Pith. sign in

REVIEW 4 major objections 6 minor 39 references

Unsupervised Anomaly Detection for Autonomous Robots via Mahalanobis SVDD with Audio-IMU Fusion

T0 review · 4 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 This paper claims that an autonomous mobile robot can detect collisions and internal mechanical faults in an unsupervised way by fusing audio and IMU signals, using Mahalanobis distance in a deep support vector data description to learn…

desk verdict Useful dataset and a sensible empirical study, but Eq. 8's soft-boundary objective is degenerate: the claimed ellipsoidal boundary is not learned, so the novelty needs re-framing. read the letter →

arxiv 2505.05811 v1 pith:2DAFVBQ2 submitted 2025-05-09 cs.RO

Yizhuo Yang , Jiulin Zhao , Xinhang Xu , Kun Cao , Shenghai Yuan , Lihua Xie This is my paper · ORCID
classification cs.RO
keywords unsupervisedanomalydetectionaudio-IMUfusionMahalanobisdistancesupportvectordatadescriptionmobilerobotfaultcollisionrobustcovarianceestimationreconstructionbranch
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 establish that an autonomous mobile robot can learn to recognize its own collisions and mechanical faults from audio and IMU signals alone, using only normal-operation data for training. The proposed method replaces the Euclidean distance in deep SVDD with Mahalanobis distance built on a robust covariance estimate, so the decision boundary becomes an ellipsoid aligned with the actual spread of normal features. A reconstruction branch keeps the learned features diverse and supplies a second anomaly signal. The paper reports that this combination beats shallow and deep baselines on a new robot dataset and generalizes to public machine-sound and time-series benchmarks.

What carries the argument

The central object is the Mahalanobis SVDD (M-SVDD) module: a deep one-class boundary in latent space measured by Mahalanobis distance instead of Euclidean distance. Its key components are (i) the robust mean and covariance estimate from the minimum covariance determinant over a batch subset $H$, (ii) the soft-boundary loss $L_{\text{MSVDD}} = R^2 + \frac{1}{N}\sum_i \max(0, D_i^2 - R^2)$ with learnable radius $R$, and (iii) a reconstruction branch with Huber loss that keeps features informative. Cross-attention fusion of audio and IMU features feeds this module. The mechanism works by down-weighting high-variance feature directions and capturing inter-feature correlations, so normal samples occupy a compact ellipsoid and anomalies fall outside it.

What would settle it

Train the method on normal data from one floor surface and test on normal data from another surface with no faults; if most test samples exceed the 95th-percentile training threshold, the single-ellipsoid normality assumption fails and the method would raise false alarms on benign surface changes.

Watch

Extended reading notes

Core claim

The central discovery claimed is that modeling the latent feature distribution with its covariance structure, rather than assuming isotropic features, substantially improves unsupervised anomaly detection for robots. The M-SVDD module computes an anomaly score as Mahalanobis distance $D_i = \sqrt{(z_i-\mu_z)^T \Sigma_z^{-1}(z_i-\mu_z)}$, where $\mu_z$ and $\Sigma_z$ are estimated from the most compact subset of each training batch via the minimum covariance determinant estimator. This produces an adaptive ellipsoidal boundary around normal audio-IMU features. A parallel reconstruction branch reconstructs audio and IMU inputs from their encodings, preventing representation collapse and yielding a reconstruction-loss term that is combined with the distance score at inference. The authors claim that on their collected mobile robot dataset the method reaches F1 92.3% and AUC 97.0%, and on the MIMII machine-sound dataset an average AUC 81.5%.

Load-bearing premise

The argument assumes that the statistical estimate of normal data's spread, taken from a small most-consistent subset of each training batch, captures the true shape of the normal audio-IMU data well enough to define one stable ellipsoid at both training and inference.

Editorial extensions

If this is right

  • Robots could monitor their own health with cheap, low-power sensors and no labeled fault data, making anomaly detection practical for deployment on resource-constrained platforms.
  • The method should transfer to other platforms that emit audio and vibration signals, since experiments on the MIMII dataset and three public time-series benchmarks show generalization beyond the collected robot data.
  • The anomaly threshold can be set from training scores alone, using the 95th percentile of the training anomaly scores, which removes the need for validation anomalies.
  • The ablation results indicate that audio is the primary carrier of anomaly information: removing audio drops recall from 97.5% to 64.4%, while IMU contributes a smaller but consistent improvement.
  • The reconstruction branch is load-bearing for precision: removing it drops precision from 87.7% to 81.7%, showing that the auxiliary task prevents the representation collapse that would otherwise degrade the SVDD boundary.

