Pith. sign in

REVIEW 4 major objections 5 minor 58 references

ST-Tree with Interpretability for Multivariate Time Series Classification

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read ST-Tree combines a Swin Transformer feature extractor with a prototype-based neural tree, reaching the best average accuracy (0.789) on ten UEA multivariate time-series benchmarks and making each classification traceable to the…

desk verdict A plausible new hybrid of Swin Transformer and prototype neural tree whose accuracy evidence is thin and whose interpretability claim is untested. read the letter →

arxiv 2411.11620 v1 pith:V22MGOCJ submitted 2024-11-18 cs.LG cs.AI

Mingsen Du , Yanxuan Wei , Yingxia Tang , Xiangwei Zheng , Shoushui Wei , Cun Ji This is my paper · ORCID
classification cs.LGcs.AI
keywords multivariatetimeseriesclassificationinterpretabilityneuraltreeSwinTransformerprototypelearningpatchesUEAbenchmarkdatasets
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

ST-Tree is a hybrid classifier: a Swin Transformer—a transformer that reads a time series through local patches—extracts features, and a prototype-based neural tree then routes those features through a binary tree toward a label. The authors' central claim is that this combination reaches transformer-level accuracy, averaging 0.789 over ten public UEA multivariate time-series benchmarks and beating every method they compare it with, while making the decision process legible: each internal tree node matches the input against a trainable prototype segment of a time series and decides left or right by that similarity. The payoff for a practitioner is a classifier that is not just accurate but has a visible decision trail, which matters in settings like medical or safety-critical monitoring where an unexplained prediction is hard to trust. Ablations show that removing either the tree or the attention module lowers average accuracy, and the paper illustrates the learned prototype segments on several datasets.

What carries the argument

The load-bearing object is a perfect binary tree of trainable prototypes placed on top of the Swin Transformer's time-patch features. Each branch node owns one prototype $P_i$; squared Euclidean distance between patches and the prototype is computed with convolution, converted to a log-similarity score, and max-pooled to a scalar routing weight (left = $N(z_i)$, right = $1 - N(z_i)$). The routing weights are then used recursively so that every leaf receives a share of the input, and the leaf's class prediction combines a tree FC layer with the global ST-layer FC output. The time-patch module with shifted-window self-attention and channel/spatial attention supplies the patches that the tree routes.

What would settle it

Set the routing weights of the single most-highlighted path to zero (or mask that path's prototypes) on the ten datasets and measure how often the predicted label changes; if the visualized path is the real decision mechanism, predictions should flip in a sizable fraction of cases, whereas if they barely move the node-level similarities are not causal and the interpretability claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a Swin Transformer feature extractor and a prototype-routing neural tree can be trained jointly so that the tree's branch decisions are based on concrete time-series segments, making the classifier's reasoning inspectable without sacrificing accuracy. At each internal node $i$, the model computes the maximum similarity $N(z_i) = \log\left(1 + \max_{\tilde{z} \in \text{patches}(z_i)} \frac{1}{\|\tilde{z} - P_i\|_2^2 + \epsilon}\right)$ between the features and the node's trainable prototype $P_i$, and uses it as the soft probability of going left, with right probability $1 - N(z_i)$. Leaves return class scores, and the final prediction is the routing-weighted sum over all leaves, $\hat{y} = \rho^T g(x)$. On the ten UEA datasets the authors report an average accuracy of 0.789, above every baseline in their comparison, and they present figures in which the prototype segment matched at each node is overlaid on the original time series to trace the decision path.

Load-bearing premise

The interpretability claim assumes the node-level similarity scores faithfully show what actually drives the final prediction, even though the final answer is a weighted blend of all leaves; if the highlighted prototypes are mere correlates rather than causes, the visualizations would be illustrative rather than explanatory.

Editorial extensions

If this is right

  • On the ten UEA datasets, ST-Tree's average accuracy is 0.789, the highest among the twelve compared classifiers, so the prototype-tree addition does not cost accuracy relative to the transformer baselines.
  • Ablations show the neural tree raises average accuracy from 0.739 to 0.789 and the attention module from 0.761 to 0.789, so both components carry weight.
  • Tree depth needs per-dataset tuning: on AWR and SR2 deeper trees hurt accuracy, while NATOPS improves when depth moves from 3 to 4, meaning the tree should be treated as a hyperparameter.
  • The model is especially effective on long series, winning on StandWalkJump (2500 time steps), consistent with the shifted-window patch design.
  • The node-prototype visualizations give a concrete, segment-level account of what the model matches at each routing decision, which is the paper's interpretability deliverable.

