Pith. sign in

REVIEW 4 major objections 4 minor 42 references

Adapting to Fragmented and Evolving Data: A Fisher Information Perspective

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read FADE adapts models to drifting data online using Fisher information, with up to 19 percent accuracy gains.

desk verdict A plausible Fisher-regularizer idea for sequential covariate shift, but the arXiv version is under-specified and internally contradictory, so the headline results cannot be trusted as written. read the letter →

arxiv 2507.18996 v1 pith:56XCTHIE submitted 2025-07-25 cs.LG stat.ML

Behraj Khan , Tahir Qasim Syed , Nouman Muhammad Durrani This is my paper · ORCID
classification cs.LGstat.ML
keywords Fisherinformationsequentialcovariateshiftonlineadaptationtest-timefederatedlearningCramér-RaoboundKLdivergencedistribution
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

FADE is an online method for keeping a model accurate when the input distribution changes from batch to batch while labels for the new data never arrive. It claims that a lightweight regularizer built from the Fisher information matrix, combined with a shift detector that multiplies KL divergence between consecutive batches by the change in that matrix, lets a fixed-memory learner adapt to sequential covariate shift without task boundaries, replay, or target labels. Across vision, language, and tabular benchmarks, the paper reports up to 19 percent higher accuracy under severe shifts than methods such as TENT and DIW, and roughly 3–5 percent gains over FedAvg-style baselines in federated settings. The paper also states bounded regret and $O(1/\sqrt{T})$ parameter consistency as theorems, though the derivations are not included in the preprint.

What carries the argument

The load-bearing object is the Fisher information matrix $I_t(\theta) = -\mathbb{E}_{x\sim P_t}[\nabla^2_\theta \log p_\theta(x)]$, used as a time-varying sensitivity map. It carries the argument in two roles: its smoothed version $I_{\mathrm{global}}$ defines the quadratic regularizer that prevents updates from destroying previously learned behavior, and its change across consecutive batches, multiplied by the KL divergence $D_{\mathrm{KL}}(P_t\|P_{t-1})$, forms the shift trigger $\tau_t$. The diagonal approximation keeps the per-batch cost linear in the number of parameters.

What would settle it

Run FADE on a synthetic stream where the true input density is known, compute the Fisher term exactly from that density, and compare the resulting shift signal to the paper's classifier-based approximation; if the approximation does not track the true signal and the accuracy gains disappear under the exact signal, the mechanism is not what the paper claims.

Watch

Extended reading notes

Core claim

The paper's core claim is that an online learner can ride out sequential covariate shift by treating the Fisher information matrix (FIM) as a sensitivity map for the current input distribution and using it in two ways: a detection signal $\tau_t = \|I_t(\theta)-I_{t-1}(\theta)\|_F \cdot D_{\mathrm{KL}}(P_t\|P_{t-1})$ that triggers adaptation only when both distributional and geometric changes are large, and a regularizer $\mathcal{L}_t = \mathbb{E}[\ell(f_\theta(x),y)] + \lambda(\theta-\mu_{t-1})^\top I_{\mathrm{global}}^{(t-1)}(\theta-\mu_{t-1})$ that anchors the update to previously learned parameters along directions the data still cares about. With the FIM smoothed over time and a diagonal approximation for deep networks, FADE updates only when the shift signal crosses a threshold, keeps only two batches in memory, and never sees target labels. The paper reports consistent gains over ERM, EWC, TENT, DIW, and federated baselines across seven datasets, and states Theorems 2.1 and 2.2 claiming bounded regret and $O(1/\sqrt{T})$ parameter consistency.

Load-bearing premise

The whole mechanism assumes that a label-only classifier yields a meaningful Fisher information matrix for the input distribution and that the KL divergence between unlabeled high-dimensional batches can be reliably estimated; if either fails, the shift signal and the regularizer are not well defined.

