An Empirical Evaluation of Factors Affecting SHAP Explanation of Time Series Classification

Davide Italo Serramazza; Georgiana Ifrim; Nikos Papadeas; Zahraa Abdallah

REVIEW 4 major objections 5 minor 28 references

An Empirical Evaluation of Factors Affecting SHAP Explanation of Time Series Classification

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read For SHAP explanations of time series, the simple equal-length segmentation beats most custom segmentation algorithms.

desk verdict A useful empirical benchmark on segmentation choices for SHAP on time series, but the central 'segment count matters more than method' claim is not actually tested as stated. read the letter →

arxiv 2509.03649 v1 pith:73AA5CM3 submitted 2025-09-03 cs.AI

Davide Italo Serramazza , Nikos Papadeas , Zahraa Abdallah , Georgiana Ifrim This is my paper · ORCID

classification cs.AI

keywords timeseriesclassificationSHAPsegmentationexplainableAIattributionnormalizationInterpretAUCDXAIevaluation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that when SHAP is applied to time series through segment-based feature aggregation, the number of segments matters more than the segmentation algorithm that produces them. It compares eight segmentation methods across five datasets and four classifiers and finds that equal-length segmentation scores as well as or better than most custom change-point-detection approaches, at a fraction of the cost. The paper also argues that dividing each segment's SHAP value by the segment length, preserving the Shapley additivity property at the timepoint level, consistently improves attribution scores under both evaluation measures. A reader should care because the paper offers a cheaper and more defensible default for explaining time series classifiers, while flagging that the evaluation methods themselves are not neutral.

What carries the argument

The load-bearing objects are: (1) segment-based SHAP, where consecutive timepoints are grouped into features so the Shapley computation becomes tractable; (2) the eight segmenters, with equal-length segmentation as the cost-free baseline; (3) the length-weighted normalisation phi_S/|S|, which restores segment-level additivity after a segment's value is spread over its timepoints; and (4) the two evaluators, InterpretTime (AUCSE and F-score) and AUCD, which perturb time series according to attribution rankings to measure explanation quality.

What would settle it

Repeat the segmentation comparison on the same datasets and classifiers using a fidelity measure that does not depend on noise-augmented retraining, for example deletion/insertion with the original model across multiple random seeds, or a domain-expert oracle on Gunpoint and KeplerLightCurves. If equal-length segmentation no longer ranks in the top group under that metric, the paper's central conclusion does not transfer.

Watch

Extended reading notes

Core claim

The central discovery is that segment-level SHAP explanations for time series do not need elaborate segmentation algorithms: the number of segments is the dominant factor in explanation quality, and equal-length segmentation is the best default, matched by most custom methods only when they produce similar partitions. A second, more general finding is that attributions remain theoretically sound only if segment-level Shapley values are distributed uniformly over the timepoints of each segment, dividing the segment's value by its length; doing so preserves segment-level additivity and empirically raises explainability scores, with the largest gains on the most uneven segmentations, such as th

Load-bearing premise

The ranking of segmentation methods stands or falls with the validity of the two evaluation scores (InterpretTime and AUCD) as measures of true explanation quality; the paper itself observes that InterpretTime's required noise augmentation degrades non-deep-learning models and that its results can vary unpredictably across perturbations for those models.

Editorial extensions

If this is right

Practitioners using segment-based SHAP should default to equal-length segmentation; fancier change-point detectors only help when they produce materially different partitions, and they come at higher computational cost.
The length-weighting normalisation should become standard for segment-based explanations: it improves evaluation scores, especially for uneven segmentations, and leaves equal-length explanations essentially unchanged.
The zero background is an acceptable default, but when it differs from the average background, the average background typically gives better or equal explanation quality.
The choice of evaluation measure matters: InterpretTime and AUCD can disagree, particularly for non-deep-learning classifiers, so conclusions about explanation quality should be qualified by the evaluation method used.
The number of segments, not the choice of segmenter, is the primary lever on explanation quality; practitioners should set the segment count first and treat algorithm choice as secondary.