Reading between the lines

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

  • A testable extension would be to replace the fixed subset fraction in the covariance estimator with an adaptive or multimodal estimate, since a single ellipsoid may miss multi-modal normal operation; the paper's own limitation section notes false positives when both modalities react to uneven terrain.
  • The same M-SVDD module could be applied to other paired sensor streams, such as vibration plus motor current in industrial equipment, wherever normal data is abundant and faults are rare.
  • The reported class-specific AUCs suggest one could test whether the ellipsoid boundary is systematically weaker on mechanical faults when audio is masked, which would reveal whether IMU data is doing the heavy lifting in that regime.
  • Because the threshold is fixed at the 95th percentile of training scores, the method implicitly targets a 5% false-positive rate on the training distribution; calibrating the threshold per operating condition could improve precision on the collision-versus-mechanical split.
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

4 major / 6 minor

Summary. The paper proposes an unsupervised anomaly detection framework for mobile robots using audio and IMU data. The architecture encodes the two modalities, fuses them with a cross-attention module, and feeds the fused feature into two branches: a reconstruction branch that reconstructs the inputs, and a Mahalanobis SVDD (M-SVDD) branch that is intended to learn an ellipsoidal boundary around normal data using Mahalanobis distance and Minimum Covariance Determinant (MCD) robust covariance estimates. The final anomaly score combines the Mahalanobis distance with the reconstruction loss, and the decision threshold is set as the 95th percentile of training scores. The method is evaluated on a newly collected mobile-robot dataset and on four public datasets (MIMII, SWaT, SMAP, MSL), with reported F1 and AUC improvements over shallow and deep baselines. The authors also release code and data.

Significance. If the method were implemented as described, it would address a practical gap: detecting collisions and mechanical faults in mobile robots without labeled anomaly data, using cheap and energy-efficient sensors. The paper has concrete strengths: it ships a new multimodal dataset, releases code and a video, reports threshold-independent AUC scores, and evaluates on public benchmarks with ablations on modalities, fusion, and the M-SVDD/reconstruction modules. The reported AUC gains on the collected dataset (97.0%) and on MIMII (average 81.5%) are consistent with a meaningful empirical improvement, assuming the implementation matches the description. However, the central mathematical object of the paper, the soft-boundary M-SVDD loss in Eq. (8), is degenerate as written, and the evaluation protocol for F1 scores leaks test-label information. These issues make the paper's central claim as currently stated unreliable, despite the apparent value of the dataset and experimental setup.

