A Scalable Approach to Covariate and Concept Drift Management via Adaptive Data Segmentation

Govind Waghmare; Siddhartha Asthana; Sonia Gupta; Vennela Yarabolu

REVIEW 4 major objections 6 minor 49 references

A Scalable Approach to Covariate and Concept Drift Management via Adaptive Data Segmentation

T0 review · 4 major / 6 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 Two cheap data-selection rules—gradient filtering and leaf ranking—can keep models accurate under both covariate and concept shift by training on a curated subset of historical data instead of discarding it.

desk verdict A reasonable engineering combination of Quilt and MatchMaker, with useful results on some datasets, but the central claim is contradicted by its own tables and the statistics are too thin to carry the conclusion. read the letter →

arxiv 2411.15616 v1 pith:ZNQCNQOY submitted 2024-11-23 cs.LG

Vennela Yarabolu , Govind Waghmare , Sonia Gupta , Siddhartha Asthana This is my paper · ORCID

classification cs.LG

keywords conceptdriftcovariateshiftdatasegmentationsubsetselectionadaptationrandomforestleafrankinggradientdisparitymachinelearningdeployment

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that drifted historical data is not waste: it is a resource that, if selected correctly, keeps models accurate as the world changes. The authors propose a framework that first discards data segments whose gradient statistics diverge from a validation set drawn from the current segment (concept drift), then ranks the remaining batches by how often they land in the same random-forest leaf nodes as validation samples (covariate shift). Training on the top-ranked subset, they claim, consistently beats training on all data, on recent data only, and on several existing drift-adaptation and subset-selection baselines, while using far fewer training samples. If the claim holds, drift management becomes a data-selection layer that can be dropped into existing pipelines, saving compute and money rather than adding it.

What carries the argument

The load-bearing mechanism is the combination of two scoring rules. For concept drift, the paper computes validation-set gradients and compares each historical segment's average gradient to them, keeping a segment only when its dot product with the validation gradient (gain) is positive and its L2 distance (disparity) is below a tuned threshold. For covariate shift, a random forest trained on all batches decides which batches are relevant: for each validation sample, batches are ranked by how many of their training points fall in the same leaf node, and the best qualifying batch is added to the training set. The first rule removes concept-drifted history; the second rule concentrates training on covariate-similar data; together they shrink the training set while keeping it aligned with the validation distribution.

What would settle it

Table 3 already contains a concrete counterexample to the claim of consistent outperformance: on Weather, training on all historical data and one leading subset-selection baseline both report higher accuracy than the method (0.800 and 0.793 versus 0.778), and a controlled synthetic stream whose validation segment is deliberately non-representative of the next batch would test whether the same failure propagates.

Watch

Extended reading notes

Core claim

The paper's central claim is that a two-stage segmentation procedure—filter out concept-drifted segments using a gradient disparity/gain test, then rank covariate-similar batches by leaf-concentration in a random forest—produces a training set that is more accurate for the next test batch than the full history, the current segment, or standard baselines. On the ten datasets reported, the method achieves the best accuracy and F1 in most rows while using between roughly 53% and 94% of the data, and the ablation shows that peak accuracy often occurs well before 100% data utilization. The authors take this as evidence that explicitly incorporating drifted data, rather than discarding it, improves accuracy and robustness.

Load-bearing premise

The whole selection procedure assumes that a validation set drawn from the current data segment is representative of the next batch of test data; if it is not, the gradient filter and the leaf rankings can point at the wrong segments and the accuracy gains disappear.

Editorial extensions

If this is right

Training-data volume can drop to roughly half while maintaining or improving accuracy, so operational cost and latency shrink with the data.
The method addresses covariate shift and concept drift in one pipeline, so deployments do not need separate detectors and adapters for each.
Because the selection rules only need last-layer gradients and random-forest leaf assignments, the framework can be added to existing supervised pipelines without new label acquisition.
The ablation's peak-before-100% pattern suggests that more history is not always better; a tuned subset can beat the full dataset.