Reading between the lines

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

If the ranking holds under other fidelity metrics, the whole category of 'smart' segmentation for SHAP may be over-engineered; future work could redirect effort from change-point detectors to choosing segment counts and validating background sets.
The length-normalisation insight generalises beyond SHAP: any group-based attribution method that assigns the same weight to every member of a group should scale by group cardinality to preserve additivity, which is a testable extension for model-agnostic explainers on tabular data.
The observed disagreement between InterpretTime and AUCD for non-deep classifiers implies that comparative XAI papers in time series should report both metrics or re-validate with an evaluation method that does not retrain with noise augmentation.
A cheaper practical default could be quantified: the paper's data suggests a plateau in explanation quality at a relatively small number of equal-length segments, but it does not derive a formula for the minimal segment count.

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.

Desk Editor's Note

read the letter

Colleague, here's my read on arXiv:2509.03649. The genuinely useful part is the systematic comparison of eight segmentation strategies for segment-based SHAP on time series classification. Prior work mostly used equal-length segmentation; this paper gives a reasonably broad experimental matrix (five datasets, four classifiers, two backgrounds, two evaluation metrics) and shows that equal-length segmentation is usually as good as or better than bespoke change-point methods, at much lower cost. The length-based normalisation is a simple, sensible fix: distributing each segment's Shapley value uniformly over its time points preserves segment-level additivity. The result that this rescaling helps most when segment lengths are uneven is plausible and empirically demonstrated, especially on the EOG and KLC datasets. Credit where due: the code and data are released, the comparison is broad, and the practical guidance is actionable.

The soft spots are real, though. The headline claim that 'the number of segments has a greater impact than the segmentation method' is not actually supported by the experiments as described. No experiment varies segment count independently of method. Different methods produce different numbers of segments (ClaSP and NNSegment auto-detect change points; equal-length uses an arbitrary n), so observed method differences are confounded with segment-count differences. This is an internal-validity gap in the central claim. As far as I can see, the paper only shows that 'segmentation method' explains little variance overall, which is not the same thing. A controlled study where each method is run with matched segment counts (or where segment count is swept within a fixed method) would be needed.

Second, the evaluation metrics for non-deep-learning classifiers are fragile. InterpretTime requires noise-based training augmentation in its original form; the authors themselves state they omitted it because it degrades accuracy for QUANT, MiniRocket, and RandomForest, and note results 'may vary unpredictably' across perturbations. Since those classifiers are a large part of the experimental matrix, the ranking-based conclusions rest on a foundation the paper itself acknowledges is shaky. AUCD is more robust in this respect, but it's a different proxy, and the two metrics sometimes disagree.

The normalisation contribution is more solid: it's derived from an additivity property, tested externally via the two metrics, and shows a clear pattern. The background comparison is minor but fine.

Overall, this is a useful benchmark paper for the time series XAI subfield. It deserves a serious referee, but the referee should push for a direct segment-count experiment and a more careful handling of the evaluation metrics for non-DL classifiers. The equal-length recommendation may hold up, but the 'number of segments matters more' claim needs proper support.

Referee Report

4 major / 5 minor

Summary. The paper investigates how the choice of time-series segmentation affects the quality of SHAP-based explanations for time-series classifiers. It compares eight segmentation methods (equal-length, ClaSP, Information Gain, Greedy Gaussian, NNSegment, Binary Segmentation, Bottom-up, KernelCPD) on four classifiers and five datasets, with two SHAP background strategies and two XAI evaluation protocols (InterpretTime and AUCD). The abstract and conclusion make two headline claims: (1) the number of segments matters more than the specific segmentation method, and (2) equal-length segmentation consistently outperforms most custom segmentation algorithms. The paper also proposes a length-based normalisation of segment-level SHAP values to preserve segment-level additivity, and reports that this normalisation consistently improves attribution-quality scores.

Significance. If the claims are correct, the paper would provide actionable practical guidance for the growing area of segment-based SHAP for time series: users could adopt cheap equal-length segmentation and a simple normalisation rule to obtain equal or better explanations than with sophisticated segmentation algorithms. The paper is well-positioned in the literature and makes its code and data publicly available, which supports reproducibility. However, the headline contribution—the relative importance of segment count versus segmentation method—is not operationalised in the experimental design, and the evaluation metrics are acknowledged by the authors themselves to be unreliable for three of the four classifiers. The proposed normalisation is justified by a definitional additivity argument, but its empirical benefit is measured with the same fragile metrics. These issues are central to the paper's claims, so the work currently requires substantive revision.

major comments (4)

