Pith. sign in

REVIEW 3 major objections 5 minor 49 references

Quantitative Analysis of Proxy Tasks for Anomalous Sound Detection

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Source separation is the only proxy task whose improvements consistently improve anomalous sound detection.

desk verdict Useful negative map of proxy-task/ASD correlations, but the only positive result (source separation) is confounded with model capacity and needs controlled experiments. read the letter →

arxiv 2601.08480 v2 pith:O47AZ6VY submitted 2026-01-13 eess.AS

Seunghyeon Shin , Seokjin Lee This is my paper
classification eess.AS
keywords anomaloussounddetectionproxytasksself-supervisedlearningsourceseparationlinearprobeMahalanobisdistancetaskalignmentrepresentation
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 improving a self-supervised proxy task—autoencoding, classification, source separation, contrastive learning, or pretrained feature extraction—actually improves anomalous sound detection (ASD). Across dozens of configurations on two public machine-sound datasets, it measures each task's own metric and correlates it with ASD performance measured by linear probes and Mahalanobis distance. The central finding is that the usual assumption does not hold in general: only source separation shows a strong, consistent positive correlation, while classification saturates at near-perfect accuracy without helping detection, contrastive learning collapses as model size grows, autoencoders correlate only with distributional compactness, and pretrained model accuracy does not transfer. If true, proxy metrics cannot be trusted as surrogates for ASD performance, and designing proxy tasks requires checking both task difficulty and objective alignment.

What carries the argument

The central instrument is the correlation analysis between proxy-task metrics and ASD metrics across controlled capacity variations. Proxy metrics are the AE reconstruction MAE, classification Macro-F1, source-separation SI-SDRi, contrastive alignment/uniformity, and pretrained AudioSet mAP. ASD performance is scored by a linear probe (trained on evaluation-set anomalous labels) and by Mahalanobis distance on normal training features, both summarized as AUC. Spearman rank correlation across configurations isolates whether proxy improvements track detection gains; the source-separation experiments vary conformer-block count and channel width, producing a monotone proxy-ASD relationship that t

What would settle it

Train separation models with identical architecture and capacity but differing separation quality—for example, by varying the training SNR range or adding a reconstruction auxiliary loss—and check whether ASD AUC follows SI-SDRi; a flat or non-monotonic ASD curve across SI-SDRi values would refute the paper's central correlation claim.

Watch

Extended reading notes

Core claim

The paper claims that improved proxy task performance does not necessarily improve anomalous sound detection performance. It reports that source separation is the only task among the five studied whose proxy metric (SI-SDRi) strongly and consistently correlates with ASD performance across all evaluation protocols (Spearman rho above 0.95, p < 0.01), such that improved separation quality translates into better detection. Classification tasks saturate at F1-scores above 97%, losing discriminative power; contrastive learning fails to learn meaningful features due to limited data diversity and collapses as capacity grows; autoencoders correlate only with Mahalanobis-distance compactness, not lin

Load-bearing premise

The source-separation correlation is measured by scaling model size, so if larger capacity alone drives both separation and detection improvements, the claim that improved separation quality itself drives ASD gains would collapse.

Editorial extensions

If this is right

  • Proxy-task metrics should not be used alone as surrogates for ASD performance; a task like classification can reach ceiling accuracy while yielding poor detection features.
  • For reconstruction-based ASD, the common assumption that better reconstruction implies better detection is only partially supported: reconstruction quality tracks distributional compactness (Mahalanobis distance) but not linear separability.
  • Source separation stands out as a proxy whose optimization directly benefits anomaly detection, suggesting separation-style objectives are a safer design choice for ASD systems.
  • Contrastive learning on small, low-diversity acoustic datasets is prone to feature collapse, so capacity scaling without data diversity can degrade ASD performance.
  • The proposed three-stage protocol—checking proxy-task health, evaluating representation suitability, and running correlation analysis across hyperparameters—provides a concrete diagnostic for future ASD proxy-task design.