Reading between the lines

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

  • Because the final prediction in Eq. (28) averages over all leaves, what the visualizations show is a soft aggregate path, not a single hard decision path; a faithful-explanations test (e.g., zeroing the top path's routing weights and watching accuracy) would be needed to confirm the prototypes are causes rather than correlates.
  • The same per-node maximum-similarity score could serve as a built-in confidence or out-of-distribution detector, since inputs that match no prototype closely should receive low routing mass everywhere; the paper does not test this.
  • The channel/spatial attention map computed inside each node could be rendered per channel to say which sensor channels drove a branch, extending the current segment-only visualizations at little extra cost.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes ST-Tree, a hybrid model for multivariate time series classification (MTS) that combines a Swin Transformer backbone with a prototype-based neural tree. The neural tree computes soft routing scores from the maximum similarity between time patches and per-node prototype vectors, and the final prediction is a weighted mixture of leaf predictions. The authors evaluate on 10 UEA benchmark datasets, reporting an average accuracy of 0.789, and they visualize tree paths and prototype segments as evidence of interpretability. The central claim is that ST-Tree improves accuracy over strong baselines while providing a transparent decision process.

Significance. If the interpretability claim were substantiated, the paper would offer a practically useful contribution: a relatively accurate deep model whose decisions can be traced to prototype time-series segments. The authors also release code and parameters, which supports reproducibility. However, the current evidence does not establish the two conjuncts of the central claim. The accuracy advantage over MTSC FF is 0.006 on average with no variance or significance testing, and the interpretability claim is asserted from visualizations without any faithfulness analysis. The architecture itself makes the interpretability claim nontrivial, because the final prediction is a soft mixture over all leaves. Given the importance of interpretable MTS classification and the public benchmark setting, the work has potential, but the current manuscript is not yet convincing on either load-bearing point.

major comments (4)
  1. [§4.4, Eq. (28)] The interpretability claim is not supported. Eq. (28) computes yhat = rho^T g(x), a weighted average over all leaves, and each leaf logit in Eq. (24) includes the global term timeFC(z0). Consequently, no single tree path determines the output. The visualizations in Figs. 6-9 do not define how the displayed path is selected, do not measure whether ignoring other leaves changes the prediction, and report no faithfulness metric or human evaluation. To support the second half of the central claim, the paper should add a faithfulness analysis: for example, measure the contribution of the top-1 path to the final logit, prune or mask non-selected leaves and report the change in prediction, or compare the visualized path with human judgment on a sample of instances.
  2. [Table 3, Avg row] The claimed accuracy improvement is not statistically substantiated. ST-Tree's average accuracy is 0.789 versus 0.783 for MTSC FF, a difference of 0.006 that comes from a single run without standard deviations or a significance test. The text in §4.2 also says 'MTSCFF has the highest average accuracy (0.789)', which contradicts Table 3, where ST-Tree attains 0.789 and MTSC FF 0.783. Please report mean and standard deviation over multiple seeds, and apply an appropriate significance test (e.g., Wilcoxon signed-rank or the CD test) before claiming improved accuracy.
  3. [Table 4] The ablation results lack error bars or significance tests. The average gain of ST-Tree over ST-Tree (without tree) is 0.050 (0.789 vs 0.739) and over ST-Tree (without attention) is 0.028 (0.789 vs 0.761), but without variance estimates it is unclear whether these differences are reliable. Moreover, the 'Win 9' count is not explained: in the comparison against ST-Tree (without tree), the full model loses on CharacterTrajectories and ties on SpokenArabicDigits. Please clarify the counting and report repeated-run statistics.
  4. [§3.4.1, Eqs. (6)-(12)] The definition of the similarity score is internally inconsistent and ambiguous. Eq. (6) defines N(zi) = log(1 + max 1/(||tilde z - Pi||^2 + epsilon)), while Eq. (12) gives a different expression for similarity involving distance + 1 × epsilon. The notation in Eq. (13), '[N(zi)]1 0', and Eq. (15), 'maxim:,0', is undefined. Because the routing score is the basis of both the decision process and the interpretability claim, the exact formula and indexing must be stated unambiguously.
minor comments (5)
  1. [§4.2] The text says 'totaling 9 methods' but then lists 12 methods in Table 3; please correct this inconsistency.
  2. [§4.2] The method is referred to as '1NN-DTW' in Table 3 but as 'DTW-1NN' in the text; please use one name consistently.
  3. [Figs. 6-9] The interpretability figures lack axis labels, class labels, and a clear explanation of how the highlighted prototype segments and arrows are chosen; the captions should be self-contained.
  4. [Algorithm 3, line 24] The learning-rate update formula 'learning rate ← learning rate × decay rate(e/decay steps)' is ambiguous; it should likely be learning rate × decay_rate^(e/decay_steps).
  5. [Eq. (24)] The sentence 'treeFC is obtained based on a certain path, and timeFC is obtained through time patch module' is unclear; please specify how the two terms are computed and combined.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: accuracy is externally benchmarked and interpretability is a faithfulness concern, not a definitional reduction.

