Multi-SpaCE: Multi-Objective Subsequence-based Sparse Counterfactual Explanations for Multivariate Time Series Classification

David Luengo; Mario Refoyo

REVIEW 4 major objections 6 minor 37 references

Multi-SpaCE: Multi-Objective Subsequence-based Sparse Counterfactual Explanations for Multivariate Time Series Classification

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Multi-SpaCE claims to be the first multi-objective counterfactual method that guarantees perfect validity for multivariate time-series classification.

desk verdict A solid, reproducible multivariate extension of Sub-SpaCE that achieves validity 1.0 on every tested dataset, but the 'ensures perfect validity' claim is a penalty-induced ordering claim, not a guarantee. read the letter →

arxiv 2501.04009 v2 pith:3JXM754C submitted 2024-12-14 cs.NE cs.LGstat.ML

Mario Refoyo , David Luengo This is my paper · ORCID

classification cs.NEcs.LGstat.ML

keywords eXplainableArtificialIntelligence(XAI)CounterfactualExplanationsGeneticAlgorithmOptimizationTimeSeriesClassificationMultivariateSparsityValidityParetofront

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

Multi-SpaCE aims to solve a practical problem in explainable AI: when a time-series classifier makes a decision, how can we produce a "what-if" version of the input that reliably changes the decision while keeping the change small, interpretable, and realistic? The paper's central claim is that by treating validity—the requirement that the counterfactual actually flips the predicted class—as a strict, uniformly penalized constraint inside a multi-objective evolutionary search, it is possible to guarantee perfect validity for multivariate time series, not just for univariate ones. The method returns a whole Pareto front of alternative counterfactuals, so different users can choose among sparser or more plausible explanations without re-tuning weights. If true, this removes the need to discard invalid explanations after the fact and makes counterfactual explanations usable in high-stakes settings such as healthcare or finance.

What carries the argument

The central object is the binary mask $M \in \{0,1\}^{L \times C}$ (a shared mask $m \in \{0,1\}^L$ is used first) that marks which time-channel cells are replaced by the corresponding values of the nearest unlike neighbor. The search engine is NSGA-II, a non-dominated sorting genetic algorithm, augmented with subsequence-level mutation operators that extend, compress, or prune whole contiguous blocks of changes. The load-bearing idea is that each of the four objectives in Eq.\ (5) carries the same strict penalty $\nu \cdot \mathbb{1}_{\text{class}}$ with $\nu = 100$; because any candidate with the wrong output class is dominated by every candidate with the correct class, the final non-dominated front can only contain valid counterfactuals. Plausibility is scored by an autoencoder's reconstruction-error outlier score, and the method combines a common-mask phase with an independent per-channel-mask phase to balance search difficulty and flexibility.

What would settle it

For a single test instance, enumerate all possible binary masks (or run Multi-SpaCE many times with different random seeds) and check whether any mask's nearest-unlike-neighbor substitution changes the classifier's output. If a single instance admits no valid mask, the method cannot return a valid counterfactual for it regardless of the penalty; finding such an instance among the test set would disprove the perfect-validity claim. Similarly, measuring validity on a dataset with many misclassified instances would expose whether the 100 percent scores reflect the algorithm or the easy datasets.

Watch

Extended reading notes

Core claim

The paper claims that Multi-SpaCE is the first multi-objective optimization method that ensures perfectly valid counterfactual explanations for multivariate time-series classification. A counterfactual is built by choosing a binary mask of positions and copying the values of the nearest unlike neighbor (the closest training instance from a different class) at those positions; the search over masks is done with NSGA-II over four objectives: adversarial probability for the desired class, sparsity, contiguity of subsequences, and plausibility measured by an autoencoder. In every objective the same large penalty term penalizes candidates that do not change the predicted class, so invalid candidates are dominated by valid ones and drop out of the Pareto front. On the paper's experiments, Multi-SpaCE attains 100 percent validity on every tested dataset, is the only method to do so in the multivariate benchmarks, and is competitive or best on proximity, sparsity, plausibility, and the joint sparsity-contiguity measure.

Load-bearing premise

The method assumes that for every instance explained there exists some binary mask whose nearest-unlike-neighbor substitution flips the classifier's output to the desired class; the strict penalty can only prefer valid masks over invalid ones, it cannot produce a valid mask where none exists.