Reading between the lines

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

  • The source-separation correlation is measured by scaling model capacity, so the claim that 'improved separation improves detection' may partly reflect a capacity effect; a test that varies separation quality at fixed capacity would clarify causality.
  • The linear-probe evaluation trains on anomalous labels from the evaluation set, so the reported LP scores measure a semi-supervised separability rather than pure unsupervised ASD; Mahalanobis distance is the closer approximation to the true unsupervised setting.
  • The separation model's own reconstruction residual could be used directly as an anomaly score, potentially making the proxy and target objectives the same; this extension is not tested in the paper.
  • The 'task difficulty' explanation predicts an inverted-U relationship: tasks too easy saturate while tasks too hard collapse; adding a curriculum or SNR control could test whether a middle difficulty yields the strongest proxy-ASD alignment.
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

3 major / 5 minor

Summary. The paper investigates whether optimizing proxy-task metrics (AE reconstruction error, classification accuracy, SI-SDRi, contrastive alignment/uniformity, and AudioSet mAP) transfers to anomalous sound detection (ASD) performance measured by linear-probe AUC and Mahalanobis-distance AUC on ToyADMOS2/MIMII under the DCASE 2022 setup. Across five proxy-task families and 41 configurations, it finds that strong proxy performance does not generally improve ASD: classification saturates, contrastive learning collapses or underperforms, and only source separation shows a strong positive Spearman correlation. The paper additionally proposes a three-stage alignment verification protocol for designing proxy tasks.

Significance. If the source-separation result were causal, the paper would be a valuable contribution: it would systematically challenge the common assumption that better proxy performance implies better ASD, and its three-stage protocol would give practitioners a concrete design checklist. The study is broad in scope and thoughtfully uses both a supervised linear probe and an unsupervised Mahalanobis score, with exact Spearman p-values. It also offers a useful taxonomy of failure modes (saturation, collapse, task misalignment). The main limitation is that the sole positive correlation is confounded with model capacity and the ASD feature-vector dimensionality, so the paper's headline claim is not yet established.

major comments (3)
  1. [Section III-D, Table IV] The headline positive result—that source separation is the only task with a strong positive proxy–ASD correlation—is confounded. The eight configurations vary the number of conformer blocks N ∈ {0,1,2,4} and channel width ∈ {64,128}. Two things change simultaneously: model capacity, and the ASD feature representation, which concatenates channel-pooled features from N+1 extraction points (Sec. III-D), so the ASD feature dimensionality grows with N. The reported Spearman coefficients (ρ ≈ 0.95–0.98, n=8) therefore cannot distinguish 'better separation causes better ASD' from 'a larger/richer feature extractor yields better ASD.' In fact, Table IV is not even consistently monotonic: 2-block/128-ch has higher SI-SDRi (3.57 dB) than 1-block/128-ch (3.40 dB) but lower MD AUC (59.24% vs 60.51%). To support the causal-sounding claim, the paper needs to control for capacity/feature dimension—for
  2. [Section IV-B1, Eqs. (1)-(4)] The LP protocol trains a linear classifier on the evaluation set with access to both normal and anomalous labels. Thus the 'ASD performance' reported for LP is not an unsupervised detector; it is a supervised representation-quality probe. The abstract and conclusions state the findings in terms of 'anomalous sound detection performance' without this caveat. This matters because the source-separation positive correlation is reported for LP as well as MD. The paper should either consistently describe LP as a supervised measure of linear separability and restrict 'unsupervised ASD' claims to the MD results, or provide evidence that LP correlates with a fully unsupervised scoring on the same features.
  3. [Tables II-VII] The quantitative conclusions are based on single runs with no seeds or error bars. All AUC, F1, and SI-SDRi values are point estimates, and the Spearman correlations are computed over only n=4–10 configurations. For example, the AE in-domain LP AUC differences of about 1% (Table II) are interpreted as trends, but run-to-run variability could be of the same magnitude. The exact p-values do not account for this. Reporting multiple random seeds with mean and standard deviation, or at least a sensitivity analysis, would substantially strengthen the reliability of the correlation findings.
minor comments (5)
  1. [Table VII] For contrastive learning, it is unclear whether the reported Spearman ρ uses alignment or uniformity as the proxy metric. The footnote explains the sign convention for 'low is better' metrics, but the row labels 'Contrastive (SimCLR)' and 'Contrastive (SimSiam)' do not state which metric was used.
  2. [Figs. 2-4] The captions refer to 'normalized proxy task performance,' but no normalization formula is given. Please specify the normalization (e.g., min-max across configurations) and whether it is applied per proxy task or globally.
  3. [Tables II and IV] The column header 'Mahalanobis(↑)' is ambiguous: it is later described as AUC, but the header does not say so. Please make the metric explicit in each table.
  4. [Section III-D] The architecture is cited to the authors' previous work [27]. For reproducibility, briefly state which hyperparameters (conformer blocks, channel widths, training budget) are newly varied here and whether the remaining settings are identical to [27].
  5. [Section VI-A] The claim 'consistent positive correlation across all evaluation metrics (ρ > 0.95, p < 0.01)' should be qualified by the small n and the capacity confound discussed above; currently the sentence appears before the confound is acknowledged.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: proxy metrics and ASD metrics are measured independently; the one self-citation is a design choice, not a load-bearing circular step.