full rationale

The paper's accuracy claim is anchored to external UEA train/test splits reported in Table 2, with results in Table 3 and a public code release, so the accuracy comparison is not circular. No parameter is fitted to a subset of data and then reported as a prediction of a closely related quantity; the tree-depth sensitivity analysis in Section 4.5 does not alter the main three-layer results in Table 3. The related-work self-citations (Du et al. 2023a,b, 2024; Ji et al. 2022a,b) are descriptive background and comparison references, not load-bearing premises, and no uniqueness theorem or ansatz is imported from the authors' prior work. The interpretability claim relies on visualizing learned prototypes and routing scores (Eqs. 6, 13, 23, 24, 28). Showing a learned prototype as an explanation is self-referential in the broad sense that the prototype is itself a learned parameter used in prediction, but this is the standard design of prototype-based interpretability rather than a derivation that reduces to its own input. A possible weakness is that Eq. 28 computes the final prediction as a weighted average over all leaves, so a single visualized path may not fully determine the output; however, that is an empirical faithfulness question, not a circularity of the kind defined by the analysis rules. No specific equation or fitted quantity can be exhibited as equivalent to another by construction, so the appropriate finding is no significant circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The central result is empirical. The main load-bearing choices are the fixed tree depth and the unvalidated equivalence between soft routing scores and interpretable decisions. No external physical entities are introduced beyond learned prototypes.

free parameters (5)
  • Tree depth d = 3 by default, varied 3 to 6 in Section 4.5
    Determines the number of prototypes and routing decisions; selected per dataset via sensitivity analysis on test accuracy, which risks fitting to the test set.
  • Prototype kernel length k = not reported
    Length of prototype vectors used as convolution kernels in Eq. 6 through Eq. 11; not specified, yet it affects the similarity computation and the interpretability of prototypes.
  • Epsilon epsilon in similarity denominator = not reported
    Added in Eq. 6 and Eq. 12 to stabilize division; the value is not given.
  • Number of ST layers and training epochs = one ST layer, 50 epochs
    Chosen by hand with the rationale that more layers cause overfitting and 50 epochs suffice; no sensitivity evidence is shown for the ST layer count.
  • Learning rate hyperparameters = lr 0.001, decay 0.9 every 5 epochs
    Standard training choices, but the learning-rate decay is described inconsistently between Section 4.1.3 and Algorithm 3.
assumptions (5)
  • domain assumption Swin Transformer self-attention and end-to-end backpropagation provide a suitable feature extractor for MTS patches.
    Adopted from Liu et al. (2021) without re-derivation; the paper relies on this for the time patch module in Section 3.2.
  • domain assumption The 10 UEA datasets and the published accuracies of baselines are directly comparable to ST-Tree's results.
    Table 3 mixes numbers from prior papers with a single run of ST-Tree; no joint experimental protocol is described in Section 4.2.1.
  • ad hoc to paper Soft routing scores computed from maximum similarity to learned prototypes make the decision process transparent.
    This is the interpretability premise in Section 3.4.1; it is asserted, not demonstrated via human or quantitative evaluation.
  • ad hoc to paper A perfect binary tree with fixed depth is an appropriate inductive bias for all MTS datasets.
    Section 3.3.2 fixes a full binary tree, while Section 4.5 shows the optimal depth varies per dataset, so the fixed structure is a convenience assumption.
  • standard math Cross-entropy loss and the Adam optimizer find parameters that satisfy both accuracy and prototype interpretability goals.
    Routine optimization assumption; the objective contains no term that explicitly encourages interpretable prototypes.
invented entities (1)
  • Per-node prototype vector P_i
    purpose: Serves as a learned template; the routing score is the maximum similarity between patch features and prototypes (Eq. 6), and visualization of the nearest patch is presented as interpretation.
    Internal learned parameters with no external falsifiable handle; their explanatory value is asserted rather than tested.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ST-Tree with Interpretability for Multivariate Time Series Classification." pith.science (2026). https://pith.science/paper/V22MGOCJ