Editorial extensions

If this is right

  • An online learner can adapt to evolving input distributions with fixed memory, no target labels, and no task boundaries, provided it can estimate a Fisher information matrix and a batch-wise KL divergence.
  • Fisher-weighted regularization should reduce catastrophic forgetting under shift; the paper reports average forgetting of 1.4 percent, lower than all baselines.
  • Treating federated clients as temporally fragmented batches gives a direct extension of sequential-shift adaptation to decentralized heterogeneity, with gains over FedAvg and SCAFFOLD without explicit personalization.
  • If the stated guarantees hold, FADE-like updates have bounded regret in online convex settings and converge to the global-distribution optimum at the usual $O(1/\sqrt{T})$ rate.

Reading between the lines

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

  • A direct test of the mechanism would be to replace the unstated density $p_\theta(x)$ in the Fisher information matrix with an explicit density model; if performance survives, the Fisher signal is the active ingredient, and if it collapses, the implementation is silently relying on a heuristic.
  • The KL divergence between unlabeled high-dimensional batches is used without a stated estimator, so in practice the threshold $\gamma$ may matter more for FADE's behavior than the theoretical framing suggests.
  • Because the two theorems are stated without derivation in this preprint, the bounded-regret and consistency claims remain open until a proof or a counterexample appears.
  • The same shift-trigger-plus-anchor recipe could be applied to label shift by replacing input KL divergence with divergence on predictive distributions, a swap the paper does not explore.
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.

Referee Report

4 major / 4 minor

Summary. The paper proposes FADE, an online method for adapting a model to sequential covariate shift (SCS) using a Fisher-information regularizer and a KL-based shift trigger. It claims bounded regret and parameter-consistency guarantees, reports large accuracy gains over ERM, EWC, TENT, and DIW on vision, language, and tabular benchmarks, and extends the idea to federated learning by treating clients as sequential batches. The central contribution is a lightweight, label-free, fixed-memory adaptation rule whose updates are supposed to be guided by the Cramér-Rao bound.

Significance. If properly substantiated, the proposal would be a useful default for online adaptation: it requires no target labels, no replay, and only two batches in memory, and the empirical results across modalities indicate that the general idea may have practical value. The paper's explicit use of Fisher geometry to combine shift detection and regularized adaptation is a reasonable design, and the ablation study attempts to isolate the contributions of the two main components. However, the theoretical guarantees and the algorithmic specification are currently unsupported, no proofs or code are supplied, and the main empirical claim is not reproducible from the manuscript as written.

major comments (4)
  1. [Algorithm 1 / §4.4] Algorithm 1 lines 5–10 update the parameters only when τ_t > γ; otherwise θ_t is set to θ_{t−1}. This contradicts Eq. (6), which defines a current-batch loss L_t(θ) for every t, and §4.4, which states models are 'trained (or adapted)' on each batch. Under the pseudocode, FADE cannot learn from non-shift batches, so the high accuracy reported for mild-shift regimes (Table 5, 94.1%) is unexplained; if the implementation trains on all batches, the published algorithm is wrong. Because no code is supplied, the main empirical claim is not reproducible from the manuscript as written.
  2. [§2.4, Eqs. (7)–(8)] Theorems 2.1 and 2.2 are stated without proofs, and Eq. (7) bounds regret by a term containing Δ_SCS, the same shift measure that drives detection and adaptation; as written the bound is a restatement of the shift magnitude rather than an informative derived guarantee. The theorem statements also omit the precise assumptions (boundedness, convexity, Lipschitz constants) needed to make the rates meaningful, and no appendix or supplementary material is provided. The theoretical component of the central claim is therefore unverified.
  3. [Eq. (2), §4.5] The Fisher information matrix is defined as -E_{x~Pt} ∇^2_θ log p_θ(x), but FADE is applied to discriminative classifiers that do not model p_θ(x); no construction for a density model is given. Section 4.5 says the FIM is approximated by 'the diagonal of the empirical Hessian computed on a subset of each batch,' but it does not specify the Hessian of which objective. Without this specification, both the shift signal τ_t in Eq. (3) and the regularizer in Eq. (6) are not well-defined for the vision and language networks tested.
  4. [Eq. (3), Algorithm 1] The shift signal relies on D_KL(P_t || P_{t−1}) between empirical batch distributions, but no estimator for this quantity is provided. In high-dimensional inputs, standard plug-in KL estimates between empirical distributions are degenerate, and the paper gives no density model or non-parametric estimator; this affects the detection step for all experiments, so the results cannot be reproduced independently. A practical estimator with its hyperparameters should be stated.
minor comments (4)
  1. [§2.2] The reference to 'Figure ??' in §2.2 should be a numbered figure reference; the figure is not resolved in the manuscript.
  2. [§5.5, Table 3] Table 3 and §5.5 introduce FIRE as a comparison method but provide no description, citation, or configuration; the comparison with FADE is therefore incomplete.
  3. [Contributions, §4.5] The contribution list promises 'complete implementation code,' but no code, repository link, or appendix is included in the manuscript.
  4. [Introduction, Sections 2–6] The introduction states that Section 2 reviews related work and Section 3 formulates the approach, but the actual paper places Related Work as Section 3 and the formulation as Section 2; the section numbering should be reconciled.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the empirical claims rest on external benchmarks and the unproved shift-dependent regret bound is a rigor gap rather than a circular reduction.