major comments (4)
  1. [Sec. III-B, Eq. (8)] The soft-boundary objective is degenerate as printed. For fixed network features, L(R)=R^2+(1/N)Σ max(0,D_i^2−R^2). For any R≤min_i D_i, all hinge terms are active and L(R)=(1/N)Σ D_i^2, independent of R; for R>min_i D_i, the derivative is 2R(1−k/N)>0 where k is the number of points with D_i^2>R^2, so the loss is minimized on the whole interval [0,min_i D_i]. Thus the learnable radius never encloses a single training point, and no adaptive ellipsoidal boundary is actually learned. Algorithm 1 confirms that R is not used in inference: the threshold is the 95th percentile of the scores in Eq. (13), which depend on Mahalanobis distance and reconstruction loss only. The standard soft-boundary deep SVDD objective from which Eq. (8) is taken includes a 1/(νN) multiplier on the hinge term; without it, the trade-off between R and violations disappears. This is the central contribution claimed in Sec. I. Please either add the missing multiplier (or otherwise constrain R) and rerun the experiments, or revise the contribution statement to describe the method as Mahalanobis-distance-plus-reconstruction scoring without claiming a learned enclosing boundary.
  2. [Sec. IV-A, evaluation metrics] The F1 scores in Tables I, II, III, and V are computed by selecting the optimal threshold with the best F1 on the test labels, as explicitly stated in Sec. IV-A. This leaks test information into the decision rule and inflates precision and recall relative to any usable unsupervised deployment. The same threshold-selection protocol is not equally meaningful for all baselines, making the F1-based comparisons unreliable. Please report thresholds chosen from training data only (e.g., the 95th percentile rule already used in Algorithm 1), or make AUC the primary metric. In addition, no standard deviations or repeated-run results are reported anywhere; with 15 training sequences and a single split, the small F1 differences in Table I (0.3–0.7%) cannot be judged significant.
  3. [Sec. III-D, Eq. (13)] As written, the anomaly score is δ_i = D_i + w (μ_T_D / μ_T_LRec) L_i^Rec. The text states that both metrics are normalized to the same scale, but D_i is not divided by μ_T_D; only the reconstruction term is scaled. If the intended formula is D_i/μ_T_D + w L_i^Rec/μ_T_LRec, please correct it; if not, the normalization claim is inaccurate. Because this formula defines the deployed anomaly score, it directly affects the threshold and all reported P/R/F1 numbers.
  4. [Sec. III-B, Eqs. (6)–(7) and Sec. IV] The MCD subset size h is never specified in the experimental section, even though h controls the robustness of the covariance estimate used in the Mahalanobis distance. Since Σ_z and μ_z are estimated from a subset H of size h<N, the method is not reproducible without reporting h (or the fraction h/N) and ideally a sensitivity study over h. This is particularly relevant because the paper's robustness claim rests on the MCD estimator.
minor comments (6)
  1. [Sec. III-B, Eq. (6)] Equation (6) defines μ_z as (1/h)Σ_{i∈H} x_i; this should presumably be z_i, the fused latent features, to be consistent with Eq. (7).
  2. [Sec. IV-C, Table II] The column header 'Percision' is a typo; it should be 'Precision'.
  3. [Sec. III-A, Eq. (3)] The cross-attention formula uses Q, K, V without defining the projection matrices that produce them from z_A and z_I. Please clarify whether Q, K, V are the raw features or learned projections.
  4. [Algorithm 1] Step 2 initializes the radius R but no initialization value or update rule is given for R. After the loss in Eq. (8) is corrected, please specify how R is initialized and updated during training.
  5. [Sec. IV-E] The time-series experiments use the point-adjust evaluation protocol, which is known to be optimistic for anomaly detection. The authors follow previous work in this regard, but the text should note this explicitly so the F1 numbers are not over-interpreted.
  6. [General terminology] The terms 'hypersphere', 'hyperspace', and 'ellipsoidal space' are used inconsistently (e.g., Sec. III-B vs. Algorithm 1). Since the proposed boundary is ellipsoidal in feature space, please use consistent geometric terminology.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: M-SVDD is evaluated against external baselines and public datasets, and no central claim reduces by construction to its inputs.

full rationale

The paper's central claim is empirical: M-SVDD with audio-IMU fusion detects collisions and mechanical faults from normal-only training. The derivation chain is explicit: fused features z are obtained from encoders and cross-attention (Eqs. 3-4); robust mean and covariance are estimated from a training batch via MCD (Eqs. 6-7); Mahalanobis distances D_i are computed (Eq. 5); training uses a soft-boundary loss (Eq. 8) plus reconstruction and regularization terms (Eqs. 9-12); inference scores samples by D_i plus reconstruction loss with a 95th-percentile training threshold (Eq. 13). None of these steps defines an output in terms of the target anomaly labels, nor fits a parameter to the test-set labels and then renames it a prediction. The comparisons against Isolation Forest, OCSVM, SVDD, DSVDD, DAGMM, and others on the collected dataset and on MIMII, SWaT, SMAP, and MSL are external benchmarks, giving the empirical claim independent content. There are no load-bearing self-citations: prior works such as DSVDD [22], MCD [25], VAE-SVDD [32], and COUTA [17] are by other groups and serve as standard external foundations. The manually tuned combination weights and the reported optimal-threshold evaluation protocol are evaluation choices, not circular derivation. The skeptic's observation that the learnable radius R in Eq. 8 can degenerate to R being no larger than the minimum training distance is a legitimate internal-consistency and correctness concern about whether the claimed ellipsoidal boundary is actually learned, but it is not a case of the paper's prediction reducing to its input by definition, so it does not raise the circularity score. No self-definitional, fitted-input-as-prediction, or self-citation-chain circularity was found.