Reading between the lines

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

A testable extension: when the current segment is not representative of the next live batch (e.g., sudden drift or seasonality), the validation-grounded selection should underperform; this is a direct consequence of the framework's dependence on the validation set and is not tested in the paper.
The random-forest leaf ranking could double as an interpretable diagnostic, showing which historical batches a deployed model still relies on, which the paper does not develop.
The gradient disparity threshold $T_d$ is tuned per segment via Bayesian optimization on the validation set; a cheaper online rule for setting it on the fly would determine whether the approach can run truly continuously rather than in batch-retrospective mode.
The method's improvement over full-data training on most datasets is consistent with viewing subset selection as a form of denoising or regularisation; whether the same gains appear for deep models beyond a single hidden layer remains open.

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

The genuinely new thing here is the integration: using Quilt's gradient-based segment filtering for concept drift and MatchMaker's random-forest leaf ranking for covariate shift in one pipeline. That combination is not in the cited prior work, and the paper shows it can beat Quilt on several datasets while using less data. The runtime numbers are also plausible, and the ablation in Table 4 gives some evidence that both components contribute. Credit where it's due: this is a sensible engineering contribution that could be useful to practitioners.

The soft spots are real and load-bearing. Section 7.4 says the method "consistently outperforms all the baselines," but their own Tables 2 and 3 show Weather where Full Data, GLISTER, and GRAD-MATCH all beat Our Method, and Covcon where it only ties Quilt. That claim is simply false as written. On top of that, there are no error bars or repeated runs anywhere, so even the wins that remain (Electricity 0.833 vs Quilt 0.831, SEA 0.899 vs 0.893) are within the sort of noise you'd expect from a single run. The per-dataset Bayesian tuning of the disparity threshold also makes the comparison less clean, though that's not fatal if the procedure is described honestly.

The weakest assumption is the validation set. Algorithm 2 computes gradients and selects batches from d_VN, and the threshold is tuned on it. If that validation slice isn't representative of the next live segment, the whole selection mechanism can be miscalibrated. The paper doesn't address this beyond treating the last segment as the test set. That's a structural limitation, not a minor omission.

No code is released, which matters here because the method is a combination of existing pieces and the empirical evaluation is the contribution. The citation pattern is fine, and the paper is honest about building on Quilt and MatchMaker. The limitations section is short but does acknowledge real-time segmentation challenges.

Who is this for? ML engineers in industry who need a practical drift-management module and want to see whether combining these two ideas works. As a research paper, the central claim needs repair. I'd send it to review only if the authors fix the overclaim and add variance-aware runs. As is, I wouldn't cite it beyond a footnote about the combination.

Referee Report

4 major / 6 minor

Summary. The paper proposes a drift-management framework that combines two existing ideas: MatchMaker-style random-forest leaf-based batch ranking for covariate shift (Algorithm 1) and Quilt-style gradient-based segment selection for concept drift (Algorithm 2). The selected training batches are used to train a small neural network, and the method is evaluated on five synthetic and five real-world datasets against a broad set of baselines. The paper's central claim, stated in Section 7.4, is that the proposed method 'consistently outperforms all the baselines in terms of accuracy' while using substantially less data. The contribution is framed as a drop-in, scalable improvement for continuous ML systems facing both covariate and concept drift.

Significance. If the reported results held, the framework would be a practical and useful contribution: it is computationally light, uses less training data than full-data training, and integrates two complementary drift mechanisms. The algorithms are clearly specified, and the ablation in Table 4 gives a transparent picture of per-module runtime. However, the significance of the paper rests almost entirely on the empirical claim of consistent superiority, and that claim is not supported by the paper's own tables. The results also lack any uncertainty quantification, and the validation protocol is not fully specified. These are not cosmetic gaps: without them, the contribution is an interesting engineering combination rather than a demonstrated improvement.

major comments (4)