[Abstract, Section 5, Section 6] The central claim that 'the number of segments has a greater impact on explanation quality than the specific segmentation method' is not testable from the described experiments. Section 5's conditional variance analysis considers backgrounds, perturbations, and segmentation methods, but never varies the number of segments independently of the method. Section 4.2 lists hyperparameters for each segmentation, yet gives no indication that segment counts were matched across methods or varied within a method. Because each method produces its own segment count (e.g., ClaSP detects change points, equal-length uses an externally set n), the observed differences between 'methods' inevitably conflate method identity with segment count. The conclusion in Section 6 therefore goes beyond what the data can identify.
[Section 3, Section 5.2] The justification for the proposed normalisation is not sufficient. Section 3 asserts that preserving segment-level additivity is 'essential' and that the proposed uniform distribution 'ensures that evaluations performed on individual timepoints remain faithful to the original segment-wise Shapley values.' This is a definitional property, not an empirical or theoretical argument that additivity at the segment level is the correct criterion for timepoint-level explanation quality. The empirical claim that normalisation 'consistently improves attribution quality' is based on InterpretTime and AUCD scores, which rank individual timepoints; changing the scale by dividing by segment length directly changes cross-segment rankings for uneven segment lengths. The paper does not rule out that the improvement is an artefact of the metric rather than a genuinely better explanation. At minimum, the
[Section 5.3, Section 6] The paper's own limitation statement undermines the generality of the empirical results. It reports that the noise-based training augmentation required by InterpretTime degrades non-deep-learning classifiers, and that without it, results 'may vary unpredictably across different perturbations.' Since QUANT, MiniRocket, and RandomForest are used in a large part of the experimental matrix, the ranking-based conclusions for these classifiers rest on a fragile evaluation foundation. The paper concludes that InterpretTime's effectiveness 'may be limited to deep learning classifiers,' yet the headline recommendations about segmentation and normalisation are made across all classifiers rather than only for ResNet. This inconsistency needs to be resolved, for instance by restricting the claims to deep classifiers or by validating the metrics further.
[Section 4.2, Section 5] The claim that 'equal-length segmentation consistently outperforms most of the custom time series segmentation algorithms' is not supported by any direct statistical comparison. The equal-length segmentation's number of segments n is not specified in Section 4.2, so it is unclear whether the comparison is fair: a poorly chosen n could handicap equal-length, while the custom methods autodetect their segment counts. Section 5's variance analysis is not a substitute for a per-dataset, per-classifier pairwise comparison with a statistical test. Figure 1 shows error bars but no significance testing is reported anywhere in Section 5. The 'consistently' qualifier is too strong based on the evidence presented.

minor comments (5)

[Section 4.3] Typo: 'Metholodologies' should be 'Methodologies'.
[Section 3 / 5.2] The spelling alternates between 'normalisation' (Section 3) and 'normalization' (Section 5.2 and elsewhere). Please standardise.
[Section 2.2] The description of NNSegment says it 'comes from LimeSegment' but the original method is cited as [26]; consider stating the relationship more explicitly.
[Section 5] The phrase 'equal-length segmentation is often the best choice' is justified by computational cost rather than by explanation quality; this is an argument about efficiency, not about the quality claim, and should be separated.
[Figure 3] The caption mentions 'normalised entropy' but the body text uses 'relative entropy'; please align the terminology.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the paper's claims are empirical measurements against external XAI benchmarks, with self-citations serving as tooling rather than as derivation.

full rationale

The paper is an empirical study rather than a derivational chain. Its central claims—equal-length segmentation is best, normalization improves scores, background choice matters—are evaluated using InterpretTime (Turbé et al., external) and AUCD (Parvatharaju et al. / Petsiuk et al., external), not by construction from the paper's own definitions. The proposed normalization is defined as dividing each segment's attribution by its length and is then tested against those external metrics; the paper does not fit a parameter to the evaluation outcome and then call it a prediction. The self-citations to [25] provide the SHAP sampling implementation, the equal-length segmentation implementation, and the extended InterpretTime routine. These are tooling dependencies from the authors' prior published, code-released work; they do not incorporate the present paper's conclusions, and the prior work's claims (e.g., that noise injection degrades non-deep classifiers) are externally falsifiable empirical statements rather than imported theorems used to force a result. The conclusion that 'the number of segments is more important than the segmentation method' is not circular, though it is under-supported because segment count is not independently varied in the experiments; that is an internal-validity concern, not a circularity. Likewise, the assertion that preserving segment-level additivity is 'essential' is a stated premise used to motivate normalization, not a result derived from the evaluation itself. No equation in the paper reduces to its own inputs, and no fitted quantity is renamed as a prediction. Therefore no circular step can be exhibited.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new unobserved entities; its contributions are a comparison and a normalisation rule. The main uncharged assumptions are the validity of the evaluation metrics and the uniform attribution within segments.