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

The central method rests on the assumption that normal fused features are unimodal and well described by an ellipsoid, that MCD subset selection yields a stable covariance, and that audio and IMU inputs are synchronized. Free parameters alpha, w, epsilon, h, and latent dimensions are empirically tuned, with h not reported. No invented physical entities are introduced.

free parameters (6)
  • alpha_1, alpha_2, alpha_3 = 1, 1, 0.001
    Loss weights in Eq. 12 for M-SVDD, reconstruction, and regularization terms. Text says they are 'manually set based on empirical tuning' (Sec. III-C, IV-A).
  • w = 0.01
    Weight balancing Mahalanobis distance and reconstruction loss in the anomaly score, Eq. 13, set in Sec. IV-A.
  • MCD subset size h = not reported
    Eq. 6-7 require a subset H of size h<N with minimum determinant, but h is never specified in the implementation details.
  • epsilon = 0.001
    Covariance regularization constant in Eq. 7, set by hand to avoid singular covariance.
  • latent dimensions d and s = 32
    Feature and ellipsoid dimensions set to 32 in Sec. IV-A; Fig. 11 shows performance tuning over s.
  • evaluation threshold for F1 = optimal threshold on test labels
    Sec. IV-A says thresholds are chosen to maximize F1 on test samples, which is a fitted value affecting precision and recall.
assumptions (5)
  • domain assumption Normal operation data is representative and anomalies are out-of-distribution in the fused feature space.
    Core premise of one-class anomaly detection, stated in Sec. I and III-B.
  • ad hoc to paper The normal data distribution in latent space is unimodal enough to be enclosed by a single ellipsoid described by mean and covariance.
    M-SVDD models only one mean and covariance (Eq. 5-8); multi-modal normal conditions such as different road surfaces or background soundscapes would violate this.
  • ad hoc to paper The MCD subset H of size h captures the true normal distribution despite unlabeled outliers in training data.
    Eq. 6-7 rely on a minimum-determinant subset; h is unspecified and robustness depends on this choice.
  • domain assumption Audio and IMU signals are temporally aligned and synchronized during deployment.
    The limitations section admits timing misalignment in ROS2 systems, yet the model assumes aligned 2-second input windows.
  • domain assumption Z-axis IMU velocity alone captures representative anomaly cues.
    Sec. IV-A restricts IMU input to Z-axis velocity 'as it captures more representative anomaly features', an empirical assumption not derived.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Anomaly Detection for Autonomous Robots via Mahalanobis SVDD with Audio-IMU Fusion." pith.science (2026). https://pith.science/paper/2DAFVBQ2

@misc{pith2026250505811,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Anomaly Detection for Autonomous Robots via Mahalanobis SVDD with Audio-IMU Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2DAFVBQ2}},
  note         = {Machine review of arXiv:2505.05811}
}
read the original abstract

Reliable anomaly detection is essential for ensuring the safety of autonomous robots, particularly when conventional detection systems based on vision or LiDAR become unreliable in adverse or unpredictable conditions. In such scenarios, alternative sensing modalities are needed to provide timely and robust feedback. To this end, we explore the use of audio and inertial measurement unit (IMU) sensors to detect underlying anomalies in autonomous mobile robots, such as collisions and internal mechanical faults. Furthermore, to address the challenge of limited labeled anomaly data, we propose an unsupervised anomaly detection framework based on Mahalanobis Support Vector Data Description (M-SVDD). In contrast to conventional SVDD methods that rely on Euclidean distance and assume isotropic feature distributions, our approach employs the Mahalanobis distance to adaptively scale feature dimensions and capture inter-feature correlations, enabling more expressive decision boundaries. In addition, a reconstruction-based auxiliary branch is introduced to preserve feature diversity and prevent representation collapse, further enhancing the robustness of anomaly detection. Extensive experiments on a collected mobile robot dataset and four public datasets demonstrate the effectiveness of the proposed method, as shown in the video https://youtu.be/yh1tn6DDD4A. Code and dataset are available at https://github.com/jamesyang7/M-SVDD.

Figures

Figures reproduced from arXiv: 2505.05811 by the authors.