full rationale

The paper's central empirical claim — up to 19% higher accuracy under severe shifts — is supported by comparisons against external baselines (ERM, EWC, TENT, DIW, FedAvg, etc.) on seven datasets, and the reported numbers are measured results, not quantities derived from FADE's equations. The Fisher-regularized objective (Eq. 6) and the KL/FIM shift trigger (Eq. 3) are design choices, not fitted to the reported test accuracies. The stated theoretical results are not circular in the required sense: Theorem 2.1 is unproved and its regret bound is expressed in terms of the paper's own shift measure Δ_SCS, but that makes the theorem loose or vacuous as a 'bounded' guarantee, not a reduction of the empirical outcome to its inputs. Theorem 2.2 is also unproved and appears inconsistent with non-stationarity, but again that is a correctness/rigor concern. The sole self-citation ([3]) is used to motivate the importance of sequential covariate shift, not as a load-bearing premise for the method; no uniqueness theorem is imported from the authors' prior work, and no fitted parameter is renamed as a prediction. Algorithm 1's freeze-on-no-shift branch conflicts with the statement in §4.4 that models are trained on every batch, but this is an internal consistency/reproducibility problem, not circularity. The FIM in Eq. (2) is underspecified for a discriminative model, yet that is a modeling gap, not a self-definitional derivation. Under the hard rule requiring an explicit reduction or fitted-input-as-prediction, no such step is present; the appropriate finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central machinery rests on a Fisher matrix over an input density p_theta(x), on KL estimates between batches, and on Lipschitz smoothness for the regret bound. Hyperparameters lambda, alpha, gamma, and epsilon are not fully specified and are partly tuned on a validation stream. No invented entities are introduced.

