Pith. sign in

REVIEW 3 major objections 4 minor 55 references

Baking process knowledge into graph masks, not better temporal modeling, is what drives anomaly-detection gains in multi-stage plants.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 22:15 UTC pith:Z2GEPFL2

load-bearing objection A promising knowledge-masking idea undermined by a GAT aggregation formula that makes the graphs irrelevant — worth a round of revision, not publication as is. the 3 major comments →

arxiv 2607.15799 v1 pith:Z2GEPFL2 submitted 2026-07-17 cs.LG cs.AI

Knowledge-Assisted Multi-Graph Dependency Learning for Multivariate Time Series Anomaly Detection in Multi-Stage Industrial Processes

Jaeyeong Lee , Taeseong Yoon , Wonmo Koo , Heeyoung Kim This is my paper
classification cs.LG cs.AI
keywords multivariate time series anomaly detectiongraph neural networksknowledge-assisted graph learningprocess knowledgemulti-stage industrial processessensor group knowledgeprocess flow knowledgeforecasting-based anomaly detection
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper aims to establish that in multi-stage industrial processes, purely data-driven graph learning misses or invents sensor dependencies, and that two cheaply available forms of process knowledge—which sensors belong to the same sub-process and which sub-processes are adjacent—can be baked directly into graph construction as masks. It builds three graphs: one uninformed, one masked by sensor-group knowledge, and one masked by process-flow knowledge, then fuses them with separate graph attention networks plus a temporal convolutional network for forecasting-based anomaly detection. On two real water-system datasets, the fused model achieves higher F1 than data-driven baselines, and ablations show each knowledge mask contributes independently. If correct, the paper shows a simple, structural way to transfer process diagrams into deep anomaly detectors without manual relationship modeling.

Core claim

On its own terms, the paper's central discovery is that knowledge-assisted graph construction—not the temporal encoder or the anomaly-scoring scheme—is what drives the performance gain. The authors claim that sensor group knowledge and process flow knowledge, encoded as hard masks over learned sensor similarities, force the model to consider intra-sub-process and inter-sub-process dependencies that data-driven graph learning under-weights. This yields F1 of 0.8233 on a six-stage water treatment dataset and 0.6485 on a three-stage water distribution dataset, the highest among the compared methods. Ablation results show that removing either knowledge-informed graph lowers F1, and removing both

What carries the argument

The central mechanism is a masked graph-construction rule built on two learnable embeddings per sensor, one for the source role and one for the target role, whose normalized dot product defines a similarity score. Edges are selected by a Top-k rule, and process knowledge is injected as hard masks: the sensor-group-informed graph only allows edges between sensors in the same sub-process, while the process-flow-informed graph only allows edges between sensors in adjacent sub-processes. These three adjacency matrices feed three separate graph attention networks, whose outputs are concatenated with the embeddings and passed through a multi-layer perceptron to forecast the next reading; anomalies

Load-bearing premise

The load-bearing premise is that the sensor groupings and sub-process adjacencies read from the process diagrams are correct, complete, and current; if a grouping is wrong or the plant has been reconfigured, the masks can exclude genuine dependencies and force spurious ones, and the claimed advantage would shrink or vanish—a point the paper's own note to practitioners concedes by saying the knowledge should be updated when the plant configuration changes.

What would settle it

A reader could settle the claim by taking a multi-stage industrial dataset, deliberately corrupting or shuffling the sensor-group and process-flow assignments, and checking whether F1 degrades as expected; if performance does not drop when the knowledge is made false, the gains are not attributable to the semantic content of the knowledge. Comparing against an identically sized random graph mask on the same datasets would also test whether the benefit comes from structural constraints rather than from the specific process semantics.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the central claim is correct, adding these two masks is enough to raise F1 to 0.8233 on the water treatment dataset and 0.6485 on the water distribution dataset, the highest among the compared methods, with the largest gains in recall.
  • Because the ablation shows removing either knowledge graph lowers F1 and removing both lowers it further, both knowledge types carry independent signal and should be retained.
  • The approach transfers to any multi-stage process—chemical processing, power generation, automated manufacturing—where sub-process boundaries and adjacencies are documented.
  • Practitioners only need to specify sensor groupings and stage adjacencies from existing diagrams; no manual relationship modeling or per-sensor domain expertise is required.
  • The paper's error analysis shows low forecasting error on normal timestamps and markedly higher error on anomalous ones, confirming that prediction error is a usable discriminative signal when the graphs are knowledge-guided.