Figure 1
Figure 1. Our proposed method integrates audio and IMU data for anomaly [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall workflow of the proposed unsupervised anomaly detection framework. In the training stage, audio and IMU data collected under normal [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The cross-attention module for audio-IMU feature fusion. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The experimental platform utilized for data collection. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Examples of selected samples from normal driving and anomalous [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Figures (a) and (b) illustrate the variance distribution in the PCA space for DSVDD and M-SVDD, respectively, while Figure (c) depicts the covariance [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: The ROC curves for different methods. The number in the bracket [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Comparison of ROC curves across different machine types on the MIMII dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Visualization of the t-SNE embedding and feature distribution for [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 22 canonical work pages

  1. [1]

    Deep industrial image anomaly detection: A survey,

    J. Liu, G. Xie, J. Wang, S. Li, C. Wang, F. Zheng, and Y . Jin, “Deep industrial image anomaly detection: A survey,” Machine Intelligence Research, vol. 21, no. 1, pp. 104–135, 2024

    2024

  2. [2]

    Machine learning for anomaly assessment in sensor networks for ndt in aerospace,

    I. Kraljevski, F. Duckhorn, C. Tsch ¨ope, and M. Wolff, “Machine learning for anomaly assessment in sensor networks for ndt in aerospace,” IEEE Sensors Journal, vol. 21, no. 9, pp. 11 000–11 008, 2021

    work page 2021

  3. [3]

    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

    work page 2018

  4. [4]

    Multimodal anomaly detection based on deep auto-encoder for object slip perception of mobile ma- nipulation robots,

    Y . Yoo, C.-Y . Lee, and B.-T. Zhang, “Multimodal anomaly detection based on deep auto-encoder for object slip perception of mobile ma- nipulation robots,” in 2021 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2021, pp. 11 443–11 449. 11

    work page 2021

  5. [5]

    Safe robot navigation via multi-modal anomaly detection,

    L. Wellhausen, R. Ranftl, and M. Hutter, “Safe robot navigation via multi-modal anomaly detection,” IEEE Robotics and Automation Letters , vol. 5, no. 2, pp. 1326–1333, 2020

    work page 2020

  6. [6]

    Proactive anomaly detection for robot navigation with multi-sensor fusion,

    T. Ji, A. N. Sivakumar, G. Chowdhary, and K. Driggs-Campbell, “Proactive anomaly detection for robot navigation with multi-sensor fusion,” IEEE Robotics and Automation Letters , vol. 7, no. 2, pp. 4975– 4982, 2022

    work page 2022

  7. [7]

    Safe autonomous driving in adverse weather: Sensor evaluation and performance monitoring,

    F. Sezgin, D. Vriesman, D. Steinhauser, R. Lugner, and T. Brandmeier, “Safe autonomous driving in adverse weather: Sensor evaluation and performance monitoring,” in 2023 IEEE Intelligent V ehicles Symposium (IV). IEEE, 2023, pp. 1–6

    work page 2023

  8. [8]

    A novel vehicle collision detection system: Integrating audio-visual fusion for enhanced performance,

    K. Li, Z. Zhao, Q. Cai, Q. Wang, N. Jing, Z. Mao, and J. Jiang, “A novel vehicle collision detection system: Integrating audio-visual fusion for enhanced performance,” Expert Systems with Applications , vol. 249, p. 123828, 2024

    work page 2024

Show all 39 references
  1. [9]

    Fault diagnosis for mobile robots based on spatial-temporal graph attention network under imbalanced data,

    L. Zhang, Z. Miao, Y . Xia, F. Zhou, and X. Yuan, “Fault diagnosis for mobile robots based on spatial-temporal graph attention network under imbalanced data,” IEEE Transactions on Instrumentation and Measurement, 2023

    2023

  2. [10]

    A semi-supervised enhanced fault diagnosis algorithm for complex equipment assisted by digital multitwins,

    S. Liu, D. Xu, C. Shen, Y . Ye, and B. Jiang, “A semi-supervised enhanced fault diagnosis algorithm for complex equipment assisted by digital multitwins,” IEEE Transactions on Instrumentation and Measure- ment, vol. 74, pp. 1–11, 2025

    2025

  3. [11]

    A few-shot machinery fault diagnosis framework based on self-supervised signal representation learning,

    H. Wang, X. Wang, Y . Yang, K. Gryllias, and Z. Liu, “A few-shot machinery fault diagnosis framework based on self-supervised signal representation learning,” IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1–14, 2024

    2024

  4. [12]

    Modeling the distribution of normal data in pre-trained deep features for anomaly detection,