[Section 7.4, Tables 2-3] The claim that 'our method consistently outperforms all the baselines in terms of accuracy' is directly contradicted by the reported numbers. In Table 3, on Weather, Full Data achieves 0.800, GLISTER 0.793, and GRAD-MATCH 0.795, all above Our Method's 0.778. In Table 2, on Covcon, Our Method ties Quilt at 0.988, which is at best parity rather than consistent outperformance. The claim must be weakened to a per-dataset comparison, or the experiments must be designed and reported in a way that supports a strict ordering.
[Tables 2 and 3] All accuracy and F1 values are reported as single numbers with no standard deviations, confidence intervals, or number of repetitions. Several wins are very small, for example Electricity 0.833 vs Quilt 0.831 and SEA 0.899 vs Quilt 0.893; on Weather the method loses by 1.5-2.2 points. Without variance-aware repeated runs or significance tests, the 'consistent' superiority assertion is not statistically supported, even on the datasets where the point estimate is favorable.
[Section 7.2 and Section 7.4, Algorithm 2] The validation protocol is underspecified and potentially leaky. Algorithm 2 uses a validation set d_VN to compute gradients, select segments, and tune the disparity threshold T_d via Bayesian optimization, while Section 7.4 states that the latest segment is used 'to evaluate' the method. The paper never states whether d_VN is a disjoint held-out portion of the current segment or whether it overlaps with the test data used to report accuracy. If d_VN is drawn from the same current segment on which accuracy is reported, then model selection and threshold tuning see test labels, and the reported gains may be artifacts. This must be clarified, and the protocol should use a genuine temporal train/validation/test split.
[Section 7.2] The disparity threshold T_d is tuned per dataset via Bayesian optimization on the validation set, and the random forest batch size is chosen by grid search also per dataset. This tuning is reasonable for a practical system, but the paper should report how sensitive the final accuracies are to T_d and batch size; otherwise the headline numbers could reflect per-dataset overfitting to the validation segment.

minor comments (6)

[Section 7.1] The word 'orientiations' in the Hyperplane description is a typo for 'orientations'.
[Section 5.2] 'on can compute' should read 'one can compute' in the gradient computation paragraph.
[Section 7.4] 'under perform compared toour method' contains a missing space and should read 'underperform compared to our method'.
[Algorithm 1] The pseudocode appears incomplete: after the line storing S_i[k_i][t], there is no closing instruction for the per-tree loop and no explanation of how the per-tree scores are aggregated into the final batch ranking. Please complete the pseudocode or describe the aggregation in the surrounding text.
[Section 7.3] The HAT baseline is cited as reference [3], but reference [3] is the ADWIN paper by Bifet and Gavaldà, not a Hoeffding Adaptive Tree source; the citation should be corrected.
[Table 4] The column labeled 'Only Alg. 1' reports accuracy, but it is not clear what model is trained in that ablation. State explicitly whether the model is trained on all batches ranked by Algorithm 1 or only on the top-ranked batch per validation sample, and how the current segment is handled without Algorithm 2.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the framework is an empirical combination of two explicitly cited external methods, benchmarked independently against those methods and other baselines.

full rationale

The paper makes no claim that reduces by construction to its own inputs. Algorithm 1 is explicitly adopted from MatchMaker [31] and Algorithm 2 explicitly builds on Quilt [26], with disparity and gain scores quoted from Quilt. The central experimental claim is an accuracy comparison against Quilt and other baselines on held-out segment evaluations; the reported accuracies are not derived from the tuned disparity threshold or from the gradient-based selection scores themselves. Tuning T_d via Bayesian optimization on a validation set is ordinary hyperparameter selection, and the paper separately reports the fraction of data used rather than presenting data usage as a fitted prediction. There is no load-bearing self-citation chain: the paper contains no self-citations, and the use of Quilt's public codebase is only to generate baseline results, which is an independent comparison. The claim in Section 7.4 that the method 'consistently outperforms all the baselines' is contradicted by some entries in the paper's own Tables 2 and 3 (e.g., Weather), but an overstatement or internal inconsistency is a correctness and reporting concern, not circularity. Therefore the appropriate circularity score is 0.

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

The method introduces no new theoretical objects. Its load-bearing components are two existing heuristics plus several tuned hyperparameters. The most important free parameter is the per-segment disparity threshold, followed by the random-forest batch size.

free parameters (4)

Disparity threshold T_d = Not reported; Bayesian optimization in interval (0,2) per data segment
Controls which historical segments are included in training (Algorithm 2 line 'if G_d > 0 and D_d < T_d'). Tuned per segment on validation data, directly shapes the selected subset and therefore accuracy.
Random forest batch size = Varies per dataset (e.g., 100 for SEA, 50,000 for Covcon_5M)
Chosen by grid search over 3 to 5 values in Section 7.2. Defines the granularity of covariate-shift ranking and which batches enter B_best.
Random forest n_estimators and max_depth = 50 estimators, max depth 20
Listed as fixed after grid search. These affect leaf structure and therefore which batches are ranked close to test points.
Neural network hyperparameters = Learning rate 1e-3, hidden layer 256 units, early stopping patience 10
Fixed by hand; less critical to the central claim but still part of the final classifier.