@misc{pith2026241111620,
  author       = {Pith},
  title        = {Pith review of: ST-Tree with Interpretability for Multivariate Time Series Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V22MGOCJ}},
  note         = {Machine review of arXiv:2411.11620}
}
read the original abstract

Multivariate time series classification is of great importance in practical applications and is a challenging task. However, deep neural network models such as Transformers exhibit high accuracy in multivariate time series classification but lack interpretability and fail to provide insights into the decision-making process. On the other hand, traditional approaches based on decision tree classifiers offer clear decision processes but relatively lower accuracy. Swin Transformer (ST) addresses these issues by leveraging self-attention mechanisms to capture both fine-grained local patterns and global patterns. It can also model multi-scale feature representation learning, thereby providing a more comprehensive representation of time series features. To tackle the aforementioned challenges, we propose ST-Tree with interpretability for multivariate time series classification. Specifically, the ST-Tree model combines ST as the backbone network with an additional neural tree model. This integration allows us to fully leverage the advantages of ST in learning time series context while providing interpretable decision processes through the neural tree. This enables researchers to gain clear insights into the model's decision-making process and extract meaningful interpretations. Through experimental evaluations on 10 UEA datasets, we demonstrate that the ST-Tree model improves accuracy in multivariate time series classification tasks and provides interpretability through visualizing the decision-making process across different datasets.

Figures

Figures reproduced from arXiv: 2411.11620 by the authors.

Figure 1
Figure 1. ST-Tree’s transparent decision-making process, with the original Epilepsy time [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Schematic of ST-Tree classification via tree structure, with the original Epilepsy [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Schematic of ST module. 9 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: CD diagram of 9 implementations on 10 UEA datasets. [PITH_FULL_IMAGE:figures/full_fig_p021_4.png]
Figure 5
Figure 5. Figure 5: Comparison of the accuracy of ST-Tree and ST-Tree (without tree), ST-Tree [PITH_FULL_IMAGE:figures/full_fig_p024_5.png]
Figure 6
Figure 6. Figure 6: CharacterTrajectories. dim 1 dim 3 dim 2 dim 4 dim 5 dim 6 dim 3 dim 2 dim 1 [PITH_FULL_IMAGE:figures/full_fig_p024_6.png]
Figure 8
Figure 8. Figure 8: AtrialFibrillation. dim 0 dim 2 dim 3 dim 1 dim 4 dim 5 dim 6 dim 6 dim 0 dim 1 [PITH_FULL_IMAGE:figures/full_fig_p025_8.png]
Figure 9
Figure 9. Figure 9: ArticularyWordRecognition. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_9.png]
Figure 10
Figure 10. Figure 10: Variation in the precision of the number of layers of the corresponding tree for [PITH_FULL_IMAGE:figures/full_fig_p026_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 25 canonical work pages

  1. [1]

    G:)PYmPY;5) 2VY ї ?f - | 12 &

    + is cited as + ESG96 +. In connection with cross-referencing and possible future hyperlinking it is not a good idea to collect more that one literature item in one + +. The so-called Harvard or author-year style of referencing is enabled by the package natbib . With this package the literature can be cited as follows: enumerate [ ] Parenthetical: + WB96 ...

    work page 1996

  2. [2]

    write newline

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

  3. [3]

    write newline

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

  4. [4]

    write newline

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

  5. [5]

    write newline

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

  6. [6]

    author Bahri, O. B. , author Li, P. , author Boubrahimi, S. F. , & author Hamdi, S. M. ( year 2022 ). title Shapelet-based temporal association rule mining for multivariate time series classification . journal 2022 IEEE International Conference on Big Data (Big Data) \/ , (pp. pages 242--251 ). :10.1109/BigData55660.2022.10020478

    arXiv 2022

  7. [7]