free parameters (4)
  • lambda (regularization strength) = not reported; selected via held-out validation stream (Sec. 4.5)
    Controls the strength of the Fisher regularizer in Eq. (6); the reported empirical gains depend on this tuned choice.
  • alpha (FIM smoothing factor) = not reported; selected via held-out validation stream (Sec. 4.5)
    Controls exponential smoothing of the global FIM in Eq. (5); no default or range is reported.
  • gamma (shift detection threshold) = not reported; described as 'predefined'
    Triggers adaptation in Algorithm 1 and Eq. (3); without a value or calibration procedure, the behavior of the detector is unspecified.
  • epsilon (SCS significance threshold) = appears in Eq. (1), not operationalized
    Used to define SCS in Eq. (1) but never set or used in the algorithm or experiments, leaving the formal definition of SCS incomplete.
assumptions (5)
  • ad hoc to paper A Fisher information matrix can be defined as -E_{x~Pt} grad^2_theta log p_theta(x) for a discriminative classifier.
    Eq. (2) uses log p_theta(x), but the tested classifiers model p_theta(y|x), not an input density. No generative model is specified, so the central signal is not well-defined for the architectures used.
  • domain assumption Lipschitz-continuous loss functions and bounded shifts are sufficient for the stated regret bound.
    Theorem 2.1 invokes Lipschitz-continuous losses, but the full assumption set is not stated and no proof is given, so the bound is not established.
  • domain assumption KL divergence between batch feature distributions can be reliably estimated from unlabeled batches in high dimensions.
    The shift signal in Eq. (3) requires D_KL(Pt || Pt-1), but no estimator, sample-size condition, or dimensionality reduction is provided, which is nontrivial for image and text data.
  • domain assumption A smoothed diagonal FIM captures the temporal sensitivity of the model.
    Eq. (5) and the diagonal approximation assume that a smoothed diagonal Fisher matrix is a sufficient sensitivity statistic; the paper acknowledges the diagonal approximation can miss inter-parameter correlations.
  • domain assumption A held-out validation stream is available for tuning alpha and lambda.
    Section 4.5 selects alpha and lambda via a held-out validation stream, but fully online deployment may not provide such a stream; this weakens the 'requires no tuning' claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adapting to Fragmented and Evolving Data: A Fisher Information Perspective." pith.science (2026). https://pith.science/paper/56XCTHIE

@misc{pith2026250718996,
  author       = {Pith},
  title        = {Pith review of: Adapting to Fragmented and Evolving Data: A Fisher Information Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/56XCTHIE}},
  note         = {Machine review of arXiv:2507.18996}
}
read the original abstract

Modern machine learning systems operating in dynamic environments often face \textit{sequential covariate shift} (SCS), where input distributions evolve over time while the conditional distribution remains stable. We introduce FADE (Fisher-based Adaptation to Dynamic Environments), a lightweight and theoretically grounded framework for robust learning under SCS. FADE employs a shift-aware regularization mechanism anchored in Fisher information geometry, guiding adaptation by modulating parameter updates based on sensitivity and stability. To detect significant distribution changes, we propose a Cramer-Rao-informed shift signal that integrates KL divergence with temporal Fisher dynamics. Unlike prior methods requiring task boundaries, target supervision, or experience replay, FADE operates online with fixed memory and no access to target labels. Evaluated on seven benchmarks spanning vision, language, and tabular data, FADE achieves up to 19\% higher accuracy under severe shifts, outperforming methods such as TENT and DIW. FADE also generalizes naturally to federated learning by treating heterogeneous clients as temporally fragmented environments, enabling scalable and stable adaptation in decentralized settings. Theoretical analysis guarantees bounded regret and parameter consistency, while empirical results demonstrate FADE's robustness across modalities and shift intensities.

Figures

Figures reproduced from arXiv: 2507.18996 by the authors.

Figure 1
Figure 1. Sequential Covariate Shift (SCS): Model performance drops as input distributions evolve over training [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Client-wise heterogeneity in federated learning causes model degradation. Each symbol denotes a different [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. SCS detection pipeline: (a) KL divergence spikes represent potential distribution shift, (b) FIM variations [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Workflow of the proposed FADE framework: FADE extends to Federated Learning by modeling client-wise [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FADE forgetting performance in comparison to benchmarking methods. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 30 canonical work pages

  1. [1]

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

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

    work page 2000

  2. [2]

    Covariate shift adaptation by importance weighted cross validation.Journal of Machine Learning Research, 8(5), 2007

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

    work page 2007

  3. [3]

    Causal covariate shift correction using fisher information penalty

    Behraj Khan, Behroz Mirza, and Tahir Syed. Causal covariate shift correction using fisher information penalty. In The Second Tiny Papers Track at ICLR 2024

    work page 2024

  4. [4]