Editorial extensions

If this is right

Users of Multi-SpaCE can take any counterfactual from the returned Pareto front and know it changes the classifier's predicted class, without post-hoc filtering.
Different stakeholders can choose among explanations that favor sparsity, contiguity, or plausibility, because the Pareto front preserves those trade-offs instead of collapsing them into one tuned solution.
The method is model-agnostic: it needs only the classifier's class probabilities, so it applies to any multivariate time-series classifier that provides them.
On the reported benchmarks, Multi-SpaCE matches the perfect validity of the single-subsequence univariate method it extends and becomes the only tested method with perfect validity in the multivariate datasets.
The method ranks first or near-first on proximity, sparsity, and the joint sparsity-contiguity metric, with execution times in the middle of the baseline spectrum.

Reading between the lines

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

Beyond the paper: the perfect-validity guarantee is conditional on the existence of a valid mask for each instance; if the classifier already misclassifies the nearest unlike neighbor or no subsequence combination flips the output, the penalty term can only rank candidates, not create a valid one.
Beyond the paper: a natural test is to apply Multi-SpaCE to harder settings—lower classifier accuracy, noisy labels, or out-of-distribution test points—and check whether validity stays at 100 percent.
Beyond the paper: replacing nearest-neighbor copying with generative substitutes (the paper's own future direction) would change the nature of the guarantee, since generated values must also stay on the data manifold for the counterfactual to remain plausible.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The paper is worth a referee's time, but the headline claim is wrong as stated. Multi-SpaCE empirically achieves validity 1.0 on every dataset it tests, and that is a real result. But 'ensures perfect validity' (abstract, Section 1, Section 5) is not supported: the strict penalty in Eq. (5) only orders candidates within the NSGA-II population. It cannot manufacture a valid counterfactual if no mask on the single NUN flips the classifier. That existential assumption is never stated or checked. The reinitialization schedule drives h toward 100%, so the strongest possible candidate is full substitution x' = xnun; if InceptionTime misclassifies the NUN, perfect validity is unattainable no matter how large nu is. What the evidence actually supports is 'achieves perfect validity on all evaluated instances,' which is still competitive with or better than the baselines. What is genuinely new: the multivariate extension (common-then-independent mask schedule), the subsequence pruning mutation, and the NSGA-II formulation with four objectives over subsequence masks. The paper is also unusually honest in places: it admits proximity is not directly optimized (Section 4.5), and it discloses the parallelization overhead in Appendix D. Code and data are public, the baselines use the original implementations, the dataset filter (F1 > 80%) is sensible, and the ablation study in Appendix A is thorough. Soft spots, in order of severity. First, the perfect-validity claim is load-bearing; it appears in the abstract, intro, and conclusions, and it is the stated novelty over TSEvo and TX-Gen. Rewording is mandatory, not cosmetic. Second, the abstract says the method balances proximity, but Eq. (5) has no proximity objective; the method optimizes adversarial probability, sparsity, contiguity, and plausibility. Minor fix. Third, hyperparameters (mask schedule, mutation probabilities, pruning) were selected in Appendix A on the same ten multivariate datasets used for the reported benchmarks, and the final configuration was confirmed on that same data. That is an in-sample selection loop, and the paper should disclose it in the main text rather than bury it in an appendix. Fourth, the plausibility objective uses the autoencoder, and one of the three plausibility evaluation metrics is that same autoencoder's outlier score; the IF and LOF evaluations mitigate this only partially. Who this is for: researchers working on time-series XAI, especially anyone who needs valid counterfactuals in multivariate settings. The comparison against original implementations of COMTE, AB-CF, DiscoX, NG, and Glacier is useful and fair. Recommendation: send to peer review. The method is solid and reproducible; the claims need tempering and the existence assumption needs a stated limitation.

Referee Report

4 major / 6 minor

Summary. Multi-SpaCE is a counterfactual explanation method for multivariate time series classification. It uses NSGA-II to optimize several objectives: adversarial class probability, sparsity, contiguity, and autoencoder-based plausibility, while searching over binary masks that substitute values from the nearest unlike neighbor (NUN). The method returns a Pareto front of candidate counterfactuals rather than a single solution. The paper reports experiments on UCR/UEA datasets, comparing against NG, Glacier, GlacierAE, AB-CF, DiscoX, and COMTE, and claims that Multi-SpaCE is the first multi-objective method that ensures perfectly valid counterfactuals for multivariate time series.