free parameters (3)

Number of segments (change points) for each segmentation method = not reported
The central comparison between segmentation methods is confounded with the number of segments produced by each algorithm; no experiment varying segment count is described.
ClaSP period length = 4
Set to 4 because the default produced only one segment; this data-dependent choice affects which segmentations ClaSP produces.
AUCD step size = 4% of series
Hand-chosen perturbation step; affects the AUCD curves and therefore the ranking of methods.

assumptions (4)

domain assumption Segment-level SHAP values are valid explanations for the time series classifier
Sections 1 and 4 treat a coarsened feature set, segments, as the SHAP feature space, which requires that grouping consecutive time points preserves explanatory meaning.
domain assumption InterpretTime and AUCD are reliable proxies for explanation quality
Sections 2.3 and 5.3: all quality claims are defined by these metrics, and the paper itself concludes InterpretTime may be unreliable for non-deep-learning classifiers.
ad hoc to paper Preserving segment-level additivity is necessary for sound explanations
Section 3 asserts this without external support; it motivates the proposed normalisation and underlies the claimed improvement.
ad hoc to paper All time points within a segment contribute equally to the segment attribution
Section 3 distributes phi_S divided by segment length uniformly across constituent time points; no evidence is given that points within a segment are equally important.

how reviews work

0 comments

Cite this review

Pith. "Pith review of An Empirical Evaluation of Factors Affecting SHAP Explanation of Time Series Classification." pith.science (2026). https://pith.science/paper/73AA5CM3

@misc{pith2026250903649,
  author       = {Pith},
  title        = {Pith review of: An Empirical Evaluation of Factors Affecting SHAP Explanation of Time Series Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/73AA5CM3}},
  note         = {Machine review of arXiv:2509.03649}
}

read the original abstract

Explainable AI (XAI) has become an increasingly important topic for understanding and attributing the predictions made by complex Time Series Classification (TSC) models. Among attribution methods, SHapley Additive exPlanations (SHAP) is widely regarded as an excellent attribution method; but its computational complexity, which scales exponentially with the number of features, limits its practicality for long time series. To address this, recent studies have shown that aggregating features via segmentation, to compute a single attribution value for a group of consecutive time points, drastically reduces SHAP running time. However, the choice of the optimal segmentation strategy remains an open question. In this work, we investigated eight different Time Series Segmentation algorithms to understand how segment compositions affect the explanation quality. We evaluate these approaches using two established XAI evaluation methodologies: InterpretTime and AUC Difference. Through experiments on both Multivariate (MTS) and Univariate Time Series (UTS), we find that the number of segments has a greater impact on explanation quality than the specific segmentation method. Notably, equal-length segmentation consistently outperforms most of the custom time series segmentation algorithms. Furthermore, we introduce a novel attribution normalisation technique that weights segments by their length and we show that it consistently improves attribution quality.

Figures

Figures reproduced from arXiv: 2509.03649 by the authors.

**Figure 1.** Aggregated AUCSE by dataset and classifier. Error bars are standard deviation from the mean. AUCSE is computed based on the SHAP attributions for each classifier-dataset pair. combination as shown in [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗

**Figure 2.** F-score on ResNet and KeplerLightCurves by Background and Perturba [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗

**Figure 3.** Mean normalised entropy by Dataset and Segmentation. Error bar de [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: AUCD for MiniRocket on Gunpoint and MilitaryPress. Both cases show [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 5.** Figure 5: Difference in AUCSE results when normalization is applied. A positive [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

**Figure 6.** Figure 6: InterpretTime’s scores for Gunpoint MiniRocket and for MP MiniRocket. [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]

**Figure 7.** Figure 7: Accuracy decay and probability of the initial predicted label evolutions [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]

**Figure 8.** Figure 8: InterpretTime’s and AUCD scores for (a) and (b) ResNet applied to KLC, [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 25 canonical work pages

[1]

Monthly Notices of the Royal Astronomical Society 514(2), 2793–2804 (2022)

Barbara, et al.: Classifying kepler light curves for 12 000 a and f stars using super- vised feature-based machine learning. Monthly Notices of the Royal Astronomical Society 514(2), 2793–2804 (2022)

work page 2022
[2]