    Fl games: A federated learning framework for distribution shifts.arXiv preprint arXiv:2205.11101, 2022

    Sharut Gupta, Kartik Ahuja, Mohammad Havaei, Niladri Chatterjee, and Yoshua Bengio. Fl games: A federated learning framework for distribution shifts.arXiv preprint arXiv:2205.11101, 2022

    arXiv 2022

  5. [5]

    Oxford university press, 1995

    Christopher M Bishop et al.Neural networks for pattern recognition. Oxford university press, 1995

    work page 1995

  6. [6]

    Fast algorithm selection using learning curves

    Jan N van Rijn, Salisu Mamman Abdulrahman, Pavel Brazdil, and Joaquin Vanschoren. Fast algorithm selection using learning curves. InAdvances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22-24, 2015. Proceedings 14, pages 298–309. Springer, 2015

    work page 2015

  7. [7]

    Sample selection bias correction theory

    Corinna Cortes, Mehryar Mohri, Michael Riley, and Afshin Rostamizadeh. Sample selection bias correction theory. InAlgorithmic Learning Theory: 19th International Conference, ALT 2008, Budapest, Hungary, October 13-16, 2008. Proceedings 19, pages 38–53. Springer, 2008

    work page 2008

  8. [8]

    The impact of changing populations on classifier performance

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

    work page 1999

Show all 42 references
  1. [9]

    Classifier technology and the illusion of progress

    David J Hand. Classifier technology and the illusion of progress. 2006

    2006

  2. [10]

    A framework for monitoring classifiers’ performance: when and why failure occurs?Knowledge and Information Systems, 18(1):83–108, 2009

    David A Cieslak and Nitesh V Chawla. A framework for monitoring classifiers’ performance: when and why failure occurs?Knowledge and Information Systems, 18(1):83–108, 2009

    2009

  3. [11]

    A unifying view on dataset shift in classification.Pattern recognition, 45(1):521–530, 2012

    Jose G Moreno-Torres, Troy Raeder, Roc´ıo Alaiz-Rodr´ıguez, Nitesh V Chawla, and Francisco Herrera. A unifying view on dataset shift in classification.Pattern recognition, 45(1):521–530, 2012

    2012

  4. [12]

    Mit Press, 2009

    Joaquin Qui˜nonero-Candela, Masashi Sugiyama, Neil D Lawrence, and Anton Schwaighofer.Dataset shift in machine learning. Mit Press, 2009

    2009

  5. [13]

    Eeg-based emotion recognition using deep learning network with principal component based covariate shift adaptation.The Scientific World Journal, 2014, 2014

    Suwicha Jirayucharoensak, Setha Pan-Ngum, and Pasin Israsena. Eeg-based emotion recognition using deep learning network with principal component based covariate shift adaptation.The Scientific World Journal, 2014, 2014

    2014

  6. [14]

    A weighted support vector machine for data classification.International Journal of Pattern Recognition and Artificial Intelligence, 21(05):961–976, 2007

    Xulei Yang, Qing Song, and Yue Wang. A weighted support vector machine for data classification.International Journal of Pattern Recognition and Artificial Intelligence, 21(05):961–976, 2007

    2007

  7. [15]

    Application of covariate shift adaptation techniques in brain–computer interfaces.IEEE Transactions on Biomedical Engineering, 57(6):1318–1324, 2010

    Yan Li, Hiroyuki Kambara, Yasuharu Koike, and Masashi Sugiyama. Application of covariate shift adaptation techniques in brain–computer interfaces.IEEE Transactions on Biomedical Engineering, 57(6):1318–1324, 2010

    2010

  8. [16]

    Dirichlet-enhanced spam filtering based on biased samples.Advances in neural information processing systems, 19, 2006

    Steffen Bickel and Tobias Scheffer. Dirichlet-enhanced spam filtering based on biased samples.Advances in neural information processing systems, 19, 2006

    2006

  9. [17]