Significance. If the central claim were fully supported, Multi-SpaCE would fill a clear gap: existing multivariate counterfactual methods for time series either enforce rigid subsequence structures, treat validity as a soft objective, or lack multivariate support. The paper's strengths include publicly available code, a broad benchmark across 25 datasets, comparison with several recent baselines, and an ablation study of mutation and mask design choices. The inclusion of Isolation Forest and Local Outlier Factor as independent plausibility evaluators is a good safeguard. However, the perfect-validity guarantee is overstated relative to what a penalty-based ranking can deliver, and several evaluation choices create in-sample optimism. The empirical results are still useful, but the headline formal claim and the interpretation of some reported metrics need substantial revision.

major comments (4)

[Section 3.2.3, Eq. (5), Algorithm 1] The claim that the penalty term -nu * 1class 'ensures that non-valid solutions are always dominated by valid counterfactuals' only imposes an ordering on the population. It cannot create a valid solution if none exists in the search space. Since the counterfactual is always a subsequence-wise copy of a single NUN, validity is possible only if some mask makes b(x') = y_nun. The reinitialization schedule in Algorithm 1 approaches the full substitution x' = x_nun, but if the classifier misclassifies x_nun (not excluded by the F1 > 80% dataset filter), even this extreme candidate can be invalid, and no value of nu repairs the absence. Thus 'ensures perfect validity' in Sections 1 and 5 is unsupported as a formal property; the defensible statement is that Multi-SpaCE achieved perfect validity on all evaluated instances. Please either add an explicit existence assumption and prove the guarantee under it, or soften the claim throughout.
[Abstract and Section 1 vs. Eq. (5)] The abstract and introduction state that Multi-SpaCE 'balances proximity, sparsity, plausibility, and contiguity', but Eq. (5) contains no proximity objective: o1 is class probability, o2 sparsity, o3 contiguity, and o4 AE-based plausibility. Proximity appears only as an evaluation metric in Section 4.4 and Table 3. Sparsity of NUN substitutions is a rough proxy for proximity but is not identical to the reported l2 distance. Please either add an explicit proximity objective to Eq. (5) or revise the claims to say that proximity is evaluated but not directly optimized.
[Appendix A and Section 4.5] The hyperparameters of Multi-SpaCE are selected using the same multivariate datasets on which the final results are reported. Tables A1-A4 and Figures A1-A2 use A WR, BasicMotions, Cricket, Epilepsy, NATOPS, RacketSports, SR-SCP1, and UWave to choose pe, pc, pp, and the common/independent mask schedule; Section 4.5 then reports the performance of the selected configuration on those same datasets. This creates an in-sample selection loop that can inflate the reported average ranks. Please either use a separate validation split for the ablation, or clearly disclose the selection loop and quantify its potential bias, for example by comparing the selected configuration with a default configuration on held-out datasets.
[Eq. (5) and Tables 4-5] The plausibility objective o4(x, x') = -e(x,x')/e_max is computed with the autoencoder f_AE, and the evaluation metric OS(AE) in Tables 4 and 5 uses the same f_AE. Consequently, the OS(AE) rankings partly measure how well the method optimizes the evaluation quantity itself. The OS(IF) and OS(LOF) columns provide independent evidence and partly mitigate this concern, but the text should state this circularity explicitly and give the OS(IF)/OS(LOF) results at least equal weight when drawing plausibility conclusions.

minor comments (6)

[Section 3.2.1] The word 'beggining' should be 'beginning'.
[Section 1] The word 'obtimization' should be 'optimization'.
[Equations (2)-(4)] The term 'Bernouilli' should be 'Bernoulli' in the descriptions of delta_e, delta_c, and delta_p.
[Algorithm 2, line 14] The typeset 'F itnessn' should be 'Fitness_n'.
[Section 3.2.3] The sentence 'Every objective is an adaptation of the terms in (5)' should refer to Eq. (1) rather than Eq. (5), since Eq. (5) itself defines the objectives being adapted.
[Appendix B, Table B6] The dataset 'Phoneme' appears in Table B6 but is not listed in Table 1 or described in Section 4.2; either add it to the dataset selection or remove the row.