assumptions (4)

domain assumption Random-forest leaf co-occurrence is a valid proxy for covariate shift between training batches and test points.
Adopted from MatchMaker [31]; used in Section 5.1 and Algorithm 1 to rank batches, with no theoretical guarantee provided.
domain assumption Gradient disparity and gain scores computed on the last layer approximate concept-drift severity and segment utility.
Adopted from Quilt [26]; the approximation replaces E[||y_t - y_v||] with ||E[g_t] - E[g_v]|| (Eq. 4) and uses gradient alignment (Eq. 5).
domain assumption A validation set drawn from the current segment is representative of the data the model will be tested on.
Algorithm 2 computes validation gradients and batch rankings on d_VN; Section 7.2 tunes the disparity threshold on this validation set before evaluating on the final segment.
ad hoc to paper Segments with positive gain and low disparity can be safely mixed for training without hurting the current segment.
Algorithm 2's selection rule (G_d > 0 and D_d < T_d) is a heuristic; no convergence or error-bound analysis is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Scalable Approach to Covariate and Concept Drift Management via Adaptive Data Segmentation." pith.science (2026). https://pith.science/paper/ZNQCNQOY

@misc{pith2026241115616,
  author       = {Pith},
  title        = {Pith review of: A Scalable Approach to Covariate and Concept Drift Management via Adaptive Data Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZNQCNQOY}},
  note         = {Machine review of arXiv:2411.15616}
}

read the original abstract

In many real-world applications, continuous machine learning (ML) systems are crucial but prone to data drift, a phenomenon where discrepancies between historical training data and future test data lead to significant performance degradation and operational inefficiencies. Traditional drift adaptation methods typically update models using ensemble techniques, often discarding drifted historical data, and focus primarily on either covariate drift or concept drift. These methods face issues such as high resource demands, inability to manage all types of drifts effectively, and neglecting the valuable context that historical data can provide. We contend that explicitly incorporating drifted data into the model training process significantly enhances model accuracy and robustness. This paper introduces an advanced framework that integrates the strengths of data-centric approaches with adaptive management of both covariate and concept drift in a scalable and efficient manner. Our framework employs sophisticated data segmentation techniques to identify optimal data batches that accurately reflect test data patterns. These data batches are then utilized for training on test data, ensuring that the models remain relevant and accurate over time. By leveraging the advantages of both data segmentation and scalable drift management, our solution ensures robust model accuracy and operational efficiency in large-scale ML deployments. It also minimizes resource consumption and computational overhead by selecting and utilizing relevant data subsets, leading to significant cost savings. Experimental results on classification task on real-world and synthetic datasets show our approach improves model accuracy while reducing operational costs and latency. This practical solution overcomes inefficiencies in current methods, providing a robust, adaptable, and scalable approach.

Figures

Figures reproduced from arXiv: 2411.15616 by the authors.

**Figure 2.** Subsegment selection approach using gradient [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Complete architecture of our method. The top branch focuses on concept drift while bottom branch ranks train [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Trade-off between % of data used vs accuracy. In our experiments with the Usenet2 and Weather dataset, we evaluated the effect of varying the proportion of data used for training on the model’s accuracy. The goal was to determine the minimal amount of data required to …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 39 canonical work pages

[1]

Manuel Baena-Garcıa, José del Campo-Ávila, Raul Fidalgo, Albert Bifet, Ricard Gavalda, and Rafael Morales-Bueno. 2006. Early drift detection method. InFourth international workshop on knowledge discovery from data streams , Vol. 6. Citeseer, 77–86

work page 2006
[2]

Michele Basseville, Igor V Nikiforov, et al . 1993. Detection of abrupt changes: theory and application. Vol. 104. prentice Hall Englewood Cliffs

work page 1993
[3]

Albert Bifet and Ricard Gavalda. 2007. Learning from time-changing data with adaptive windowing. In Proceedings of the 2007 SIAM international conference on data mining. SIAM, 443–448