Where Pith is reading between the lines

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

  • A direct way to stress-test the claim would be to shuffle or corrupt the sensor groupings and stage adjacencies supplied to the masks; if F1 does not drop, the benefit is structural—sparsity or regularization—rather than semantic.
  • Because the masks are hard constraints, plant reconfiguration is a failure mode the paper acknowledges only in its practitioner note; a soft-mask or a forgetting mechanism could make the approach adaptive to outdated process diagrams.
  • The same mask-encoding could be attached to reconstruction-based or contrastive anomaly detectors, not just forecasting, which would show whether the knowledge injection transfers across detection paradigms.
  • The paper itself flags that it does not model forecast stochasticity, which can lead to overconfident predictions; coupling graph masking with probabilistic forecasting is a natural next step and could further improve reliability.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes a knowledge-assisted multi-graph framework for multivariate time series anomaly detection in multi-stage industrial processes. The method constructs three graphs—an uninformed data-driven graph, a sensor-group-knowledge-masked graph, and a process-flow-knowledge-masked graph—then uses a TCN temporal encoder and three separate GAT branches to produce forecasts. Anomaly scores are derived from per-sensor normalized prediction errors, with a threshold selected to maximize F1. Experiments on SWaT and WADI report the highest average F1 among compared baselines (0.8233 on SWaT, 0.6485 on WADI), with ablations showing performance drops when either knowledge-informed graph is removed. The central claim is that injecting process knowledge into graph construction improves dependency modeling and anomaly detection.

Significance. If the claims hold, the contribution is a practical and modest one: it shows that simple, diagram-derived knowledge (sensor groupings and process-stage adjacencies) can be baked into graph-structure learning to improve forecasting-based anomaly detection on industrial testbeds. The paper ships an anonymous code link and reports means and standard deviations over 10 runs, which are helpful for reproducibility. However, the printed formulation of the GAT branch in Section IV-C2 is internally inconsistent: as written, the neighbor aggregation collapses to a per-sensor linear transform, so the graph structure and neighbor features have no effect on the aggregated output. This is load-bearing for the central claim. The evaluation also suffers from test-label-based threshold tuning and a lack of significance testing, which weakens the quantitative conclusions even if the mechanism is repaired.