    Failing loudly: An empirical study of methods for detecting dataset shift.Advances in Neural Information Processing Systems, 32, 2019

    Stephan Rabanser, Stephan G¨unnemann, and Zachary Lipton. Failing loudly: An empirical study of methods for detecting dataset shift.Advances in Neural Information Processing Systems, 32, 2019. 13 FADE:Fisher-based Adaptation to Dynamic EnvironmentsA PREPRINT

    2019

  10. [18]

    A distribution-free test of covariate shift using conformal prediction.arXiv preprint arXiv:2010.07147, 2020

    Xiaoyu Hu and Jing Lei. A distribution-free test of covariate shift using conformal prediction.arXiv preprint arXiv:2010.07147, 2020

    2010 arXiv

  11. [19]

    Study on the impact of partition-induced dataset shift onk-fold cross-validation.IEEE Transactions on Neural Networks and Learning Systems, 23(8):1304–1312, 2012

    Jose Garc´ıa Moreno-Torres, Jos´e A S´aez, and Francisco Herrera. Study on the impact of partition-induced dataset shift onk-fold cross-validation.IEEE Transactions on Neural Networks and Learning Systems, 23(8):1304–1312, 2012

    2012

  12. [20]

    Testing for concept shift online.arXiv preprint arXiv:2012.14246, 2020

    Vladimir V ovk. Testing for concept shift online.arXiv preprint arXiv:2012.14246, 2020

    2012 arXiv

  13. [21]

    Testing randomness online.Statistical Science, 36(4):595–611, 2021

    Vladimir V ovk. Testing randomness online.Statistical Science, 36(4):595–611, 2021

    2021

  14. [22]

    Cam- bridge University Press, 2012

    Masashi Sugiyama, Taiji Suzuki, and Takafumi Kanamori.Density ratio estimation in machine learning. Cam- bridge University Press, 2012

    2012

  15. [23]

    Scaffold: Stochastic controlled averaging for federated learning

    Sai Praneeth Karimireddy, Satyen Kale, Mehryar Mohri, Sashank Reddi, Sebastian Stich, and Ananda Theertha Suresh. Scaffold: Stochastic controlled averaging for federated learning. InInternational Conference on Machine Learning, pages 5132–5143. PMLR, 2020

    2020

  16. [24]

    Federated optimization in heterogeneous networks.Proceedings of Machine learning and systems, 2:429–450, 2020

    Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, and Virginia Smith. Federated optimization in heterogeneous networks.Proceedings of Machine learning and systems, 2:429–450, 2020

    2020

  17. [25]

    Big data: Principles and best practices of scalable realtime data systems

    N Marz and J Warren. Big data: Principles and best practices of scalable realtime data systems. 2015.Citado, 6:9–23, 2013

    2015

  18. [26]

    Spark: Cluster computing with working sets.HotCloud, 10(10-10):95, 2010

    Matei Zaharia, Mosharaf Chowdhury, Michael J Franklin, Scott Shenker, Ion Stoica, et al. Spark: Cluster computing with working sets.HotCloud, 10(10-10):95, 2010

    2010

  19. [27]

    Mahout in action: Manning shelter island

    S Owen, R Anil, T Dunning, and E Friedman. Mahout in action: Manning shelter island. 2011

    2011

  20. [28]

    Domain-adversarial training of neural networks.Journal of machine learning research, 17(59):1–35, 2016

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, Franc ¸ois Laviolette, Mario March, and Victor Lempitsky. Domain-adversarial training of neural networks.Journal of machine learning research, 17(59):1–35, 2016

    2016

  21. [29]

    Communication- efficient learning of deep networks from decentralized data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. Communication- efficient learning of deep networks from decentralized data. InArtificial intelligence and statistics, pages 1273–1282. PMLR, 2017

    2017

  22. [30]

    These studies, however, do not belong to history yet

    Vladimir Vapnik.Statistical Learning Theory now plays a more active role: after the general analysis of learning processes, the research in the area of synthesis of optimal algorithms was started. These studies, however, do not belong to history yet. They are a subject of toda...