2007
[4]

Albert Bifet and Ricard Gavaldà. 2009. Adaptive Learning from Evolving Data Streams. In Advances in Intelligent Data Analysis VIII , Niall M. Adams, Céline Robardet, Arno Siebes, and Jean-François Boulicaut (Eds.). Springer Berlin Hei- delberg, Berlin, Heidelberg, 249–260

work page 2009
[5]

Albert Bifet and Ricard Gavaldà. 2007. Learning from Time-Changing Data with Adaptive Windowing. Proceedings of the 7th SIAM International Conference on Data Mining

work page 2007
[6]

Dariusz Brzezinski and Jerzy Stefanowski. 2014. Reacting to Different Types of Concept Drift: The Accuracy Updated Ensemble Algorithm. IEEE Transactions on Neural Networks and Learning Systems (2014)

work page 2014
[7]

Alex Davies and Zoubin Ghahramani. 2014. The random forest kernel and other kernels for big data from random partitions.arXiv preprint arXiv:1402.4293 (2014)

arXiv 2014
[8]

Gregory Ditzler and Robi Polikar. 2011. Hellinger distance based drift detec- tion for nonstationary environments. In 2011 IEEE symposium on computational intelligence in dynamic and uncertain environments (CIDUE) . IEEE, 41–48

work page 2011

Show all 49 references

[9]

Gregory Ditzler, Manuel Roveri, Cesare Alippi, and Robi Polikar. 2015. Learning in nonstationary environments: A survey.IEEE Computational Intelligence Magazine 10, 4 (2015), 12–25

2015
[10]

Pedro Domingos and Geoff Hulten. 2000. Mining high-speed data streams. In Proceedings of the Sixth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

2000
[11]

Fan Dong, Jie Lu, Yiliao Song, Feng Liu, and Guangquan Zhang. 2021. A Drift Region-Based Data Sample Filtering Method. IEEE Transactions on Cybernetics (2021)

2021
[12]

Ryan Elwell and Robi Polikar. 2011. Incremental Learning of Concept Drift in Nonstationary Environments. IEEE Transactions on Neural Networks (2011)

2011
[13]

Wei Fan. 2004. Systematic data selection to mine concept-drifting data streams. In Proceedings of the Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining . Association for Computing Machinery

2004
[14]

Fischler and Robert C

Martin A. Fischler and Robert C. Bolles. 1981. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM (1981)

1981
[15]

Joao Gama, Pedro Medas, Gladys Castillo, and Pedro Rodrigues. 2004. Learning with drift detection. InAdvances in Artificial Intelligence–SBIA 2004: 17th Brazilian Symposium on Artificial Intelligence, Sao Luis, Maranhao, Brazil, September 29- Ocotber 1, 2004. Proceedings 17 . ...

2004
[16]

João Gama, Ricardo Rocha, and Pedro Medas. 2003. Accurate decision trees for mining high-speed data streams. Association for Computing Machinery

2003
[17]

João Gama, Indrundefined Žliobaitundefined, Albert Bifet, Mykola Pechenizkiy, and Abdelhamid Bouchachia. 2014. A survey on concept drift adaptation. ACM Comput. Surv. 46, 4, Article 44 (mar 2014), 37 pages. https://doi.org/10.1145/ 2523813

2014
[18]

Widmer Gerhard and Kubat Miroslav. 1996. Learning in the presence of concept drift and hidden contexts. Machine Learning (1996)

1996
[19]

Heitor Murilo Gomes, Albert Bifet, Jesse Read, Jean Paul Barddal, Fabrício Enem- breck, Bernhard Pfahringer, Geoff Holmes, and Talel Abdessalem. 2017. Adaptive random forests for evolving data stream classification. Machine Learning (2017)

2017
[20]

Arthur Gretton, Alex Smola, Jiayuan Huang, Marcel Schmittfull, Karsten Borg- wardt, and Bernhard Schölkopf. 2008. Covariate shift by kernel mean matching. (2008)

2008
[21]

Fabian Hinder, André Artelt, and Barbara Hammer. 2020. Towards non-parametric drift detection via dynamic adapting window independence drift detection (daw- idd). In International Conference on Machine Learning . PMLR, 4249–4259

2020
[22]