    , author Galtier, M

    author Chambon, S. , author Galtier, M. N. , author Arnal, P. J. , author Wainrib, G. , & author Gramfort, A. ( year 2017 ). title A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series . journal IEEE Transactions on Neural Systems and Rehabilitation Engineering \/ , volume 26 \/ , pages 758--769...

    work page arXiv 2017

  8. [8]

    , author Li, O

    author Chen, C. , author Li, O. , author Barnett, A. J. , author Su, J. , & author Rudin, C. ( year 2018 ). title This looks like that: deep learning for interpretable image recognition . In booktitle Neural Information Processing Systems \/

    work page 2018

Show all 58 references
  1. [9]

    , & author Wan, Y

    author Chen, J. , & author Wan, Y. ( year 2023 ). title Localized shapelets selection for interpretable time series classification . journal Applied Intelligence \/ , volume 53 \/ , pages 17985--18001 . :10.1007/s10489-022-04422-2

    2023 doi

  2. [10]

    , & author Shi, K

    author Chen, W. , & author Shi, K. ( year 2021 ). title Multi-scale attention convolutional neural network for time series classification . journal Neural Networks \/ , volume 136 \/ , pages 126--140 . :10.1016/j.neunet.2021.01.001

    2021 doi

  3. [11]

    , author Hu, B

    author Chen, Y. , author Hu, B. , author Keogh, E. J. , & author Batista, G. E. A. P. A. ( year 2013 ). title Dtw-d: time series semi-supervised learning from a single example . journal Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data...

    2013

  4. [12]

    , author Liu, Q

    author Cheng, M. , author Liu, Q. , author Liu, Z. , author Li, Z. , author Luo, Y. , & author Chen, E. ( year 2023 ). title Formertime: Hierarchical multi-scale representations for multivariate time series classification . journal Proceedings of the ACM Web Conference 2023 \/...

    2023

  5. [13]

    , author Greene, D

    author Delaney, E. , author Greene, D. , & author Keane, M. T. ( year 2020 ). title Instance-based counterfactual explanations for time series classification . journal ArXiv \/ , volume abs/2009.13211 \/ . :10.1007/978-3-030-86957-1_3

    2020 arXiv

  6. [14]

    , author Petitjean, F

    author Dempster, A. , author Petitjean, F. , & author Webb, G. I. ( year 2019 ). title Rocket: exceptionally fast and accurate time series classification using random convolutional kernels . journal Data Mining and Knowledge Discovery \/ , volume 34 \/ , pages 1454 -- 1495 . :...

    2019 doi

  7. [15]

    , author Wei, Y

    author Du, M. , author Wei, Y. , author Hu, Y. , author Zheng, X. , & author Ji, C. ( year 2023 a ). title Automatic component identification based on time series classification for intelligent devices . journal 2023 IEEE International Conference on Data Mining Workshops (ICDM...

    2023

  8. [16]

    , author Wei, Y

    author Du, M. , author Wei, Y. , author Hu, Y. , author Zheng, X. , & author Ji, C. ( year 2024 ). title Multivariate time series classification based on fusion features . journal Expert Systems with Applications \/ , . :10.1016/j.eswa.2024.123452

    2024

  9. [17]

    , author Wei, Y

    author Du, M. , author Wei, Y. , author Zheng, X. , & author Ji, C. ( year 2023 b ). title Multi-feature based network for multivariate time series classification . journal Inf. Sci. \/ , volume 639 \/ , pages 119009 . :10.1016/j.ins.2023.119009

    2023

  10. [18]

    , author Ragab, M

    author Eldele, E. , author Ragab, M. , author Chen, Z. , author Wu, M. , author Kwoh, C. , author Li, X. , & author Guan, C. ( year 2021 ). title Time-series representation learning via temporal and contextual contrasting . In booktitle International Joint Conference on Artifi...

    2021 doi

  11. [19]

    , author Wang, P

    author Fang, Z. , author Wang, P. , & author Wang, W. ( year 2018 ). title Efficient learning interpretable shapelets for accurate time series classification . journal 2018 IEEE 34th International Conference on Data Engineering (ICDE) \/ , (pp. pages 497--508 ). :10.1109/ICDE....