Circularity Check

2 steps flagged · score 4.0 of 10

The 'perfect validity' guarantee is an unsupported existence claim rather than a circular derivation, but two evaluation loops are circular: AE-based plausibility is optimized and then reported with the same autoencoder, and final hyperparameters were tuned on datasets that later appear in the main results.

self definitional [Section 3.1 (Plausibility Loss), Section 3.2.3 Eq. (5), Section 4.4 (Plausibility metric)]
"Plausibility Loss: measured as the Increase in Outlier Score Loss Lios = e(x,x′)/emax ... Using an autoencoder fAE, trained to reconstruct the data, this term measures how far x′ is from the data distribution compared to the original instance x. ... o4(x, x′) = − e(x, x′)/emax − ν · 1class(x′, M, ynun) ... Plausibility: measured by the Outlier Score (OS) of the generated counterfactuals. As introduced in Section 4.1, we use an Autoencoder, an Isolation Forest and a Local Outlier Factor."

The o4 objective in Eq. (5) is defined directly from the same autoencoder reconstruction error e(x,x′) that underlies the reported AE Outlier Score. Multi-SpaCE explicitly maximizes −e/emax during the search, so the OS(AE) plausibility values reported in Tables 4 and 5 are evaluations of the very quantity being optimized, not an independent check. The AE-based plausibility ranking is therefore forced by construction; only the separate IF and LOF evaluations provide independent evidence.
fitted input called prediction [Section 4.1 (Setup), Appendix A.1-A.3 (ablations), Section 4.5 (Results)]
"The parameter values of Multi-SpaCE were empirically set and kept consistent across datasets to ensure robustness in multiple scenarios. Instead of extensive fine-tuning of all possible parameters, we focused on evaluating the influence of mutation parameters and the different types of masks proposed in Section 3.2.1. Based on these evaluations (see A), the final Multi-SpaCE uses a population size N = 100 ... Based on these results, a mutation probability of 0.75 was selected as optimal for the common mask setting, while a probability of 0.5 was chosen for the independent mask setting."

Appendix A selects the final mask schedule and mutation probabilities by comparing configurations on the same UCR and UEA datasets (AWR, BasicMotions, Cricket, Epilepsy, NATOPS, RacketSports, SR-SCP1, UWave) that later appear in the main result tables in Section 4.5. The reported proximity, plausibility, and sparsity results for those datasets are thus in-sample: the configuration was chosen because it scored well on these very datasets, so the benchmark comparison is not an out-of-sample prediction. The validity rows are less affected because all ablated configurations already reached validity 1.0.

full rationale

The paper contains no formal derivation, so most of its argument is empirical rather than definitional. I find no load-bearing self-citation: the references to Sub-SpaCE [25] describe the predecessor method and do not substitute for the penalty argument in Eq. (5). The central claim that Multi-SpaCE 'ensures perfect validity' is not itself circular: the penalty term −ν·1class only guarantees that valid solutions dominate invalid ones, and it cannot create a valid mask when none exists, notably when the single nearest unlike neighbor is itself misclassified. That is an unsupported existence assumption, not an equation-level reduction, so it is a correctness risk rather than a circularity. The concrete circularities are in the evaluation loop: o4 in Eq. (5) uses the same autoencoder reconstruction error that is later reported as the AE Outlier Score, so the AE-based plausibility results are partly constructed rather than measured; and Appendix A selects the final mutation parameters on datasets that subsequently appear in the main result tables, making those comparisons in-sample. The independent IF and LOF plausibility evaluations, the external baselines, and the fact that all ablated configurations already reach validity 1.0 keep the core validity claim partially grounded. Overall score 4: partial, localized circularity in supporting evaluations, with the central validity claim still resting on independent empirical evidence.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The method depends on a large set of hand-set hyperparameters (penalty, mutation probabilities, generation counts, initialization schedule) and on domain assumptions about NUN substitution being an effective way to create counterfactuals and the autoencoder being a valid plausibility measure. No new physical entities are introduced.