Mark G Kelly, David J Hand, and Niall M Adams. 1999. The impact of changing populations on classifier performance. In Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining . 367–371

1999
[23]

Daniel Kifer, Shai Ben-David, and Johannes Gehrke. 2004. Detecting Change in Data Streams

2004
[24]

Krishnateja Killamsetty, Sivasubramanian Durga, Ganesh Ramakrishnan, Abir De, and Rishabh Iyer. 2021. Grad-match: Gradient matching based data subset selection for efficient deep model training. InInternational Conference on Machine Learning. PMLR, 5464–5474

2021
[25]

Krishnateja Killamsetty, Durga Sivasubramanian, Ganesh Ramakrishnan, and Rishabh Iyer. 2021. GLISTER: Generalization based Data Subset Selection for Efficient and Robust Learning. arXiv:2012.10630 [cs.LG] https://arxiv.org/abs/ 2012.10630

2021 arXiv
[26]

Minsu Kim, Seong-Hyeon Hwang, and Steven Euijong Whang. 2024. Quilt: Robust Data Segment Selection against Concept Drifts. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 38. 21249–21257

2024
[27]

Ralf Klinkenberg. 2004. Learning drifting concepts: Example selection vs. example weighting. Intelligent Data Analysis 8 (2004)

2004
[28]

Ivan Koychev and Robert Lothian. 2006. Tracking Drifting Concepts by Time Window Optimisation. In Research and Development in Intelligent Systems XXII

2006
[29]

Krawczyk, Leandro L

B. Krawczyk, Leandro L. Minku, João Gama, Jerzy Stefanowski, and Michal Wozniak. 2017. Ensemble learning for data stream analysis: A survey. Inf. Fusion 37 (2017), 132–156. https://api.semanticscholar.org/CorpusID:1372281

2017
[30]

Jie Lu, Anjin Liu, Fan Dong, Feng Gu, Joao Gama, and Guangquan Zhang. 2018. Learning under concept drift: A review. IEEE transactions on knowledge and data engineering 31, 12 (2018), 2346–2363

2018
[31]

Ankur Mallick, Kevin Hsieh, Behnaz Arzani, and Gauri Joshi. 2022. Matchmaker: Data Drift Mitigation in Machine Learning for Large-Scale Systems. InProceedings of Machine Learning and Systems

2022
[32]

Yishay Mansour, Mehryar Mohri, and Afshin Rostamizadeh. 2008. Domain adaptation with multiple sources. Advances in neural information processing systems 21 (2008)

2008
[33]

Leandro L Minku, Allan P White, and Xin Yao. 2009. The impact of diversity on online ensemble learning in the presence of concept drift. IEEE Transactions on knowledge and Data Engineering 22, 5 (2009), 730–742

2009
[34]

Moreno-Torres, Troy Raeder, Rocío Alaiz-Rodríguez, Nitesh V

Jose G. Moreno-Torres, Troy Raeder, Rocío Alaiz-Rodríguez, Nitesh V. Chawla, and Francisco Herrera. 2012. A unifying view on dataset shift in classification. Pattern Recognition (2012)

2012
[35]

Ali Pesaranghader and Herna L. Viktor. 2016. Fast Hoeffding Drift Detection Method for Evolving Data Streams. InMachine Learning and Knowledge Discovery in Databases

2016
[36]

Ali Pesaranghader, Herna L Viktor, and Eric Paquet. 2018. McDiarmid drift detection methods for evolving data streams. In2018 International joint conference on neural networks (IJCNN) . IEEE, 1–9

2018
[37]

Sergio Ramírez-Gallego, Bartosz Krawczyk, Salvador García, Michal Wozniak, and Francisco Herrera. 2017. A survey on Data Preprocessing for Data Stream Mining: Current status and future directions. (2017)

2017
[38]

Hidetoshi Shimodaira. 2000. Improving predictive inference under covariate shift by weighting the log-likelihood function. Journal of statistical planning and inference 90, 2 (2000), 227–244

2000
[39]

Yiliao Song, Jie Lu, Anjin Liu, Haiyan Lu, and Guangquan Zhang. 2021. A Segment-Based Drift Adaptation Method for Data Streams. IEEE Transactions on Neural Networks and Learning Systems (2021)

2021
[40]