    2018

  12. [20]

    , author Dieuleveut, A

    author Franceschi, J.-Y. , author Dieuleveut, A. , & author Jaggi, M. ( year 2019 ). title Unsupervised scalable representation learning for multivariate time series . journal ArXiv \/ , volume abs/1901.10738 \/

    2019 arXiv

  13. [21]

    , author Li, H

    author Fu, K. , author Li, H. , & author Shi, X. ( year 2024 ). title Ctf-former: A novel simplified multi-task learning strategy for simultaneous multivariate chaotic time series prediction . journal Neural Networks \/ , volume 174 \/ , pages 106234 . :10.1016/j.neunet.2024.106234

    2024

  14. [22]

    , & author Cook, D

    author Ghods, A. , & author Cook, D. J. ( year 2022 ). title Pip: Pictorial interpretable prototype learning for time series classification . journal IEEE Computational Intelligence Magazine \/ , volume 17 \/ , pages 34--45 . :10.1109/MCI.2021.3129957

    2022

  15. [23]

    , author Kulbach, C

    author H \"o llig, J. , author Kulbach, C. , & author Thoma, S. ( year 2022 ). title Tsevo: Evolutionary counterfactual explanations for time series classification . journal 2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA) \/ , (pp. pages 29...

    2022

  16. [24]

    , author Kwok, J

    author Hou, L. , author Kwok, J. T.-Y. , & author Zurada, J. M. ( year 2016 ). title Efficient learning of timeseries shapelets . In booktitle AAAI Conference on Artificial Intelligence \/ . :10.1609/aaai.v30i1.10178

    2016 doi

  17. [25]

    , & author Deng, Y

    author Huang, F. , & author Deng, Y. ( year 2023 ). title Tcgan: Convolutional generative adversarial network for time series classification and clustering . journal Neural Networks \/ , volume 165 \/ , pages 868--883 . :10.1016/j.neunet.2023.06.033

    2023 doi

  18. [26]

    author Ismail, A. A. , author Gunady, M. K. , author Bravo, H. C. , & author Feizi, S. ( year 2020 ). title Benchmarking deep learning interpretability in time series predictions . journal ArXiv \/ , volume abs/2010.13924 \/

    2020 arXiv

  19. [27]

    , author Du, M

    author Ji, C. , author Du, M. , author Hu, Y. , author Liu, S. , author Pan, L. , & author Zheng, X. ( year 2022 a ). title Time series classification based on temporal features . journal Appl. Soft Comput. \/ , volume 128 \/ , pages 109494 . :10.1016/j.asoc.2022.109494

    2022

  20. [28]

    , author Hu, Y

    author Ji, C. , author Hu, Y. , author Liu, S. , author Pan, L. , author Li, B. , & author Zheng, X. ( year 2022 b ). title Fully convolutional networks with shapelet features for time series classification . journal Inf. Sci. \/ , volume 612 \/ , pages 835--847 . :10.1016/j.i...

    2022 doi

  21. [29]

    , author Zhao, C

    author Ji, C. , author Zhao, C. , author Liu, S. , author Yang, C. , author Pan, L. , author Wu, L. , & author Meng, X. ( year 2019 ). title A fast shapelet selection algorithm for time series classification . journal Comput. Networks \/ , volume 148 \/ , pages 231--240 . :10....

    2019 doi

  22. [30]

    , author Majumdar, S

    author Karim, F. , author Majumdar, S. , author Darabi, H. , & author Harford, S. ( year 2018 ). title Multivariate lstm-fcns for time series classification . journal Neural networks : the official journal of the International Neural Network Society \/ , volume 116 \/ , pages ...

    2018 doi

  23. [31]

    , author Nam, J

    author Kim, S. , author Nam, J. Y. , & author Ko, B. ( year 2022 ). title Vit-net: Interpretable vision transformers with neural tree decoder . In booktitle International Conference on Machine Learning \/

    2022

  24. [32]

    , author Lindgren, T

    author Lee, Z. , author Lindgren, T. , & author Papapetrou, P. ( year 2023 ). title Z-time: efficient and effective interpretable multivariate time series classification . journal Data Mining and Knowledge Discovery \/ , (pp. pages 1--31 ). :10.1007/s10618-023-00969-x

    2023 doi

  25. [33]

    , author Choi, B

    author Li, G. , author Choi, B. , author Xu, J. , author Bhowmick, S. S. , author Chun, K.-P. , & author Wong, G. L. ( year 2021 ). title Shapenet: A shapelet-neural network approach for multivariate time series classification . In booktitle AAAI Conference on Artificial Intel...