free parameters (8)

validity penalty nu = 100
Large constant added to every objective to enforce validity; chosen by hand, not derived.
extension/compression probabilities pe, pc = 0.75 common mask; 0.5 independent mask
Selected in Appendix A ablation on the same benchmark datasets used for final results.
pruning probability pp = 0.75 in final stage
Selected in Appendix A.3 ablation on the same datasets.
generations G1, G2 = 75, 25
Set empirically; the final pruning stage design was chosen from ablation results.
initialization schedule h, hinc, Greinit = 20%, 20%, 50
Reinitialization schedule chosen by hand; affects convergence and empirical validity.
population size N = 100
Empirically set; no sensitivity analysis reported.
utility weights for Pareto front selection = 0.1, 0.3, 0.4, 0.2
Taken from Sub-SpaCE to pick one Multi-SpaCE solution for comparison; these weights determine the reported metrics.
gamma in contiguity objective = 0.25
Inherited from Sub-SpaCE; controls the penalty curvature for the number of subsequences.

assumptions (5)

domain assumption There exists a binary mask m such that replacing selected values with NUN values changes the classifier output to ynun.
Without such a mask, no valid counterfactual exists in the search space; the paper does not verify existence for each dataset/instance. Invoked in Sections 3.1 and 3.2.3.
domain assumption Autoencoder reconstruction error is a reliable proxy for plausibility and outlier score.
The plausibility objective o4 and the evaluation metric OS(AE) both use reconstruction error; if the autoencoder is a poor density model, the plausibility objective is mis-specified. Sections 3.1 and 4.4.
domain assumption NSGA-II with penalty and reinitialization converges to valid solutions for all instances.
NSGA-II is a stochastic metaheuristic with no constraint-satisfaction guarantee; 'perfect validity' is empirical, not formal. Section 3.2.3 and Algorithm 1.
standard math Euclidean nearest unlike neighbor provides an appropriate substitution source.
The method uses Euclidean distance for the NUN and assumes it captures relevant time series similarity. Section 3.1.
domain assumption F1-score above 80% for the classifier ensures meaningful counterfactual explanations.
Datasets were selected only where InceptionTime exceeded 80% F1; results may not transfer to weaker classifiers. Section 4.2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Multi-SpaCE: Multi-Objective Subsequence-based Sparse Counterfactual Explanations for Multivariate Time Series Classification." pith.science (2026). https://pith.science/paper/3JXM754C

@misc{pith2026250104009,
  author       = {Pith},
  title        = {Pith review of: Multi-SpaCE: Multi-Objective Subsequence-based Sparse Counterfactual Explanations for Multivariate Time Series Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3JXM754C}},
  note         = {Machine review of arXiv:2501.04009}
}

read the original abstract

Deep Learning systems excel in complex tasks but often lack transparency, limiting their use in critical applications. Counterfactual explanations, a core tool within eXplainable Artificial Intelligence (XAI), offer insights into model decisions by identifying minimal changes to an input to alter its predicted outcome. However, existing methods for time series data are limited by univariate assumptions, rigid constraints on modifications, or lack of validity guarantees. This paper introduces Multi-SpaCE, a multi-objective counterfactual explanation method for multivariate time series. Using non-dominated ranking genetic algorithm II (NSGA-II), Multi-SpaCE balances proximity, sparsity, plausibility, and contiguity. Unlike most methods, it ensures perfect validity, supports multivariate data and provides a Pareto front of solutions, enabling flexibility to different end-user needs. Comprehensive experiments in diverse datasets demonstrate the ability of Multi-SpaCE to consistently achieve perfect validity and deliver superior performance compared to existing methods.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 35 canonical work pages

[1]

SN Computer Science 3, 158–178 (2022)

Sarker, I.: AI-Based Modeling: Techniques, Applications and Research Issues Towards Automation, Intelligent and Smart Systems. SN Computer Science 3, 158–178 (2022)

work page 2022
[2]

Information Fusion 58, 82–115 (2020)

Barredo Arrieta, A., D ˜Aaz-Rodr ˜Aguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., Chatila, R., Herrera, F.: Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion 58, 82–115 (2020)

work page 2020
[3]

https://christophm.github

Molnar, C.: Interpretable Machine Learning, (2022). https://christophm.github. io/interpretable-ml-book

work page 2022
[4]

Harvard Journal of Law & Technology 31, 841–888 (2017)

