    author Van Merri \"e nboer, B. , author Hamer, J. , author Dumoulin, V. , author Triantafillou, E. , author Denton, T. , year 2024 . title Birds, bats and beyond: Evaluating generalization in bioacoustics models . journal Frontiers in Bird Science volume 3 , pages 1369756

    2024

  20. [28]

    , author Ren, J

    author Xie, Y. , author Ren, J. , author Xu, J. , year 2023 . title Guiding the underwater acoustic target recognition with interpretable contrastive learning , in: booktitle OCEANS 2023-Limerick , organization IEEE . pp. pages 1--6

    2023

  21. [29]

    , author Jiang, J

    author Xu, Q. , author Jiang, J. , author Xu, K. , author Dou, Y. , author Gao, C. , author Zhu, B. , author You, K. , author Zhu, Q. , year 2024 . title Self-supervised learning-for underwater acoustic signal classification with mixup . journal IEEE Journal of Selected Topics...

    2024

  22. [30]

    , author Gitman, I

    author You, Y. , author Gitman, I. , author Ginsburg, B. , year 2017 . title Scaling SGD batch size to 32k for imagenet training . journal CoRR volume abs/1708.03888

    2017 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.