In: Proceedings of the 2022 International Conference on Management of Data

Boniol, et al.: dcam: Dimension-wise class activation map for explaining multivari- ate data series classification. In: Proceedings of the 2022 International Conference on Management of Data. pp. 1175–1189 (2022)

work page 2022
[3]

Computational Statistics & Data Analysis128, 200–220 (2018)

Celisse, et al.: New efficient algorithms for multiple change-point detection with re- producing kernels. Computational Statistics & Data Analysis128, 200–220 (2018)

work page 2018
[4]

In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge D

Dempster, et al.: Minirocket: A very fast (almost) deterministic transform for time series classification. In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge D. & Data Mining. Association for Computing Machinery (2021)

work page 2021
[5]

Data Mining and Knowledge Discovery38(4), 2377–2402 (2024)

Dempster, et al.: Quant: A minimalist interval method for time series classification. Data Mining and Knowledge Discovery38(4), 2377–2402 (2024)

work page 2024
[6]

In: Advanced Analytics and Learning on Temporal Data

Dhariyal, et al.: Back to basics: A sanity check on modern time series classification algorithms. In: Advanced Analytics and Learning on Temporal Data. pp. 205–229. Springer Nature Switzerland, Cham (2023) 16 D. Serramazza, N. Papadeas et al

work page 2023
[7]

Data Mining and Knowledge D.37(2), 1010–1054 (2023)

Dhariyal, et al.: Scalable classifier-agnostic channel selection for multivariate time series classification. Data Mining and Knowledge D.37(2), 1010–1054 (2023)

work page 2023
[8]

Data Mining and Knowledge Discovery 37(3), 1262–1300 (2023)

Ermshaus, et al.: Clasp: parameter-free time series segmentation. Data Mining and Knowledge Discovery 37(3), 1262–1300 (2023)

work page 2023

Show all 28 references

[9]

PloS one13(2), e0192684 (2018)

Fang, et al.: Electrooculography-based continuous eye-writing recognition system for efficient assistive communication systems. PloS one13(2), e0192684 (2018)

2018
[10]

Geurts, et al.: Extremely randomized trees. Mach. Learn.63(1), 3–42 (Apr 2006)

2006
[11]

Advances in Data Analysis and Classification13(3), 727–751 (2019)

Hallac, et al.: Greedy gaussian segmentation of multivariate time series. Advances in Data Analysis and Classification13(3), 727–751 (2019)

2019
[12]

In: Proceedings of the IEEE conference on computer vision and pattern recognition

He, et al.: Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 770–778 (2016)

2016
[13]

Journal of the American Statistical Association107(500), 1590–1598 (2012)

Killick, et al.: Optimal detection of changepoints with a linear computational cost. Journal of the American Statistical Association107(500), 1590–1598 (2012)

2012
[14]

arXiv preprint arXiv:2009.07896 (2020)

Kokhlikyan, et al.: Captum: A unified and generic model interpretability library for pytorch. arXiv preprint arXiv:2009.07896 (2020)

2009 arXiv
[15]

Pervasive and Mobile Computing5(6), 657–675 (2009)

Liu, et al.: uwave: Accelerometer-based personalized gesture recognition and its applications. Pervasive and Mobile Computing5(6), 657–675 (2009)

2009
[16]

arXiv preprint arXiv:1909.07872 (2019)

Löning, M., Bagnall, A., Ganesh, S., Kazakov, V., Lines, J., Király, F.J.: sk- time: A unified interface for machine learning with time series. arXiv preprint arXiv:1909.07872 (2019)

1909 arXiv
[17]

Advances in neural information processing systems30 (2017)

Lundberg, et al.: A unified approach to interpreting model predictions. Advances in neural information processing systems30 (2017)

2017
[18]

Journal of Machine Learning Research25(289), 1–10 (2024)

Middlehurst, M., et al.: aeon: a python toolkit for learning from time series. Journal of Machine Learning Research25(289), 1–10 (2024)

2024
[19]

In: Proceedings of the 30th ACM International Conference on Information & Knowledge Management

Parvatharaju, P.S., Doddaiah, R., Hartvigsen, T., Rundensteiner, E.A.: Learn- ing saliency maps to explain deep time series classifiers. In: Proceedings of the 30th ACM International Conference on Information & Knowledge Management. p. 1406–1415. CIKM ’21, Association for Comp...