    2021 doi

  26. [34]

    , author Zhang, L

    author Li, Y. , author Zhang, L. , author Zhu, L. , author Liu, L. , author Han, B. , author Zhang, Y. , & author Wei, S. ( year 2023 ). title Diagnosis of atrial fibrillation using self-complementary attentional convolutional neural network . journal Computer Methods and Prog...

    2023

  27. [35]

    , author Jin, X

    author Li, Z. , author Jin, X. , & author Zhao, X. ( year 2015 ). title Drunk driving detection based on classification of multivariate time series. journal Journal of safety research \/ , volume 54 \/ , pages 61--4 . :10.1016/j.jsr.2015.06.007

    2015 doi

  28. [36]

    , author Lin, Y

    author Liu, Z. , author Lin, Y. , author Cao, Y. , author Hu, H. , author Wei, Y. , author Zhang, Z. , author Lin, S. , & author Guo, B. ( year 2021 ). title Swin transformer: Hierarchical vision transformer using shifted windows . journal 2021 IEEE/CVF International Conferenc...

    2021

  29. [37]

    , author Li, W

    author Ma, H. , author Li, W. , author Zhang, X. , author Gao, S. , & author Lu, S. ( year 2019 ). title Attnsense: Multi-level attention mechanism for multimodal human activity recognition . In booktitle International Joint Conference on Artificial Intelligence \/ . :10.24963...

    2019 doi

  30. [38]

    , author Zhuang, W

    author Ma, Q. , author Zhuang, W. , author Li, S. , author Huang, D. , & author Cottrell, G. ( year 2020 ). title Adversarial dynamic shapelet networks . In booktitle AAAI Conference on Artificial Intelligence \/ . :10.1609/AAAI.V34I04.5948

    2020 doi

  31. [39]

    , author Pagliarini, G

    author Manzella, F. , author Pagliarini, G. , author Sciavicco, G. , & author Stan, E. I. ( year 2021 ). title Interval temporal random forests with an application to covid-19 diagnosis . In booktitle Time \/ . :10.4230/LIPIcs.TIME.2021.7

    2021 doi

  32. [40]

    , author Vickers, W

    author Middlehurst, M. , author Vickers, W. D. , & author Bagnall, A. ( year 2019 ). title Scalable dictionary classifiers for time series classification . journal ArXiv \/ , volume abs/1907.11815 \/ . :10.1007/978-3-030-33607-3_2

    2019 arXiv

  33. [41]

    author Nguyen, T. L. , author Gsponer, S. , author Ilie, I. , author O'Reilly, M. , & author Ifrim, G. ( year 2019 ). title Interpretable time series classification using linear models and multi-resolution multi-domain symbolic representations . journal Data Mining and Knowled...

    2019 doi

  34. [42]

    , author Scaboro, S

    author Pagliarini, G. , author Scaboro, S. , author Serra, G. , author Sciavicco, G. , & author Stan, E. I. ( year 2022 ). title Neural-symbolic temporal decision trees for multivariate time series classification . In booktitle Time \/ . :10.4230/LIPIcs.TIME.2022.13

    2022 doi

  35. [43]

    , author Puget, R

    author Rojat, T. , author Puget, R. , author Filliat, D. , author Ser, J. D. , author Gelin, R. , & author D'iaz-Rodr'iguez, N. ( year 2021 ). title Explainable artificial intelligence (xai) on timeseries data: A survey . journal ArXiv \/ , volume abs/2104.00950 \/

    2021 arXiv

  36. [44]

    author Ruiz, A. P. , author Flynn, M. , author Large, J. , author Middlehurst, M. , & author Bagnall, A. ( year 2020 ). title The great multivariate time series classification bake off: a review and experimental evaluation of recent algorithmic advances . journal Data Mining a...

