Navigating Data Corruption in Machine Learning: Balancing Quality, Quantity, and Imputation Strategies

Qi Liu; Wanjing Ma

REVIEW 4 major objections 5 minor 1 cited by

Navigating Data Corruption in Machine Learning: Balancing Quality, Quantity, and Imputation Strategies

T0 review · 4 major / 5 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 Model performance under data corruption follows a single exponential diminishing-return curve across NLP and reinforcement learning, this paper argues.

desk verdict Two-task empirical study overclaims a universal exponential law; the derivation has a false algebraic step and the flagship fit contradicts its own baseline table. read the letter →

arxiv 2412.18296 v2 pith:6EVA74DG submitted 2024-12-24 cs.LG cs.AI

Qi Liu , Wanjing Ma This is my paper · ORCID

classification cs.LGcs.AI

keywords datacorruptionmissingnoisyimputationdiminishingreturnsexponentialmodelreinforcementlearningNLPsupervised

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

Data corruption—missing entries or added noise—degrades model performance along a universal diminishing-return curve, the paper argues. Across two very different setups, fine-tuning BERT on GLUE tasks (NLP-SL) and training a deep Q-network for traffic signal control (Signal-RL), the fitted relation $S = a(1 - e^{-\lambda(1-p)})$ holds with $R^2$ values of 0.995 and 0.956, where $p$ is the corruption ratio. The paper claims that noise is consistently more harmful than missing data, that imputation helps only when the missing ratio is high and the imputation itself is accurate, and that enlarging the dataset cannot fully recover lost performance. A sympathetic reader would take this as evidence for a quantitative, two-parameter law linking data quality to model utility, with implications for when to clean data versus when to collect more.

What carries the argument

The central object is the exponential decay law $S = a(1 - e^{-\lambda(1-p)})$, which the paper derives from a Poisson rare-event argument. Each useful pattern is assumed to appear with rate $\lambda$; the probability that a pattern is never recovered under corruption level $p$ is $e^{-\lambda(1-p)}$, and the model score is taken as proportional to the number of recovered patterns. This yields the differential equation $dS/dx = a\lambda e^{-\lambda x}$ with $x = 1-p$, whose solution is the exponential CDF. A second piece of machinery is the imputation advantage $A(p,q) = \tilde{S}(p,q) - S(p)$, the score difference with and without imputation at missing ratio $p$ and imputation noise $q$; its zero contour defines the decision boundary separating beneficial from harmful imputation.

What would settle it

Train one model on the same task under two corruption mechanisms at the same ratio $p$—random token masking versus adversarial masking of the most informative tokens. If the fitted $\lambda$ differs substantially or the curve departs from $S = a(1 - e^{-\lambda(1-p)})$, the claimed universal law is contradicted. Equivalently, measure the number of recovered patterns directly (for instance with probing tasks) and check whether model score is linear in that count.

Watch

Extended reading notes

Core claim

The paper's central claim is that, for any model trained on corrupted data, performance follows $S = a(1 - e^{-\lambda(1-p)})$ with $p$ the corruption ratio and $\lambda$ a task-specific decay rate. Fitting this to NLP-SL gives $a = 0.475$, $\lambda = 3.517$, $R^2 = 0.995$; fitting to Signal-RL gives $a = 395.8$, $\lambda = 7.493$, $R^2 = 0.956$. The larger $\lambda$ for the RL task is interpreted as higher sensitivity to corruption, and the paper classifies tasks as noise-sensitive (exponential decision boundary on an imputation-advantage heatmap, RL-like) or noise-insensitive (logistic boundary, NLP-like). It also claims that noisy data causes more severe and unstable degradation than missing data, that imputation is advantageous only in a region of high missing ratio and low imputation noise (the 'imputation advantageous corner') and harmful elsewhere, and that increasing dataset size yields exponentially diminishing returns so that corruption cannot be fully bought off with more data. A further empirical observation is that in traffic signal control roughly the closest 30% of the detection range determines performance, while the remaining 70% can be masked with little effect.

Load-bearing premise

The exponential law rests on the assumption that model score is exactly proportional to the number of independently recovered patterns, each with an equal corruption probability; if that link breaks, the curve is only an empirical fit, not a derived law.

Editorial extensions

If this is right

With two fitted parameters ($a$ and $\lambda$), practitioners can predict the performance loss for any corruption ratio without retraining the model.
In noise-sensitive tasks such as reinforcement learning, even a small amount of injected noise can push performance below a fixed-timing baseline, so preventing noise matters more than adding data.
Imputation should be applied selectively: it pays off only when the missing ratio is high and the imputation error is low; otherwise it can make performance worse.
To offset a given corruption level, the required training data grows roughly exponentially, so cleaning the critical fraction of data is often cheaper than expanding the dataset.
In traffic signal control, focusing detection and data collection on the nearest 30% of each lane preserves most of the learning signal.