major comments (3)
  1. [§IV-C2, equations for a^U_{t,i}] The displayed update for the uninformed graph is a^U_{t,i} = ReLU(β^U_ii W^U z_{t,i} + Σ_{j∈N^U(i)} β^U_ij W^U z_{t,i}). Because the β coefficients are normalized over N^U(i)∪{i}, their sum is 1, so the entire parenthesized sum equals W^U z_{t,i} irrespective of the neighbor set, the attention scores, and the neighbors' features. Thus a^U_{t,i} = ReLU(W^U z_{t,i}), a per-sensor MLP with no graph dependence. The same construction is used for G^G and G^F, so the three graph branches are identical in structure and none of them performs inter-sensor aggregation. This contradicts the qualitative analysis: e.g., in Case 2, the claim that FIT401's stable pattern is used to predict AIT504 is impossible under the printed formula. Please correct the neighbor term to use z_{t,j} (as in the background GAT in §III-B), and either re-verify the experiments or document that the implementation differs fr
  2. [§V-C, Evaluation Metrics] The threshold for anomaly detection is described as 'determined using a grid search method to maximize F1'. As written, this appears to use the test-set labels for threshold selection. Tuning a threshold on test labels is a form of evaluation leakage; it inflates the absolute F1 values and makes comparisons with baselines unfair if those baselines do not use the same test-label tuning. Please specify whether the threshold is selected on a validation set (as the hyperparameters are) or on the test set. If the threshold is selected on the test set, the headline F1 numbers are optimistically biased and should be re-computed with an unsupervised thresholding procedure or a validation-set criterion, with sensitivity reported.
  3. [§V-E, Table II and §V-E1] The claim that the proposed method 'consistently outperforms all baselines' is not statistically supported. On SWaT, the proposed F1 is 0.8233±0.0227 versus 0.8089±0.0434 for ECNU-GNN; these are within one standard deviation, so the difference may be noise. No significance tests (e.g., paired bootstrap or permutation tests over the 10 runs) are reported. Additionally, many baseline numbers are imported from [15] (per the table footnote), and it is unclear whether those runs use the same train/test split, threshold protocol, and preprocessing as the authors' runs. Please provide paired significance tests and ensure all compared methods are evaluated under a common protocol, or clearly state the conditions under which the imported numbers were produced.
minor comments (4)
  1. [Note to Practitioners] The limitation that the knowledge-informed masks depend on the correctness and currency of the process diagram is acknowledged only in the Note to Practitioners ('should be updated if the plant configuration changes'). Since this dependency is central to the method's advantage, it would be useful to also state it in the main body and, ideally, test robustness to corrupted or outdated knowledge.
  2. [§V-E2, Table III] In the ablation study, removing a graph branch also reduces the capacity of the concatenated forecast module. To strengthen the claim that the knowledge-informed graphs specifically improve dependency modeling, consider comparing against a variant with an additional equal-capacity branch that receives no knowledge constraint, rather than simply removing the branch.
  3. [Abstract / §V-E1] The abstract and results section use the phrase 'substantially enhances' without reporting effect sizes or confidence intervals. Given the overlap in standard deviations noted above, a more measured wording would be more appropriate.
  4. [Throughout] There are minor typographical and formatting issues, including inconsistent spacing in 'W ADI' and 'SW AT' in Section V, and the use of 'Top k' without a subscript in Eq. (2). These do not affect the substance of the work but should be cleaned up.

Circularity Check

0 steps flagged

No significant circularity: the main claim is an empirical comparison supported by external benchmarks and ablations; self-citations are background, not load-bearing.

full rationale

The paper's central assertion—that adding sensor-group and process-flow knowledge as graph masks improves anomaly detection—is validated by comparisons on SWaT/WADI and by ablations (Tables II and III). The knowledge masks (Definitions 1-2, Eqs. 3-4) are taken from iTrust process diagrams, not from anomaly labels or from the forecast targets, so the reported gain is not a restatement of the input. The forecasting loss and anomaly scoring are standard error-based quantities (Sections IV-C, IV-D). Self-citations [3], [5], [8], [23], [24], [26] are background or domain-support citations; none is a uniqueness theorem or a fitted parameter that forces the result. The test-label-based threshold search is an evaluation-protocol issue, not circular reasoning. One non-circular correctness concern: in Section IV-C2 the displayed GAT update uses the target feature z_{t,i} in every summand with normalized weights, so algebraically a^U_{t,i}=ReLU(W^U z_{t,i}) and the graph structure cancels; if this equation is the implemented one, the knowledge-graph mechanism is not actually exercised. That is an internal-consistency/implementation issue to correct or verify, but it does not make the paper's derivation circular.