full rationale

The paper's central empirical claim is based on independent measurements: proxy-task performance is quantified by reconstruction MAE (Eq. 6), classification F1 (Eq. 7), SI-SDRi (Eq. 8), or mAP from external pretraining, while ASD performance is computed from frozen representations h=F(x) using linear probe (Eqs. 1–4) or Mahalanobis distance (Eq. 5) on held-out evaluation sections. No fitted parameter is later renamed as a prediction; model selection uses training loss only, and the LP explicitly trains on labeled evaluation data, which the paper itself acknowledges ('Unlike LP, which requires a labeled subset of the evaluation data'). The only author self-citation, [27], supplies the source-separation architecture, but the paper does not invoke [27] as evidence for the correlation—it is a network design choice. The proposed three-stage verification protocol is a recommendation and is not used to generate the reported results. The correlation in Table IV may be vulnerable to a capacity/feature-dimension confound because the ASD feature concatenates N+1 extraction points as N grows, but that is a threat to causal interpretation, not a circular derivation in which an output equals an input by construction. Therefore no circularity step meeting the quoted-evidence standard is present.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

No theoretical constants are fitted and no new entities are introduced. The entries above are the experimental choices and domain assumptions the correlation claims rest on; none are independently validated with ablations or code release.

free parameters (5)
  • Source separation model capacity = conformer blocks 0,1,2,4; channels 64,128
    Capacity is varied to change SI-SDRi; this confounds the proxy-ASD correlation.
  • AE bottleneck/hidden dimensions = latent 4,8,16; hidden 64,128,256
    Nine capacity configurations; MAE and AUC both vary with capacity.
  • Classification backbone and loss = ResNet-18/34/50/101/152 x CE/ArcFace
    Ten configurations; F1 saturates, so capacity effects on AUC are visible.
  • Contrastive framework and augmentations = SimCLR/SimSiam with hand-picked augmentations; ResNet variants
    Augmentations and hyperparameters chosen by hand; observed collapse may be a tuning artifact.
  • Pretrained model selection = BEATs-iter3/iter3+, EAT-base/large
    Only four models; mAP is taken from original papers as the proxy metric.
assumptions (6)
  • domain assumption Anomalous inputs produce higher reconstruction error in AEs trained on normal data.
    Foundational premise of AE-based ASD (Sec. III-B); the paper tests only the extension that lower MAE improves ASD.
  • domain assumption DCASE 2022 Task 2 setup on ToyADMOS2/MIMII is representative of ASD; findings generalize beyond seven machine types.
    Authors concede limited generalizability in Sec. VII.
  • domain assumption LP trained with normal+anomalous labels and Mahalanobis distance on normal features are valid measures of ASD capability.
    LP uses anomalous labels (Sec. IV-B1); this is a supervised evaluation, not purely unsupervised ASD.
  • domain assumption The proxy metrics (MAE, MacroF1, SI-SDRi, alignment/uniformity, AudioSet mAP) faithfully reflect proxy-task quality.
    For pretrained models, mAP is taken from original papers (Sec. IV-C).
  • ad hoc to paper Capacity variations do not systematically confound the proxy-ASD correlation.
    The source-separation correlation is observed while model size changes; no control experiment holds capacity fixed (Sec. V-C).
  • domain assumption A single training run per configuration without seeds is adequate to estimate AUC and correlations.
    No repeated seeds or error bars reported (Tables II-VI).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantitative Analysis of Proxy Tasks for Anomalous Sound Detection." pith.science (2026). https://pith.science/paper/O47AZ6VY