    O. Rippel, P. Mertens, and D. Merhof, “Modeling the distribution of normal data in pre-trained deep features for anomaly detection,” in 2020 25th International Conference on Pattern Recognition (ICPR) . IEEE, 2021, pp. 6726–6733

    2020

  5. [13]

    The voraus-ad dataset for anomaly detection in robot applications,

    J. T. Brockmann, M. Rudolph, B. Rosenhahn, and B. Wandt, “The voraus-ad dataset for anomaly detection in robot applications,” IEEE Transactions on Robotics , vol. 40, pp. 438–451, 2023

    2023

  6. [14]

    Reconstruction-based deep unsupervised adaptive threshold support vector data description for wind turbine anomaly detection,

    D. Peng, W. Desmet, and K. Gryllias, “Reconstruction-based deep unsupervised adaptive threshold support vector data description for wind turbine anomaly detection,” Reliability Engineering & System Safety , p. 110995, 2025

    2025

  7. [15]

    A novel methodology for unsupervised anomaly detection in industrial electrical systems,

    M. Carrat `u, V . Gallo, S. D. Iacono, P. Sommella, A. Bartolini, F. Grasso, L. Ciani, and G. Patrizi, “A novel methodology for unsupervised anomaly detection in industrial electrical systems,” IEEE Transactions on Instrumentation and Measurement , vol. 72, pp. 1–12, 2023

    2023

  8. [16]

    Data-driven anomaly detection using ocsvm with boundary optimzation,

    K. Guo, D. Liu, Y . Peng, and X. Peng, “Data-driven anomaly detection using ocsvm with boundary optimzation,” in 2018 Prognostics and System Health Management Conference (PHM-Chongqing) , 2018, pp. 244–248

    2018

  9. [17]

    Calibrated one-class classification for unsupervised time series anomaly detection,

    H. Xu, Y . Wang, S. Jian, Q. Liao, Y . Wang, and G. Pang, “Calibrated one-class classification for unsupervised time series anomaly detection,” IEEE Transactions on Knowledge and Data Engineering , 2024

    2024

  10. [18]

    Pyramid reconstruction assisted deep autoencoding gaussian mixture model for industrial fault detection,

    Y . Tian, J. Li, Q. Song, Z. Li, and X. Huang, “Pyramid reconstruction assisted deep autoencoding gaussian mixture model for industrial fault detection,” Information Sciences , vol. 649, p. 119682, 2023

    2023

  11. [19]

    Unsuper- vised anomaly detection for self-flying delivery drones,

    V . Sindhwani, H. Sidahmed, K. Choromanski, and B. Jones, “Unsuper- vised anomaly detection for self-flying delivery drones,” in 2020 IEEE international conference on robotics and automation (ICRA) . IEEE, 2020, pp. 186–192

    2020

  12. [20]

    Robust spatial-temporal autoencoder for unsupervised anomaly detection of unmanned aerial vehicle with flight data,

    G. Jiang, P. Nan, J. Zhang, Y . Li, and X. Li, “Robust spatial-temporal autoencoder for unsupervised anomaly detection of unmanned aerial vehicle with flight data,” IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1–14, 2024

    2024

  13. [21]

    Support vector data description,

    D. M. Tax and R. P. Duin, “Support vector data description,” Machine learning, vol. 54, pp. 45–66, 2004

    2004

  14. [22]

    Deep one-class classification,

    L. Ruff, R. Vandermeulen, N. Goernitz, L. Deecke, S. A. Siddiqui, A. Binder, E. M ¨uller, and M. Kloft, “Deep one-class classification,” in International conference on machine learning . PMLR, 2018, pp. 4393–4402

    2018

  15. [23]

    Patch svdd: Patch-level svdd for anomaly detection and segmentation,

    J. Yi and S. Yoon, “Patch svdd: Patch-level svdd for anomaly detection and segmentation,” in Proceedings of the Asian conference on computer vision, 2020

    2020

  16. [24]

    Is someone speaking? exploring long-term temporal features for audio- visual active speaker detection,

    R. Tao, Z. Pan, R. K. Das, X. Qian, M. Z. Shou, and H. Li, “Is someone speaking? exploring long-term temporal features for audio- visual active speaker detection,” in Proceedings of the 29th ACM international conference on multimedia , 2021, pp. 3927–3935