Axiom & Free-Parameter Ledger

6 free parameters · 4 axioms · 0 invented entities

No new physical entities. The free parameters are model hyperparameters plus the test-set-fitted anomaly threshold. The key domain assumption is that diagram-derived process knowledge is correct and current; this assumption carries the claimed improvement. The masking and shared-α design choices are ad hoc but standard for empirical ML papers.

free parameters (6)
  • anomaly threshold θ = grid-searched to maximize F1 on test set
    Section V-C: 'The threshold for anomaly detection was determined using a grid search method to maximize F1.' This uses test labels and directly inflates the headline F1.
  • graph sparsity k = 10 (both datasets)
    Section V-D: joint grid search over {5,10,15,20,25} on validation F1; controls number of incoming edges per node in A^U.
  • knowledge-graph sparsity α = 0.6 (SWaT), 0.3 (WADI)
    Section V-D: grid-searched over {0.1,0.3,0.6,0.9}; sets adaptive neighborhood sizes k_i^G and k_i^F.
  • embedding dimension d = 128
    Section V-D: grid-searched over {32,64,128,256}; dimension of source/target sensor embeddings and GAT features.
  • sliding window size w = 15
    Section V-D: set for both datasets; defines input history length for forecasting.
  • learning rate / batch size = 5e-4/32 (SWaT), 1e-3/64 (WADI)
    Section V-D: chosen by validation; standard hyperparameters.
axioms (4)
  • standard math GAT and TCN formulations (Sections III-A, III-B) compute features as described; attention normalization and dilation behave as standard.
    Background used without proof; standard in literature.
  • domain assumption Sensor group knowledge and process flow knowledge derived from iTrust process diagrams are accurate and consistent with true physical dependencies.
    Load-bearing: masks in Eqs. (3)-(4) restrict edges to same-sub-process or adjacent-sub-process sets. Wrong groupings break the method.
  • domain assumption Forecasting errors of anomalous timestamps are larger than normal ones, making prediction-error-based scoring valid.
    Basis of anomaly scoring in Section IV-D; standard for forecasting-based MTAD but not proven for these datasets.
  • ad hoc to paper The masking strategy (set C_ji=-∞ outside knowledge sets, then top-k) is a suitable way to inject knowledge; shared α for G^G and G^F is appropriate.
    No principled derivation; design choice validated only empirically.

pith-pipeline@v1.3.0-alltime-deepseek · 18199 in / 11986 out tokens · 100637 ms · 2026-08-01T22:15:46.856610+00:00 · methodology

0 comments
read the original abstract

Industrial processes often generate complex, interdependent time-series data from multiple sensors across multiple stages, forming complex dependencies among variables and process stages. Effective monitoring and timely anomaly detection of these time series through multivariate time series anomaly detection (MTAD) is crucial for preventing failures and ensuring the reliability of automated systems. Graph neural networks (GNNs) have advanced MTAD by leveraging data-driven graphs to model complex dependencies among variables, effectively capturing relational structures within multivariate time series to enhance anomaly detection performance. However, existing GNN-based approaches often overlook critical process knowledge, and even when this knowledge is considered, seamlessly incorporating it into existing models remains inherently challenging, leading to suboptimal performance. To address this limitation, we propose a knowledge-assisted multi-graph framework for modeling sensor dependencies in multi-stage industrial processes for MTAD, which explicitly incorporates process knowledge into graph learning to enhance dependency modeling and improve anomaly detection performance. Our method constructs three complementary graphs: one purely data-driven and two refined by integrating structural constraints derived from process knowledge. To effectively leverage these graphs for anomaly detection, we employ a multi-graph attention network, enabling a more accurate and robust representation of complex dependencies. Comprehensive experiments on two real-world, multi-stage industrial datasets demonstrate that incorporating process knowledge substantially enhances anomaly detection performance.

Figures

Figures reproduced from arXiv: 2607.15799 by Heeyoung Kim, Jaeyeong Lee, Taeseong Yoon, Wonmo Koo.