@misc{pith2026260108480,
  author       = {Pith},
  title        = {Pith review of: Quantitative Analysis of Proxy Tasks for Anomalous Sound Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O47AZ6VY}},
  note         = {Machine review of arXiv:2601.08480}
}
read the original abstract

Anomalous sound detection (ASD) typically involves self-supervised proxy tasks to learn feature representations from normal sound data, owing to the scarcity of anomalous samples. In ASD research, proxy tasks such as AutoEncoders operate under the explicit assumption that models trained on normal data will increase the reconstruction errors related to anomalies. A natural extension suggests that improved proxy task performance should improve ASD capability; however, this relationship has received little systematic attention. This study addresses this research gap by quantitatively analyzing the relationship between proxy task metrics and ASD performance across five configurations, namely, AutoEncoders, classification, source separation, contrastive learning, and pre-trained models. We evaluate the learned representations using linear probe (linear separability) and Mahalanobis distance (distributional compactness). Our experiments reveal that strong proxy performance does not necessarily improve anomalous sound detection performance. Specifically, classification tasks experience performance saturation owing to insufficient task difficulty, whereas contrastive learning fails to learn meaningful features owing to limited data diversity. Notably, source separation is the only task demonstrating a strong positive correlation, such that improved separation consistently improves anomaly detection. Based on these findings, we highlight the critical importance of task difficulty and objective alignment. Finally, we propose a three-stage alignment verification protocol to guide the design of highly effective proxy tasks for ASD systems.

Figures

Figures reproduced from arXiv: 2601.08480 by the authors.

Figure 1
Figure 1. Dataset configuration of the DCASE 2022 challenge. (a) Overall [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Relationship between the proxy task and in-domain ASD perfor [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Relationship between the proxy task and out-domain ASD perfor [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Representations of the gearbox class from the trained AE (a)–(b) [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

49 extracted references · 8 linked inside Pith

  1. [1]

    Scream and gunshot detection and localization for audio-surveillance systems,

    G. Valenzise, L. Gerosa, M. Tagliasacchi, F. Antonacci, and A. Sarti, “Scream and gunshot detection and localization for audio-surveillance systems,” in2007 IEEE Conference on Advanced Video and Signal Based Surveillance, 2007, pp. 21–26

    2007

  2. [2]

    De- tection and analysis of abnormal situations through fear-type acoustic manifestations,

    C. Clavel, L. Devillers, G. Richard, I. Vasilescu, and T. Ehrette, “De- tection and analysis of abnormal situations through fear-type acoustic manifestations,” in2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP ’07, vol. 4, 2007, pp. IV–21– IV–24

    2007

  3. [3]

    Unsu- pervised detection of anomalous sound based on deep learning and the neyman–pearson lemma,

    Y . Koizumi, S. Saito, H. Uematsu, Y . Kawachi, and N. Harada, “Unsu- pervised detection of anomalous sound based on deep learning and the neyman–pearson lemma,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 27, no. 1, pp. 212–224, 2018

    2018

  4. [4]

    A review on machinery di- agnostics and prognostics implementing condition-based maintenance,

    A. K. Jardine, D. Lin, and D. Banjevic, “A review on machinery di- agnostics and prognostics implementing condition-based maintenance,” Mechanical Systems and Signal Processing, vol. 20, no. 7, pp. 1483– 1510, 2006

    2006

  5. [5]

    Description and discussion on DCASE2020 challenge task2: Unsuper- vised anomalous sound detection for machine condition monitoring,

    Y . Koizumi, Y . Kawaguchi, K. Imoto, T. Nakamura, Y . Nikaido, R. Tan- abe, H. Purohit, K. Suefusa, T. Endo, M. Yasuda, and N. Harada, “Description and discussion on DCASE2020 challenge task2: Unsuper- vised anomalous sound detection for machine condition monitoring,” in Proceedings of the Detection and Classification of Acoustic Scenes and Events 2020 W...