Wachter, S., Mittelstadt, B., Russell, C.: Counterfactual explanations without opening the black box: Automated decisions and the GDPR. Harvard Journal of Law & Technology 31, 841–888 (2017)

work page 2017
[5]

In: Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19, pp

Byrne, R.M.J.: Counterfactuals in Explainable Artificial Intelligence (XAI): Evi- dence from human reasoning. In: Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19, pp. 6276–6282 (2019)

work page 2019
[6]

Artificial Intelligence 267, 1–38 (2019) 20

Miller, T.: Explanation in Artificial Intelligence: Insights from the social sciences. Artificial Intelligence 267, 1–38 (2019) 20

work page 2019
[7]

Data Mining and Knowledge Discovery, 2770–2824 (2022)

Guidotti, R.: Counterfactual explanations and how to find them: literature review and benchmarking. Data Mining and Knowledge Discovery, 2770–2824 (2022)

work page 2022
[8]

ACM Computing Surveys 56(12) (2024)

Verma, S., Boonsanong, V., Hoang, M., Hines, K., Dickerson, J., Shah, C.: Coun- terfactual explanations and algorithmic recourses for machine learning: A review. ACM Computing Surveys 56(12) (2024)

work page 2024

Show all 37 references

[9]

Advances in neural information processing systems 31, 590–601 (2018)

Dhurandhar, A., Chen, P.-Y., Luss, R., Tu, C.-C., Ting, P., Shanmugam, K., Das, P.: Explanations based on the missing: Towards contrastive explanations with pertinent negatives. Advances in neural information processing systems 31, 590–601 (2018)

2018
[10]

In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pp

Van Looveren, A., Klaise, J.: Interpretable counterfactual explanations guided by prototypes. In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pp. 650–665 (2021). Springer

2021
[11]

In: Artificial Neural Networks and Machine Learning (ICANN), pp

Lang, J., Giese, M.A., Ilg, W., Otte, S.: Generating sparse counterfactual expla- nations for multivariate time series. In: Artificial Neural Networks and Machine Learning (ICANN), pp. 180–193 (2023)

2023
[12]

In: 2020 Conference on Fairness, Accountability, and Transparency

Mothilal, R.K., Sharma, A., Tan, C.: Explaining machine learning classifiers through diverse counterfactual explanations. In: 2020 Conference on Fairness, Accountability, and Transparency. F AT* ’20, pp. 607–617 (2020)

2020
[13]

In: 34th International Conference on Neural Information Processing Systems

Karimi, A.-H., K¨ ugelgen, J., Sch¨ olkopf, B., Valera, I.: Algorithmic recourse under imperfect causal knowledge: A probabilistic approach. In: 34th International Conference on Neural Information Processing Systems. NIPS’20, pp. 265–277 (2020)

2020
[14]

In: International Conference on Parallel Problem Solving From Nature, pp

Dandl, S., Molnar, C., Binder, M., Bischl, B.: Multi-Objective Counterfactual Explanations. In: International Conference on Parallel Problem Solving From Nature, pp. 448–469 (2020). Springer

2020
[15]

In: International Joint Conference on Neural Networks (IJCNN), pp

Barredo-Arrieta, A., Del Ser, J.: Plausible counterfactuals: Auditing deep learning classifiers with realistic adversarial examples. In: International Joint Conference on Neural Networks (IJCNN), pp. 1–7 (2020)

2020
[16]

In: 21st IEEE International Conference on Machine Learning and Applications (ICMLA), pp

H¨ ollig, J., Kulbach, C., Thoma, S.: TSEvo: Evolutionary Counterfactual Expla- nations for Time Series Classification. In: 21st IEEE International Conference on Machine Learning and Applications (ICMLA), pp. 29–36 (2022)

2022
[17]

In: 29th International Conference on Case-Based Reasoning Research and Development (ICCBR), pp

Delaney, E., Greene, D., Keane, M.T.: Instance-based counterfactual explanations for time series classification. In: 29th International Conference on Case-Based Reasoning Research and Development (ICCBR), pp. 32–47 (2021)

2021
[18]

In: 2021 International Conference on Applied Artificial 21 Intelligence (ICAPAI), pp