2021
[20]

Journal of Machine Learning Research 12, 2825–2830 (2011)

Pedregosa, et al.: Scikit-learn: Machine learning in Python. Journal of Machine Learning Research 12, 2825–2830 (2011)

2011
[21]

arXiv preprint arXiv:1806.07421 (2018)

Petsiuk, V., Das, A., Saenko, K.: Rise: Randomized input sampling for explanation of black-box models. arXiv preprint arXiv:1806.07421 (2018)

2018 arXiv
[22]

In: Proceedings of the 2005 SIAM international conference on data mining

Ratanamahatana, et al.: Three myths about dynamic time warping data mining. In: Proceedings of the 2005 SIAM international conference on data mining. pp. 506–510. SIAM (2005)

2005
[23]

Pervasive and Mobile Computing38, 92–109 (2017)

Sadri, et al.: Information gain-based metric for recognizing transitions in human activities. Pervasive and Mobile Computing38, 92–109 (2017)

2017
[24]

The Annals of statistics pp

Sen, et al.: On tests for detecting change in mean. The Annals of statistics pp. 98–108 (1975)

1975
[25]

In: Machine Learning and Knowledge Discovery in Databases

Serramazza, et al.: Improving the evaluation and actionability of explanation meth- ods for multivariate time series classification. In: Machine Learning and Knowledge Discovery in Databases. Research Track. Cham (2024)

2024
[26]

In: Camps-Valls, G., Ruiz, F.J.R., Valera, I

Sivill, et al.: Limesegment: Meaningful, realistic time series explanations. In: Camps-Valls, G., Ruiz, F.J.R., Valera, I. (eds.) Proceedings of The 25th Interna- tional Conference on Artificial Intelligence and Statistics. Proceedings of Machine Learning Research, vol. 151, p...

2022
[27]

Signal Processing 167, 107299 (2020)

Truong, et al.: Selective review of offline change point detection methods. Signal Processing 167, 107299 (2020)

2020
[28]

Nature Machine Intelligence5(3), 250–260 (2023)

Turbé, et al.: Evaluation of post-hoc interpretability methods in time-series classi- fication. Nature Machine Intelligence5(3), 250–260 (2023)

2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Monthly Notices of the Royal Astronomical Society 514(2), 2793–2804 (2022)

Barbara, et al.: Classifying kepler light curves for 12 000 a and f stars using super- vised feature-based machine learning. Monthly Notices of the Royal Astronomical Society 514(2), 2793–2804 (2022)

work page 2022

[2] [2]

In: Proceedings of the 2022 International Conference on Management of Data

Boniol, et al.: dcam: Dimension-wise class activation map for explaining multivari- ate data series classification. In: Proceedings of the 2022 International Conference on Management of Data. pp. 1175–1189 (2022)

work page 2022

[3] [3]

Computational Statistics & Data Analysis128, 200–220 (2018)

Celisse, et al.: New efficient algorithms for multiple change-point detection with re- producing kernels. Computational Statistics & Data Analysis128, 200–220 (2018)

work page 2018

[4] [4]

In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge D

Dempster, et al.: Minirocket: A very fast (almost) deterministic transform for time series classification. In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge D. & Data Mining. Association for Computing Machinery (2021)

work page 2021

[5] [5]

Data Mining and Knowledge Discovery38(4), 2377–2402 (2024)

Dempster, et al.: Quant: A minimalist interval method for time series classification. Data Mining and Knowledge Discovery38(4), 2377–2402 (2024)

work page 2024

[6] [6]

In: Advanced Analytics and Learning on Temporal Data

Dhariyal, et al.: Back to basics: A sanity check on modern time series classification algorithms. In: Advanced Analytics and Learning on Temporal Data. pp. 205–229. Springer Nature Switzerland, Cham (2023) 16 D. Serramazza, N. Papadeas et al

work page 2023

[7] [7]

Data Mining and Knowledge D.37(2), 1010–1054 (2023)

Dhariyal, et al.: Scalable classifier-agnostic channel selection for multivariate time series classification. Data Mining and Knowledge D.37(2), 1010–1054 (2023)

work page 2023

[8] [8]

Data Mining and Knowledge Discovery 37(3), 1262–1300 (2023)

Ermshaus, et al.: Clasp: parameter-free time series segmentation. Data Mining and Knowledge Discovery 37(3), 1262–1300 (2023)

work page 2023