    2020

  6. [6]

    Complementary set varia- tional autoencoder for supervised anomaly detection,

    Y . Kawachi, Y . Koizumi, and N. Harada, “Complementary set varia- tional autoencoder for supervised anomaly detection,” in2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018, pp. 2366–2370

    2018

  7. [7]

    Non-linear prediction with lstm recurrent neural networks for acoustic novelty detection,

    E. Marchi, F. Vesperini, F. Weninger, F. Eyben, S. Squartini, and B. Schuller, “Non-linear prediction with lstm recurrent neural networks for acoustic novelty detection,” in2015 International Joint Conference on Neural Networks (IJCNN), 2015, pp. 1–7

    2015

  8. [8]

    Self-supervised classification for detecting anomalous sounds,

    R. Giri, S. V . Tenneti, F. Cheng, K. Helwani, U. Isik, and A. Kr- ishnaswamy, “Self-supervised classification for detecting anomalous sounds,” inProceedings of the Detection and Classification of Acous- tic Scenes and Events 2020 Workshop (DCASE2020), Tokyo, Japan, November 2020, pp. 46–50

    2020

Show all 49 references
  1. [9]

    Self-supervised acoustic anomaly detec- tion via contrastive learning,

    H. Hojjati and N. Armanfard, “Self-supervised acoustic anomaly detec- tion via contrastive learning,” inICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 3253–3257

    2022

  2. [10]

    Activity-guided industrial anomalous sound detection against interferences,

    Y . Lee, J. Kim, and J. Ok, “Activity-guided industrial anomalous sound detection against interferences,”arXiv preprint arXiv:2409.01885, 2024

    2024 arXiv

  3. [11]

    A novel application of deep transfer learning with audio pre-trained models in pump audio fault detection,

    A. A. T. Anvar and H. Mohammadi, “A novel application of deep transfer learning with audio pre-trained models in pump audio fault detection,” Computers in Industry, vol. 147, p. 103872, 2023

    2023

  4. [12]

    A novel approach for automatic acoustic novelty detection using a denoising autoencoder with bidirectional lstm neural networks,

    E. Marchi, F. Vesperini, F. Eyben, S. Squartini, and B. Schuller, “A novel approach for automatic acoustic novelty detection using a denoising autoencoder with bidirectional lstm neural networks,” in2015 IEEE International Conference on Acoustics, Speech and Signal Processing ...

    2015

  5. [13]

    Arcface: Additive angular margin loss for deep face recognition,

    J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “Arcface: Additive angular margin loss for deep face recognition,” in2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 4685– 4694

    2019

  6. [14]

    Revisiting self-supervised visual representation learning,

    A. Kolesnikov, X. Zhai, and L. Beyer, “Revisiting self-supervised visual representation learning,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 1920–1929

    2019

  7. [15]

    A simple unified framework for detecting out-of-distribution samples and adversarial attacks,

    K. Lee, K. Lee, H. Lee, and J. Shin, “A simple unified framework for detecting out-of-distribution samples and adversarial attacks,”Advances in neural information processing systems, vol. 31, 2018

    2018

  8. [16]

    Events detection for an audio- based surveillance system,

    C. Clavel, T. Ehrette, and G. Richard, “Events detection for an audio- based surveillance system,” in2005 IEEE International Conference on Multimedia and Expo, 2005, pp. 1306–1309

    2005

  9. [17]

    Audio surveillance of roads: A system for detecting anomalous sounds,

    P. Foggia, N. Petkov, A. Saggese, N. Strisciuglio, and M. Vento, “Audio surveillance of roads: A system for detecting anomalous sounds,”IEEE Transactions on Intelligent Transportation Systems, vol. 17, no. 1, pp. 279–288, 2016

    2016

  10. [18]

    Recognizing abnormal heart sounds using deep learning,

    J. Rubin, R. Abreu, A. Ganguli, S. Nelaturi, I. Matei, and K. Sricha- ran, “Recognizing abnormal heart sounds using deep learning,”arXiv preprint arXiv:1707.04642, 2017

    2017 arXiv

  11. [19]

    Fault detection and diagnosis of railway point machines by sound analysis,

    J. Lee, H. Choi, D. Park, Y . Chung, H.-Y . Kim, and S. Yoon, “Fault detection and diagnosis of railway point machines by sound analysis,” Sensors, vol. 16, no. 4, 2016

    2016

  12. [20]

    A multimodal anomaly detector for robot-assisted feeding using an lstm-based variational autoencoder,

    D. Park, Y . Hoshi, and C. C. Kemp, “A multimodal anomaly detector for robot-assisted feeding using an lstm-based variational autoencoder,” IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 1544–1551, 2018

    2018

  13. [21]

    ToyAD- MOS: A dataset of miniature-machine operating sounds for anomalous sound detection,

    Y . Koizumi, S. Saito, H. Uematsu, N. Harada, and K. Imoto, “ToyAD- MOS: A dataset of miniature-machine operating sounds for anomalous sound detection,” in2019 IEEE Workshop on Applications of Signal JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13 Processing to Au...