Ates, E., Aksar, B., Leung, V.J., Coskun, A.K.: Counterfactual explanations for multivariate time series. In: 2021 International Conference on Applied Artificial 21 Intelligence (ICAPAI), pp. 1–8 (2021)

2021
[19]

In: 2020 IEEE Second International Conference on Cognitive Machine Intelligence (CogMI), pp

Guidotti, R., Monreale, A., Spinnato, F., Pedreschi, D., Giannotti, F.: Explain- ing any time series classifier. In: 2020 IEEE Second International Conference on Cognitive Machine Intelligence (CogMI), pp. 167–176 (2020)

2020
[20]

ACM Transactions on Knowledge Discovery from Data 18(2), 1–34 (2023)

Spinnato, F., Guidotti, R., Monreale, A., Nanni, M., Pedreschi, D., Giannotti, F.: Understanding any time series classifier with a subsequence-based explainer. ACM Transactions on Knowledge Discovery from Data 18(2), 1–34 (2023)

2023
[21]

Machine Learning, 1–31 (2024)

Wang, Z., Samsten, I., Miliou, I., Mochaourab, R., Papapetrou, P.: Glacier: guided locally constrained counterfactual explanations for time series classification. Machine Learning, 1–31 (2024)

2024
[22]

In: Wrembel, R., Gamper, J., Kotsis, G., Tjoa, A.M., Khalil, I

Li, P., Bahri, O., Boubrahimi, S.F., Hamdi, S.M.: Attention-based counterfactual explanation for multivariate time series. In: Wrembel, R., Gamper, J., Kotsis, G., Tjoa, A.M., Khalil, I. (eds.) Big Data Analytics and Knowledge Discovery, pp. 287–293. Springer, Cham (2023)

2023
[23]

Data Mining and Knowledge Discovery 38(6), 3347–3371 (2024)

Bahri, O., Li, P., Filali Boubrahimi, S., Hamdi, S.M.: Discord-based counter- factual explanations for time series classification. Data Mining and Knowledge Discovery 38(6), 3347–3371 (2024)

2024
[24]

arXiv preprint arXiv:2409.09461 (2024)

Huang, Q., Kitharidis, S., B¨ ack, T., Stein, N.: TX-Gen: Multi-Objective Opti- mization for Sparse Counterfactual Explanations for time-series classification. arXiv preprint arXiv:2409.09461 (2024)

2024 arXiv
[25]

In: World Conference on Explainable Artificial Intelligence (XAI), pp

Refoyo, M., Luengo, D.: Sub-SpaCE: Subsequence-Based Sparse Counterfactual Explanations for Time Series Classification Problems. In: World Conference on Explainable Artificial Intelligence (XAI), pp. 3–17 (2024). Springer

2024
[26]

IEEE Transactions on Evolutionary Computation 6(2), 182–197 (2002)

Deb, K., Pratap, A., Agarwal, S., Meyarivan, T.: A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation 6(2), 182–197 (2002)

2002
[27]

Knowledge-Based Systems 263, 110273 (2023)

Saeed, W., Omlin, C.: Explainable AI (XAI): A systematic meta-survey of cur- rent challenges and future opportunities. Knowledge-Based Systems 263, 110273 (2023)

2023
[28]

arXiv preprint arXiv:2104.00950 (2021)

Rojat, T., Puget, R., Filliat, D., Del Ser, J., Gelin, R., D ´ ıaz-Rodr ´ ıguez, N.: Explainable Artificial Intelligence (XAI) on timeseries data: A survey. arXiv preprint arXiv:2104.00950 (2021)

2021 arXiv
[29]

Pattern Recognition 150, 110309–110320 (2024) 22

Vielhaben, J., Lapuschkin, S., Montavon, G., Samek, W.: Explainable AI for time series via virtual inspection layers. Pattern Recognition 150, 110309–110320 (2024) 22

2024
[30]

IEEE Access 10, 100700–100724 (2022)

Theissler, A., Spinnato, F., Schlegel, U., Guidotti, R.: Explainable AI for time series classification: a review, taxonomy and research directions. IEEE Access 10, 100700–100724 (2022)

2022
[31]