Masashi Sugiyama, Matthias Krauledat, and Klaus-Robert Müller. 2007. Covariate shift adaptation by importance weighted cross validation. Journal of Machine Learning Research 8, 5 (2007)

2007
[41]

Yu Sun, Ke Tang, Zexuan Zhu, and Xin Yao. 2018. Concept drift adaptation by exploiting historical knowledge. IEEE transactions on neural networks and learning systems 29, 10 (2018), 4822–4832

2018
[42]

Abhijit Suprem, Joy Arulraj, Calton Pu, and Joao Ferreira. 2020. Odin: Automated drift detection and recovery in video analytics. arXiv preprint arXiv:2009.05440 (2020)

2020 arXiv
[43]

Ashraf Tahmasbi, Ellango Jothimurugesan, Srikanta Tirthapura, and Phillip B. Gibbons. 2020. DriftSurf: A Risk-competitive Learning Algorithm under Concept Drift. arXiv:2003.06508 [cs.LG] https://arxiv.org/abs/2003.06508

2020 arXiv
[44]

Alexey Tsymbal. 2004. The problem of concept drift: definitions and related work. Computer Science Department, Trinity College Dublin 106, 2 (2004), 58

2004
[45]

Geoffrey I Webb, Roy Hyde, Hong Cao, Hai Long Nguyen, and Francois Petitjean
[46]

Keisuke Yamazaki, Motoaki Kawanabe, Sumio Watanabe, Masashi Sugiyama, and Klaus-Robert Müller. 2007. Asymptotic bayesian generalization error when training and test distributions are different. InProceedings of the 24th international conference on Machine learning . 1079–1086

2007
[47]

Peng Zhao, Le-Wen Cai, and Zhi-Hua Zhou. 2020. Handling concept drift via model reuse. Machine Learning (2020)

2020
[48]

Indre Zliobaite. 2013. How good is the electricity benchmark for evaluating concept drift adaptation. arXiv preprint arXiv:1301.3524 (2013)

2013 arXiv
[2016]

Data Mining and Knowledge Discovery 30, 4 (2016), 964–994

Characterizing concept drift. Data Mining and Knowledge Discovery 30, 4 (2016), 964–994

2016

Pith tools

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

[1] [1]

Manuel Baena-Garcıa, José del Campo-Ávila, Raul Fidalgo, Albert Bifet, Ricard Gavalda, and Rafael Morales-Bueno. 2006. Early drift detection method. InFourth international workshop on knowledge discovery from data streams , Vol. 6. Citeseer, 77–86

work page 2006

[2] [2]

Michele Basseville, Igor V Nikiforov, et al . 1993. Detection of abrupt changes: theory and application. Vol. 104. prentice Hall Englewood Cliffs

work page 1993

[3] [3]

Albert Bifet and Ricard Gavalda. 2007. Learning from time-changing data with adaptive windowing. In Proceedings of the 2007 SIAM international conference on data mining. SIAM, 443–448

2007

[4] [4]

Albert Bifet and Ricard Gavaldà. 2009. Adaptive Learning from Evolving Data Streams. In Advances in Intelligent Data Analysis VIII , Niall M. Adams, Céline Robardet, Arno Siebes, and Jean-François Boulicaut (Eds.). Springer Berlin Hei- delberg, Berlin, Heidelberg, 249–260

work page 2009

[5] [5]

Albert Bifet and Ricard Gavaldà. 2007. Learning from Time-Changing Data with Adaptive Windowing. Proceedings of the 7th SIAM International Conference on Data Mining

work page 2007

[6] [6]

Dariusz Brzezinski and Jerzy Stefanowski. 2014. Reacting to Different Types of Concept Drift: The Accuracy Updated Ensemble Algorithm. IEEE Transactions on Neural Networks and Learning Systems (2014)

work page 2014

[7] [7]

Alex Davies and Zoubin Ghahramani. 2014. The random forest kernel and other kernels for big data from random partitions.arXiv preprint arXiv:1402.4293 (2014)

arXiv 2014

[8] [8]

Gregory Ditzler and Robi Polikar. 2011. Hellinger distance based drift detec- tion for nonstationary environments. In 2011 IEEE symposium on computational intelligence in dynamic and uncertain environments (CIDUE) . IEEE, 41–48

work page 2011