Figure 1
Figure 1. Figure 1: (a) The process flow diagram of the Secure Water Treatment (SWaT) process and (b) a one-hour-long multivariate time series from four sensors in [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Relationships among four sensors captured through graph structure learning in (a) the GDN [7] and (b) the proposed model. Edges are weighted by [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of a TCN with three dilated causal convolutional layers. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of the proposed method 2) Sensor Group Knowledge-Informed Graph: To model intra-sub-process relationships, we use sensor group knowl￾edge KG = {Ps} Np s=1. The sensor group knowledge-informed graph G G focuses on dependencies among sensors within the same sub-process. We define the masked similarity between the j-th sensor (source) and the i-th sensor (target) as follows: C G ji = ( Cji, if j ∈ Pϕ… view at source ↗
Figure 1
Figure 1. Figure 1: It includes measurements from 51 sensors over 11 [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 5
Figure 5. Figure 5: (a) Predicted and observed values during the attack on MV303 for [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: (a) Predicted and observed values during the attack on AIT504 for [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Results of the hyperparameter sensitivity analysis: (a) [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

55 extracted references · 8 linked inside Pith

  1. [1]

    Smart actuation for end-edge industrial control systems,

    Y . Ma et al., “Smart actuation for end-edge industrial control systems,” IEEE Trans. Autom. Sci. Eng., vol. 21, no. 1, pp. 269–283, Jan. 2022

    2022

  2. [2]

    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,” inProc. Int. Workshop Cyber- Physical Syst. Smart Water Netw. (CySWater), Vienna, Austria, 2016, pp. 31–36

    2016

  3. [3]

    & Kim, H

    Park, S., Kim, K. & Kim, H. Prediction of highly imbalanced semi- conductor chip-level defects using uncertainty-based adaptive margin learning.IISE Transactions.55, 147-155, 2022

    2022

  4. [4]

    Overview of industrial process automation,

    K. Sharma, “Overview of industrial process automation,”Elsevier, 2016

    2016

  5. [5]

    & Kim, H

    Cho, H., Koo, W. & Kim, H. Prediction of highly imbalanced semi- conductor chip-level defects in module tests using multimodal fusion and logit adjustment.IEEE Transactions on Semiconductor Manufacturing. 36, 425-433, 2023

    2023

  6. [6]

    Multivariate time-series anomaly detection via graph attention network,

    H. Zhao et al., “Multivariate time-series anomaly detection via graph attention network,” inProc. IEEE Int. Conf. Data Mining (ICDM), 2020, pp. 841–850

    2020

  7. [7]

    Graph neural network-based anomaly detection in multivariate time series,

    A. Deng and B. Hooi, “Graph neural network-based anomaly detection in multivariate time series,” inProc. AAAI Conf. Artif. Intell., 2021, pp. 4027–4035

    2021

  8. [8]

    & Kim, H

    Ma, E., Jung, J. & Kim, H. Order-based Causal Discovery for Multistage Processes.IEEE Transactions on Knowledge And Data Engineering. doi=10.1109/TKDE.2026.3710776. 2026

    arXiv 2026

  9. [9]

    Learning graph structures with transformer for multi- variate time-series anomaly detection in IoT,

    Z. Chen et al., “Learning graph structures with transformer for multi- variate time-series anomaly detection in IoT,”IEEE Internet Things J., vol. 9, no. 12, pp. 9179–9189, 2021

    2021

  10. [10]

    Deep variational graph convolutional recurrent network for multivariate time series anomaly detection,

    W. Chen et al., “Deep variational graph convolutional recurrent network for multivariate time series anomaly detection,” inProc. Int. Conf. Mach. Learn. (ICML), 2022, pp. 3621–3633

    2022

  11. [11]

    Learning sparse latent graph representations for anomaly detection in multivariate time series,

    S. Han and S. S. Woo, “Learning sparse latent graph representations for anomaly detection in multivariate time series,” inProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2022, pp. 2977–2986

    2022

  12. [12]

    Graph-augmented normalizing flows for anomaly detection of multiple time series,

    E. Dai and J. Chen, “Graph-augmented normalizing flows for anomaly detection of multiple time series,”arXiv preprint arXiv:2202.07857, 2022