    2021

  14. [22]

    MIMII dataset: Sound dataset for malfunctioning industrial machine investigation and inspection,

    H. Purohit, R. Tanabe, K. Ichige, T. Endo, Y . Nikaido, K. Suefusa, and Y . Kawaguchi, “MIMII dataset: Sound dataset for malfunctioning industrial machine investigation and inspection,” inProc. Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019...

    2019

  15. [23]

    Toyadmos2: Another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions,

    N. Harada, D. Niizumi, D. Takeuchi, Y . Ohishi, M. Yasuda, and S. Saito, “Toyadmos2: Another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions,” in Proceedings of the 6th Detection and Classification of Acoustic Scenes an...

    2021

  16. [24]

    Description and discussion on DCASE 2022 challenge task 2: Unsu- pervised anomalous sound detection for machine condition monitoring applying domain generalization techniques,

    K. Dohi, K. Imoto, N. Harada, D. Niizumi, Y . Koizumi, T. Nishida, H. Purohit, R. Tanabe, T. Endo, M. Yamamoto, and Y . Kawaguchi, “Description and discussion on DCASE 2022 challenge task 2: Unsu- pervised anomalous sound detection for machine condition monitoring applying dom...

    2022

  17. [25]

    Transformer-based autoencoder with id constraint for unsupervised anomalous sound detection,

    J. Guan, Y . Liu, Q. Kong, F. Xiao, Q. Zhu, J. Tian, and W. Wang, “Transformer-based autoencoder with id constraint for unsupervised anomalous sound detection,”EURASIP journal on audio, speech, and music processing, vol. 2023, no. 1, p. 42, 2023

    2023

  18. [26]

    Anomalous sound detection with ensemble of autoencoder and binary classification approaches,

    I. Kuroyanagi, T. Hayashi, Y . Adachi, T. Yoshimura, K. Takeda, and T. Toda, “Anomalous sound detection with ensemble of autoencoder and binary classification approaches,”DCASE2021 Challenge, 2021

    2021

  19. [27]

    Representational learning for an anomalous sound detection system with source separation model,

    S. Shin and S. Lee, “Representational learning for an anomalous sound detection system with source separation model,” inProceedings of the Detection and Classification of Acoustic Scenes and Events 2024 Workshop (DCASE2024), Tokyo, Japan, October 2024, pp. 146–150

    2024

  20. [28]

    Unsupervised anomalous sound detection for industrial monitoring based on arcface classifier and gaussian mixture model,

    J. Wu, F. Yang, and W. Hu, “Unsupervised anomalous sound detection for industrial monitoring based on arcface classifier and gaussian mixture model,”Applied Acoustics, vol. 203, p. 109188, 2023

    2023

  21. [29]

    Why do angular margin losses work well for semi-supervised anomalous sound detection?

    K. Wilkinghoff and F. Kurth, “Why do angular margin losses work well for semi-supervised anomalous sound detection?”IEEE/ACM Transac- tions on Audio, Speech, and Language Processing, vol. 32, pp. 608–622, 2023

    2023

  22. [30]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” inInternational conference on machine learning. PmLR, 2020, pp. 1597–1607

    2020

  23. [31]

    Exploring simple siamese representation learning,

    X. Chen and K. He, “Exploring simple siamese representation learning,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 15 750–15 758

    2021

  24. [32]

    Anomalous sound detection using audio representation with machine id based contrastive learning pretraining,

    J. Guan, F. Xiao, Y . Liu, Q. Zhu, and W. Wang, “Anomalous sound detection using audio representation with machine id based contrastive learning pretraining,” inICASSP 2023-2023 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–5

    2023

  25. [33]