    1998

  23. [31]

    Relative density- ratio estimation for robust distribution comparison.Neural computation, 25(5):1324–1370, 2013

    Makoto Yamada, Taiji Suzuki, Takafumi Kanamori, Hirotaka Hachiya, and Masashi Sugiyama. Relative density- ratio estimation for robust distribution comparison.Neural computation, 25(5):1324–1370, 2013

    2013

  24. [32]

    Rethinking importance weighting for deep learning under distribution shift.Advances in neural information processing systems, 33:11996–12007, 2020

    Tongtong Fang, Nan Lu, Gang Niu, and Masashi Sugiyama. Rethinking importance weighting for deep learning under distribution shift.Advances in neural information processing systems, 33:11996–12007, 2020

    2020

  25. [33]

    Overcoming catastrophic forgetting in neural networks.Proceedings of the national academy of sciences, 114(13):3521–3526, 2017

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. Overcoming catastrophic forgetting in neural networks.Proceedings of the national academy of scienc...

    2017

  26. [34]

    Tent: Fully test-time adaptation by entropy minimization.arXiv preprint arXiv:2006.10726, 2020

    Dequan Wang, Evan Shelhamer, Shaoteng Liu, Bruno Olshausen, and Trevor Darrell. Tent: Fully test-time adaptation by entropy minimization.arXiv preprint arXiv:2006.10726, 2020

    2006 arXiv

  27. [35]

    Online convex programming and generalized infinitesimal gradient ascent

    Martin Zinkevich. Online convex programming and generalized infinitesimal gradient ascent. InProceedings of the 20th international conference on machine learning (icml-03), pages 928–936, 2003

    2003

  28. [36]

    The mnist database of handwritten digits.http://yann

    Yann LeCun. The mnist database of handwritten digits.http://yann. lecun. com/exdb/mnist/, 1998

    1998

  29. [37]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  30. [38]

    Bridging language and items for retrieval and recommendation.arXiv preprint arXiv:2403.03952, 2024

    Yupeng Hou, Jiacheng Li, Zhankui He, An Yan, Xiusi Chen, and Julian McAuley. Bridging language and items for retrieval and recommendation.arXiv preprint arXiv:2403.03952, 2024

    2024 arXiv

  31. [39]

    Credit-card-fraud detection-imbalanced-dataset.Kaggle

    S Yadav. Credit-card-fraud detection-imbalanced-dataset.Kaggle. Available online: https://www. kaggle. com/datasets/dark0 6thunder/credit-card-dataset (accessed on 29 July 2023)

    2023

  32. [40]

    Openml benchmarking suites.arXiv preprint arXiv:1708.03731, 2017

    Bernd Bischl, Giuseppe Casalicchio, Matthias Feurer, Pieter Gijsbers, Frank Hutter, Michel Lang, Rafael G Man- tovani, Jan N van Rijn, and Joaquin Vanschoren. Openml benchmarking suites.arXiv preprint arXiv:1708.03731, 2017

    2017 arXiv

  33. [41]

    Leaf: A benchmark for federated settings.arXiv preprint arXiv:1812.01097, 2018

    Sebastian Caldas, Sai Meher Karthik Duddu, Peter Wu, Tian Li, Jakub Koneˇcn`y, H Brendan McMahan, Virginia Smith, and Ameet Talwalkar. Leaf: A benchmark for federated settings.arXiv preprint arXiv:1812.01097, 2018. 14 FADE:Fisher-based Adaptation to Dynamic EnvironmentsA PREPRINT

    2018 arXiv

  34. [42]

    Reading digits in natural images with unsupervised feature learning

    Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, Baolin Wu, Andrew Y Ng, et al. Reading digits in natural images with unsupervised feature learning. InNIPS workshop on deep learning and unsupervised feature learning, volume 2011, page 7. Granada, 2011. 15

    2011

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.