    Pith/arXiv arXiv 2022

  13. [13]

    Robust anomaly detection for multivariate time series through temporal GCNs and attention-based V AE,

    Y . Shi et al., “Robust anomaly detection for multivariate time series through temporal GCNs and attention-based V AE,”Knowl.-Based Syst., vol. 275, p. 110725, 2023

    2023

  14. [14]

    MST-GAT: A multimodal spatial- temporal graph attention network for time series anomaly detection,

    C. Ding, S. Sun, and J. Zhao, “MST-GAT: A multimodal spatial- temporal graph attention network for time series anomaly detection,” Inf. Fusion, vol. 89, pp. 527–536, 2023

    2023

  15. [15]

    Edge conditional node update graph neural network for multi-variate time series anomaly detection,

    H. Jo and S.-W. Lee, “Edge conditional node update graph neural network for multi-variate time series anomaly detection,”Inf. Sci., p. 121062, 2024

    2024

  16. [16]

    Multivariate time- series anomaly detection based on enhancing graph attention networks with topological analysis,

    Z. Liu, X. Huang, J. Zhang, Z. Hao, J. Li, and R. Hu, “Multivariate time- series anomaly detection based on enhancing graph attention networks with topological analysis,” inProc. 33rd ACM Int. Conf. Inf. Knowl. Manag. (CIKM), 2024, pp. 1480–1490

    2024

  17. [17]

    PGRF-Net: A prototype-guided relational fusion network for diagnostic multivariate time-series anomaly detection,

    J. Jeong and H. Yoon, “PGRF-Net: A prototype-guided relational fusion network for diagnostic multivariate time-series anomaly detection,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2026

    2026

  18. [18]

    Graph attention networks,

    P. Velickovic et al., “Graph attention networks,”stat, vol. 1050, no. 20, pp. 10–48550, 2017

    2017

  19. [19]

    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

    Pith/arXiv arXiv 2018

  20. [20]

    Anomaly detection: A survey,

    V . Chandola, A. Banerjee, and V . Kumar, “Anomaly detection: A survey,” ACM Comput. Surv., vol. 41, no. 3, pp. 1–58, 2009. IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING 13

    2009

  21. [21]

    Deep learning for anomaly detection in time-series data: Review, analysis, and guidelines,

    K. Choi et al., “Deep learning for anomaly detection in time-series data: Review, analysis, and guidelines,”IEEE Access, vol. 9, pp. 120043– 120065, 2021

    2021

  22. [22]

    Deep learning for time series anomaly detection: A survey,

    Z. Z. Darban et al., “Deep learning for time series anomaly detection: A survey,”ACM Comput. Surv., 2022

    2022

  23. [23]

    Contextual anomaly detection for high- dimensional data using Dirichlet process variational autoencoder,

    H. Kim and H. Kim, “Contextual anomaly detection for high- dimensional data using Dirichlet process variational autoencoder,”IISE Transactions, vol. 55, no. 5, pp. 433–444, 2023

    2023

  24. [24]

    Contextual anomaly detection for multivariate time series data,

    H. Kim and H. Kim, “Contextual anomaly detection for multivariate time series data,”Quality Engineering, vol. 35, no. 4, pp. 686–695, 2023

    2023

  25. [25]

    Deep learning for anomaly detection in multivariate time series: Approaches, applications, and challenges,

    G. Li and J. J. Jung, “Deep learning for anomaly detection in multivariate time series: Approaches, applications, and challenges,”Inf. Fusion, vol. 91, pp. 93–102, 2023

    2023

  26. [26]

    & Kim, H

    Kim, H. & Kim, H. Deep embedding kernel mixture networks for conditional anomaly detection in high-dimensional data.International Journal of Production Research.61, 1101-1113, 2023

    2023

  27. [27]

    Detecting spacecraft anomalies using LSTMs and nonparametric dynamic thresholding,

    K. Hundman et al., “Detecting spacecraft anomalies using LSTMs and nonparametric dynamic thresholding,” inProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2018, pp. 387–395