    Beats: Audio pre-training with acoustic tokenizers,

    S. Chen, Y . Wu, C. Wang, S. Liu, D. Tompkins, Z. Chen, and F. Wei, “Beats: Audio pre-training with acoustic tokenizers,”arXiv preprint arXiv:2212.09058, 2022

    2022 arXiv

  26. [34]

    Eat: Self- supervised pre-training with efficient audio transformer,

    W. Chen, Y . Liang, Z. Ma, Z. Zheng, and X. Chen, “Eat: Self- supervised pre-training with efficient audio transformer,”arXiv preprint arXiv:2401.03497, 2024

    2024 arXiv

  27. [35]

    Pre-trained model enhanced anomalous sound detection system for dcase2025 task2,

    L. Wang, “Pre-trained model enhanced anomalous sound detection system for dcase2025 task2,” DCASE2025 Challenge, Tech. Rep., June 2025

    2025

  28. [36]

    Genrep for first-shot unsupervised anomalous sound detection of dcase 2025 challenge,

    P. Saengthong and T. Shinozaki, “Genrep for first-shot unsupervised anomalous sound detection of dcase 2025 challenge,” DCASE2025 Challenge, Tech. Rep., June 2025

    2025

  29. [37]

    Visualizing data using t-sne,

    L. v. d. Maaten and G. Hinton, “Visualizing data using t-sne,”Journal of machine learning research, vol. 9, no. Nov, pp. 2579–2605, 2008

    2008

  30. [38]

    Umap: Uniform manifold approximation and projection for dimension reduction,

    L. McInnes, J. Healy, and J. Melville, “Umap: Uniform manifold approximation and projection for dimension reduction,”arXiv preprint arXiv:1802.03426, 2018

    2018 arXiv

  31. [39]

    Understanding intermediate layers using linear classifier probes,

    G. Alain and Y . Bengio, “Understanding intermediate layers using linear classifier probes,”arXiv preprint arXiv:1610.01644, 2016

    2016 arXiv

  32. [40]

    Designing and interpreting probes with control tasks,

    J. Hewitt and P. Liang, “Designing and interpreting probes with control tasks,”arXiv preprint arXiv:1909.03368, 2019

    1909 arXiv

  33. [41]

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere,

    T. Wang and P. Isola, “Understanding contrastive representation learning through alignment and uniformity on the hypersphere,” inInternational conference on machine learning. PMLR, 2020, pp. 9929–9939

    2020

  34. [42]

    How well do self- supervised models transfer?

    L. Ericsson, H. Gouk, and T. M. Hospedales, “How well do self- supervised models transfer?” inProceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition, 2021, pp. 5414–5423

    2021

  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]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  37. [45]

    Anomalous sound detection based on sound separation,

    K. Shimonishi, K. Dohi, and Y . Kawaguchi, “Anomalous sound detection based on sound separation,”INTERSPEECH 2023, 2023

    2023

  38. [46]

    Cmgan: Conformer-based metric- gan for monaural speech enhancement,

    S. Abdulatif, R. Cao, and B. Yang, “Cmgan: Conformer-based metric- gan for monaural speech enhancement,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 32, pp. 2477–2493, 2024

    2024

  39. [47]

    Conformer: Convolution-augmented transformer for speech recognition,

    A. Gulati, J. Qin, C.-C. Chiu, N. Parmar, Y . Zhang, J. Yu, W. Han, S. Wang, Z. Zhang, Y . Wuet al., “Conformer: Convolution-augmented transformer for speech recognition,” inProc. Interspeech 2020, 2020, pp. 5036–5040

    2020

  40. [48]

    Audio set: An ontology and human- labeled dataset for audio events,

    J. F. Gemmeke, D. P. Ellis, D. Freedman, A. Jansen, W. Lawrence, R. C. Moore, M. Plakal, and M. Ritter, “Audio set: An ontology and human- labeled dataset for audio events,” in2017 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2017, p...

    2017

  41. [49]

    Sdr–half-baked or well done?

    J. Le Roux, S. Wisdom, H. Erdogan, and J. R. Hershey, “Sdr–half-baked or well done?” inICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019, pp. 626–630. SEUNGHYEON SHINreceived the B.S. degree in mechanical engineering...

    2019

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.