    2020 doi

  37. [45]

    , author Liu, L

    author Tang, W. , author Liu, L. , & author Long, G. ( year 2019 ). title Few-shot time-series classification with dual interpretability . https://api.semanticscholar.org/CorpusID:211167393

    2019

  38. [46]

    , author Yang, J

    author Tao, Q. , author Yang, J. , & author Jing, S. ( year 2022 ). title Profile-matrix-based shapelet discovery for time series binary classification . journal 2022 18th International Conference on Computational Intelligence and Security (CIS) \/ , (pp. pages 297--301 ). :10...

    2022

  39. [47]

    , author Spinnato, F

    author Theissler, A. , author Spinnato, F. , author Schlegel, U. , & author Guidotti, R. ( year 2022 ). title Explainable ai for time series classification: A review, taxonomy and research directions . journal IEEE Access \/ , volume 10 \/ , pages 100700--100724 . :10.1109/ACC...

    2022

  40. [48]

    , author Eytan, D

    author Tonekaboni, S. , author Eytan, D. , & author Goldenberg, A. ( year 2021 ). title Unsupervised representation learning for time series with temporal neighborhood coding . journal ArXiv \/ , volume abs/2106.00750 \/

    2021 arXiv

  41. [49]

    , author Bjelogrlic, M

    author Turb'e, H. , author Bjelogrlic, M. , author Lovis, C. , & author Mengaldo, G. ( year 2022 ). title Interprettime: a new approach for the systematic evaluation of neural-network interpretability in time series classification . journal ArXiv \/ , volume abs/2202.05656 \/

    2022 arXiv

  42. [50]

    , author Xia, N

    author Wan, J. , author Xia, N. , author Yin, Y. , author Pan, X. , author Hu, J. , & author Yi, J. ( year 2024 ). title Tcdformer: A transformer framework for non-stationary time series forecasting based on trend and change-point detection . journal Neural Networks \/ , volum...

    2024

  43. [51]

    , author Cen, L

    author Wan, X. , author Cen, L. , author Chen, X. , author Xie, Y. , & author Gui, W. ( year 2023 ). title Memory shapelet learning for early classification of streaming time series. journal IEEE transactions on cybernetics \/ , volume PP \/ . :10.1109/TCYB.2023.3337550

    2023

  44. [52]

    , author Yan, W

    author Wang, Z. , author Yan, W. , & author Oates, T. ( year 2016 ). title Time series classification from scratch with deep neural networks: A strong baseline . journal 2017 International Joint Conference on Neural Networks (IJCNN) \/ , (pp. pages 1578--1585 ). :10.1109/IJCNN...

    2016

  45. [53]

    , author Wang, X

    author Yang, C. , author Wang, X. , author Yao, L. , author Long, G. , & author Xu, G. ( year 2023 ). title Dyformer: A dynamic transformer-based architecture for multivariate time series classification . journal Inf. Sci. \/ , volume 656 \/ , pages 119881 . :10.2139/ssrn.4362666

    2023 doi

  46. [54]

    , author Ahmadi, Z

    author Younis, R. , author Ahmadi, Z. , author Hakmeh, A. , & author Fisichella, M. ( year 2023 ). title Flames2graph: An interpretable federated multivariate time series classification framework . journal Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery an...

    2023

  47. [55]

    , author Zeng, X

    author Yu, Y. , author Zeng, X. , author Xue, X. , & author Ma, J. ( year 2022 ). title Lstm-based intrusion detection system for vanets: A time series classification approach to false message detection . journal IEEE Transactions on Intelligent Transportation Systems \/ , vol...

    2022

  48. [56]

    , author Jayaraman, S

    author Zerveas, G. , author Jayaraman, S. , author Patel, D. , author Bhamidipaty, A. , & author Eickhoff, C. ( year 2020 ). title A transformer-based framework for multivariate time series representation learning . journal Proceedings of the 27th ACM SIGKDD Conference on Know...

    2020

  49. [57]

    , & author Hill, D

    author Zhu, L. , & author Hill, D. ( year 2021 ). title Networked time series shapelet learning for power system transient stability assessment . journal IEEE Transactions on Power Systems \/ , volume 37 \/ , pages 416--428 . :10.1109/tpwrs.2021.3093423

    2021

  50. [58]

    , author Li, G

    author Zuo, R. , author Li, G. , author Choi, B. , author Bhowmick, S. S. , author yin Mah, D. N. , & author Wong, G. L. ( year 2023 ). title Svp-t: A shape-level variable-position transformer for multivariate time series classification . In booktitle AAAI Conference on Artifi...

    2023 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.