    2018

  28. [28]

    Developing an unsupervised real-time anomaly detection scheme for time series with multi-seasonality,

    W. Wu et al., “Developing an unsupervised real-time anomaly detection scheme for time series with multi-seasonality,”IEEE Trans. Knowl. Data Eng., vol. 34, no. 9, pp. 4147–4160, 2020

    2020

  29. [29]

    Spacecraft anomaly detection with attention temporal convolution networks,

    L. Liu et al., “Spacecraft anomaly detection with attention temporal convolution networks,”Neural Comput. Appl., pp. 1–9, 2023

    2023

  30. [30]

    Time-series anomaly detection with stacked transformer representations and 1D convolutional network,

    J. Kim, H. Kang, and P. Kang, “Time-series anomaly detection with stacked transformer representations and 1D convolutional network,”Eng. Appl. Artif. Intell., vol. 120, p. 105964, 2023

    2023

  31. [31]

    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 Robot. Autom. Lett., vol. 3, no. 3, pp. 1544–1551, 2018

    2018

  32. [32]

    MAD-GAN: Multivariate anomaly detection for time series data with generative adversarial networks,

    D. Li et al., “MAD-GAN: Multivariate anomaly detection for time series data with generative adversarial networks,” inProc. Int. Conf. Artif. Neural Netw., 2019, pp. 703–716

    2019

  33. [33]

    Robust anomaly detection for multivariate time series through stochastic recurrent neural network,

    Y . Su et al., “Robust anomaly detection for multivariate time series through stochastic recurrent neural network,” inProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2019, pp. 2828–2837

    2019

  34. [34]

    STAD-GAN: Unsupervised anomaly detection on mul- tivariate time series with self-training generative adversarial networks,

    Z. Zhang et al., “STAD-GAN: Unsupervised anomaly detection on mul- tivariate time series with self-training generative adversarial networks,” ACM Trans. Knowl. Discovery Data, vol. 17, no. 5, pp. 1–18, 2023

    2023

  35. [35]

    USAD: UnSupervised anomaly detection on multi- variate time series,

    J. Audibert et al., “USAD: UnSupervised anomaly detection on multi- variate time series,” inProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2020, pp. 3395–3404

    2020

  36. [36]

    Temporal convolutional autoencoder for unsupervised anomaly detection in time series,

    M. Thill, W. Konen, H. Wang, and T. B ¨ack, “Temporal convolutional autoencoder for unsupervised anomaly detection in time series,”Appl. Soft Comput., vol. 112, p. 107751, 2021

    2021

  37. [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,”arXiv preprint arXiv:2201.07284, 2022

    Pith/arXiv arXiv 2022

  38. [38]

    Variational transformer-based anomaly detection ap- proach for multivariate time series,

    X. Wang et al., “Variational transformer-based anomaly detection ap- proach for multivariate time series,”Measurement, vol. 191, p. 110791, 2022

    2022

  39. [39]

    SARAD: Spatial association- aware anomaly detection and diagnosis for multivariate time series,

    Z. Dai, L. He, S.-H. Yang, and M. Leeke, “SARAD: Spatial association- aware anomaly detection and diagnosis for multivariate time series,” in Proc. 38th Conf. Neural Inf. Process. Syst. (NeurIPS), 2024, pp. 48371– 48410

    2024

  40. [40]

    CATCH: Channel-aware multivariate time series anomaly detection via frequency patching,

    X. Wu, X. Qiu, Z. Li, Y . Wang, J. Hu, C. Guo, H. Xiong, and B. Yang, “CATCH: Channel-aware multivariate time series anomaly detection via frequency patching,” inProc. 13th Int. Conf. Learn. Represent. (ICLR), 2025, pp. 17017–17045.“

    2025

  41. [41]

    Multi-resolution decomposable diffusion model for non-stationary time series anomaly detection,