In: 34th International Conference on Neural Information Processing Systems

Ismail, A.A., Gunady, M., Bravo, H.C., Feizi, S.: Benchmarking Deep Learning Interpretability in Time Series Predictions. In: 34th International Conference on Neural Information Processing Systems. NIPS’20, pp. 6441–6452 (2020)

2020
[32]

Pattern Recognition 151, 110418–110436 (2024)

Olivas-Padilla, B.E., Manitsaris, S., Glushkova, A.: Explainable AI in human motion: A comprehensive approach to analysis, modeling, and generation. Pattern Recognition 151, 110418–110436 (2024)

2024
[33]

arXiv preprint arXiv:2208.10462 (2022)

Bahri, O., Boubrahimi, S.F., Hamdi, S.M.: Shapelet-based counterfactual expla- nations for multivariate time series. arXiv preprint arXiv:2208.10462 (2022)

2022 arXiv
[34]

In: International Conference on Artificial Intelligence and Statistics, pp

Sivill, T., Flach, P.: LIMESegment: Meaningful, Realistic Time Series Explana- tions. In: International Conference on Artificial Intelligence and Statistics, pp. 3418–3433 (2022). PMLR

2022
[35]

Data Mining and Knowledge Discovery 34(6), 1936–1962 (2020)

Ismail Fawaz, H., Lucas, B., Forestier, G., Pelletier, C., Schmidt, D.F., Weber, J., Webb, G.I., Idoumghar, L., Muller, P.-A., Petitjean, F.: Inceptiontime: Find- ing Alexnet for time series classification. Data Mining and Knowledge Discovery 34(6), 1936–1962 (2020)

2020
[36]

IEEE/CAA Journal of Automatica Sinica 6(6), 1293–1305 (2019)

Dau, H.A., Bagnall, A., Kamgar, K., Yeh, C.-C.M., Zhu, Y., Gharghabi, S., Ratanamahatana, C.A., Keogh, E.: The UCR time series archive. IEEE/CAA Journal of Automatica Sinica 6(6), 1293–1305 (2019)

2019
[37]

Bagnall, A., Dau, H.A., Lines, J., Flynn, M., Large, J., Bostrom, A., Southam, P., Keogh, E.: The UEA multivariate time series classification archive, 2018. arXiv preprint arXiv:1811.00075 (2018) 23 Appendix A Mask and mutation ablation study We evaluated the influence of muta...

2018 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

SN Computer Science 3, 158–178 (2022)

Sarker, I.: AI-Based Modeling: Techniques, Applications and Research Issues Towards Automation, Intelligent and Smart Systems. SN Computer Science 3, 158–178 (2022)

work page 2022

[2] [2]

Information Fusion 58, 82–115 (2020)

Barredo Arrieta, A., D ˜Aaz-Rodr ˜Aguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., Chatila, R., Herrera, F.: Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion 58, 82–115 (2020)

work page 2020

[3] [3]

https://christophm.github

Molnar, C.: Interpretable Machine Learning, (2022). https://christophm.github. io/interpretable-ml-book

work page 2022

[4] [4]

Harvard Journal of Law & Technology 31, 841–888 (2017)

Wachter, S., Mittelstadt, B., Russell, C.: Counterfactual explanations without opening the black box: Automated decisions and the GDPR. Harvard Journal of Law & Technology 31, 841–888 (2017)

work page 2017

[5] [5]

In: Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19, pp

Byrne, R.M.J.: Counterfactuals in Explainable Artificial Intelligence (XAI): Evi- dence from human reasoning. In: Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19, pp. 6276–6282 (2019)

work page 2019

[6] [6]

Artificial Intelligence 267, 1–38 (2019) 20

Miller, T.: Explanation in Artificial Intelligence: Insights from the social sciences. Artificial Intelligence 267, 1–38 (2019) 20

work page 2019

[7] [7]

Data Mining and Knowledge Discovery, 2770–2824 (2022)

Guidotti, R.: Counterfactual explanations and how to find them: literature review and benchmarking. Data Mining and Knowledge Discovery, 2770–2824 (2022)

work page 2022

[8] [8]

ACM Computing Surveys 56(12) (2024)

Verma, S., Boonsanong, V., Hoang, M., Hines, K., Dickerson, J., Shah, C.: Coun- terfactual explanations and algorithmic recourses for machine learning: A review. ACM Computing Surveys 56(12) (2024)

work page 2024