    2021

  17. [25]

    A fast algorithm for the minimum covariance determinant estimator,

    P. J. Rousseeuw and K. van Driessen, “A fast algorithm for the minimum covariance determinant estimator,” Technometrics, vol. 41, no. 3, pp. 212–223, 1999

    1999

  18. [26]

    Usad: Unsupervised anomaly detection on multivariate time series,

    J. Audibert, P. Michiardi, F. Guyard, S. Marti, and M. A. Zuluaga, “Usad: Unsupervised anomaly detection on multivariate time series,” in Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining , 2020, pp. 3395–3404

    2020

  19. [27]

    Multivariate time series anomaly detection and interpretation using hierarchical inter- metric and temporal embedding,

    Z. Li, Y . Zhao, J. Han, Y . Su, R. Jiao, X. Wen, and D. Pei, “Multivariate time series anomaly detection and interpretation using hierarchical inter- metric and temporal embedding,” in Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining , 2021, ...

    2021

  20. [28]

    Isolation forest,

    F. T. Liu, K. M. Ting, and Z.-H. Zhou, “Isolation forest,” in 2008 eighth ieee international conference on data mining . IEEE, 2008, pp. 413–422

    2008

  21. [29]

    One-class svms for document clas- sification,

    L. M. Manevitz and M. Yousef, “One-class svms for document clas- sification,” Journal of machine Learning research , vol. 2, no. Dec, pp. 139–154, 2001

    2001

  22. [30]

    Deep isolation forest for anomaly detection,

    H. Xu, G. Pang, Y . Wang, and Y . Wang, “Deep isolation forest for anomaly detection,” IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 12, pp. 12 591–12 604, 2023

    2023

  23. [31]

    Deep autoencoding gaussian mixture model for unsupervised anomaly detection,

    B. Zong, Q. Song, M. R. Min, W. Cheng, C. Lumezanu, D. Cho, and H. Chen, “Deep autoencoding gaussian mixture model for unsupervised anomaly detection,” in International conference on learning representa- tions, 2018

    2018

  24. [32]

    Vae-based deep svdd for anomaly detection,

    Y . Zhou, X. Liang, W. Zhang, L. Zhang, and X. Song, “Vae-based deep svdd for anomaly detection,” Neurocomputing, vol. 453, pp. 131–140, 2021

    2021

  25. [33]

    Mimii dataset: Sound dataset for malfunction- ing 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 malfunction- ing industrial machine investigation and inspection,” arXiv preprint arXiv:1909.09347, 2019

    1909 arXiv

  26. [34]

    Swat: A water treatment testbed for research and training on ics security,

    A. P. Mathur and N. O. Tippenhauer, “Swat: A water treatment testbed for research and training on ics security,” in2016 international workshop on cyber-physical systems for smart water networks (CySWater) . IEEE, 2016, pp. 31–36

    2016

  27. [35]

    Detecting spacecraft anomalies using lstms and nonparametric dynamic thresholding,

    K. Hundman, V . Constantinou, C. Laporte, I. Colwell, and T. Soder- strom, “Detecting spacecraft anomalies using lstms and nonparametric dynamic thresholding,” in Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining , 2018, pp. 387–395

    2018

  28. [36]

    An empirical evaluation of generic convolutional and recurrent networks for sequence modeling,

    S. Bai, J. Z. Kolter, and V . Koltun, “An empirical evaluation of generic convolutional and recurrent networks for sequence modeling,” arXiv preprint arXiv:1803.01271, 2018

    2018 arXiv

  29. [37]

    TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data,

    S. Tuli, G. Casale, and N. R. Jennings, “TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data,” Proceedings of VLDB , vol. 15, no. 6, pp. 1201–1214, 2022

    2022

  30. [38]

    Angle-based outlier de- tection in high-dimensional data,

    H.-P. Kriegel, M. Schubert, and A. Zimek, “Angle-based outlier de- tection in high-dimensional data,” in Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, 2008, pp. 444–452

    2008

  31. [39]

    Robospike: Fully utilizing the heterogeneous system with subcallback scheduling in ros 2,

    H. Li, Q. Yang, S. Ma, R. Zhao, and X. Ji, “Robospike: Fully utilizing the heterogeneous system with subcallback scheduling in ros 2,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2025

    2025

Pith tools

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