    G. Zhong, P. Wang, J. Yuan, Z. Li, and L. Chen, “Multi-resolution decomposable diffusion model for non-stationary time series anomaly detection,” inProc. 13th Int. Conf. Learn. Represent. (ICLR), 2025

    2025

  42. [42]

    GTAD: Graph and temporal neural network for multi- variate time series anomaly detection,

    S. Guan et al., “GTAD: Graph and temporal neural network for multi- variate time series anomaly detection,”Entropy, vol. 24, no. 6, p. 759, 2022

    2022

  43. [43]

    Causality- aware contrastive learning for robust multivariate time-series anomaly detection,

    H. Kim, J. Mok, D. Lee, J. Lew, S. Kim, and S. Yoon, “Causality- aware contrastive learning for robust multivariate time-series anomaly detection,” inProc. 42nd Int. Conf. Mach. Learn. (ICML), 2025, pp. 30591–30608

    2025

  44. [44]

    CARLA: Self-supervised contrastive representation learning for time series anomaly detection,

    Z. Z. Darban, G. I. Webb, S. Pan, C. C. Aggarwal, and M. Salehi, “CARLA: Self-supervised contrastive representation learning for time series anomaly detection,”Pattern Recognit., vol. 157, 2025, Art. no. 110874

    2025

  45. [45]

    Attention is all you need,

    A. Vaswani et al., “Attention is all you need,” inAdv. Neural Inf. Process. Syst., vol. 30, 2017

    2017

  46. [46]

    Categorical reparameterization with Gumbel-softmax,

    E. Jang, S. Gu, and B. Poole, “Categorical reparameterization with Gumbel-softmax,”arXiv preprint arXiv:1611.01144, 2016

    Pith/arXiv arXiv 2016

  47. [47]

    DAGs with NO TEARS: Continuous optimization for structure learning,

    X. Zheng et al., “DAGs with NO TEARS: Continuous optimization for structure learning,” inAdv. Neural Inf. Process. Syst., vol. 31, 2018

    2018

  48. [48]

    Entropy causal graphs for multivariate time series anomaly detection,

    F. G. Febrinanto et al., “Entropy causal graphs for multivariate time series anomaly detection,”ACM Trans. Intell. Syst. Technol., vol. 16, no. 6, pp. 1–25, 2025

    2025

  49. [49]

    Semi-supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,”arXiv preprint arXiv:1609.02907, 2016

    Pith/arXiv arXiv 2016

  50. [50]

    Rectified linear units improve restricted Boltzmann machines,

    V . Nair and G. E. Hinton, “Rectified linear units improve restricted Boltzmann machines,” inProc. Int. Conf. Mach. Learn. (ICML), 2010, pp. 807–814

    2010

  51. [51]

    Weight normalization: A simple reparameterization to accelerate training of deep neural networks,

    T. Salimans and D. P. Kingma, “Weight normalization: A simple reparameterization to accelerate training of deep neural networks,” in Adv. Neural Inf. Process. Syst., vol. 29, 2016

    2016

  52. [52]

    Empirical evaluation of rectified activations in convolu- tional network,

    B. Xu et al., “Empirical evaluation of rectified activations in convolu- tional network,”arXiv preprint arXiv:1505.00853, 2015

    Pith/arXiv arXiv 2015

  53. [53]

    W ADI: A water distribution testbed for research in the design of secure cyber physical systems,

    C. M. Ahmed, V . R. Palleti, and A. P. Mathur, “W ADI: A water distribution testbed for research in the design of secure cyber physical systems,” inProc. 3rd Int. Workshop Cyber-Physical Syst. Smart Water Netw., 2017, pp. 25–28

    2017

  54. [54]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    Pith/arXiv arXiv 2014

  55. [55]

    WaveNet: A generative model for raw audio,

    A. van den Oordet al., “WaveNet: A generative model for raw audio,” 2016, arXiv:1609.03499

    Pith/arXiv arXiv 2016