Reading between the lines

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

The paper does not test whether the exponential law holds when corruption is correlated or structured (e.g., entire regions masked); an extension would be to fit the same curve under adversarial or clustered corruption and see whether $\lambda$ remains constant.
The proposed 'noise-sensitive vs noise-insensitive' dichotomy could be recast as a property of the loss landscape or of how errors compound over time; if so, the same heatmap analysis could transfer to transformer-based RL or model-based planners.
A direct test of the proportionality assumption—decoding whether model score is truly linear in the number of recovered patterns (e.g., via probing classifiers)—would either confirm the derivation or reveal that the exponential fit is purely empirical.
The 30%-critical-data rule may be a special case of a more general feature-importance law; one could test whether the critical fraction shrinks as state features become more redundant, for example in image-based tasks.

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

Quick take: this is an honest two-task empirical study that overreaches into a "universal rule." The experiments look real, the code is out, and the imputation-advantage heatmap is a genuinely useful way to frame the missing-data/imputation-noise trade-off. But the central claim—Equation 2 as a law—is not supported. The derivation in §3.3 is circular and contains a bad algebra step: the text says λe^{-λx} = λ(a−S), which is off by a factor of a; the correct identity is aλe^{-λx} = λ(a−S). That error doesn't just polish the math, it breaks the 'explanation' of why the exponential should hold.

More seriously, the flagship Signal-RL fit is internally inconsistent with the paper's own Table 3. With a=395.8 and λ=7.493, Equation 2 gives S(0) ≈ 395.6, but Table 3 reports the p=0 mean as 409.86. The formula a = S0/(1−e^{-λ}) would require a ≈ 410.1, not 395.8. So the one example meant to demonstrate the universal law fails on its own numbers. That's a load-bearing error; any referee would catch it.

What else? The law is fit on two tasks, 20 and 10 points respectively, with no error bars on parameters, no model comparison, no held-out validation. A saturating exponential will fit many monotone sequences, so R²=0.995 on one curve and 0.956 on another is not evidence of a mechanism. The 'striking coincidence' is not tested against anything.

The paper does get some things right. The imputation advantage framework (A(p,q)) and the decision-boundary view are pragmatic and could be useful to practitioners. The finding that noisy data is worse than missing data in these settings, with RL more fragile, is plausible and consistent with prior work. The '30% critical data' is explicitly limited to the traffic task, which is appropriately hedged.

Net: the empirical skeleton is worth a referee's time—there is real data, code, and a testable idea. But the current version overclaims, and the internal inconsistency needs to be fixed. I'd invite revision with a request to temper the 'universal law' language, correct the derivation, and add out-of-sample or at least bootstrap validation. A serious editor shouldn't desk-reject it; it should go to review, with the expectation of major revision.

Referee Report

4 major / 5 minor

Summary. The paper studies how missing and noisy data affect model performance in two very different learning setups: a BERT-based supervised NLP pipeline on GLUE tasks (NLP-SL) and a DQN-based traffic signal control task (Signal-RL). It proposes that performance under corruption follows a universal exponential law S = a(1 - e^{-\lambda(1-p)}) (Eq. 2), examines imputation trade-offs through heatmaps and decision boundaries, and investigates whether enlarging datasets can offset corruption. The main claims are that noise is more harmful than missing data, that imputation is useful only in certain regimes, that larger datasets give diminishing returns, and that roughly 30% of the data is critical in the traffic signal task.

Significance. If the exponential law were established, it would give practitioners a quantitative rule for predicting performance loss under corruption and for deciding when imputation or additional data collection is worthwhile. The paper has concrete strengths: it uses two genuinely different learning paradigms, reports model scores in Tables 2 and 3, releases code for reproducibility, and makes a number of practically relevant qualitative observations about noise sensitivity and imputation. However, the central universal-law claim is not currently supported. The derivation in Section 3.3 is circular and algebraically flawed, the flagship Signal-RL fit is internally inconsistent with the paper's own baseline table, and the law is never tested out of sample or against alternative functional forms. As it stands, the paper is best read as a two-task empirical study with suggestive observations, not as a demonstration of a universal rule.

major comments (4)

[Section 3.3, Eq. (4)] The 'explanation' of Eq. (2) is not a derivation. The text assumes that S is proportional to the number of identified patterns and that the probability of failing to recover a pattern is e^{-λx}; the rate of S is then declared proportional to λe^{-λx}, which integrates directly to Eq. (2). Because λ is fitted from the same curves that Eq. (2) is supposed to explain, the Poisson argument is a restatement of the fit rather than an independent mechanism. Moreover, the identity 'λe^{-λx} = λ(a − S)' is algebraically false; from Eq. (2) the correct identity is aλe^{-λx} = λ(a − S). This needs to be corrected, and the explanatory claim should be substantially weakened or replaced by an independent test.
[Section 3.2, Table 3 and Eq. (2)] The reported Signal-RL fit is internally inconsistent with the paper's own baseline. With a = 395.8 and λ = 7.493, Eq. (2) gives S(0) = 395.8(1 − e^{−7.493}) ≈ 395.6, while Table 3 reports the p = 0 mean score as 409.86 with standard deviation 3.83; the defining relation a = S0/(1 − e^{−λ}) would require a ≈ 410.1, not 395.8. The NLP-SL fit shows a smaller but similar discrepancy (a = 0.475 predicts S(0) ≈ 0.461 versus 0.4669 in Table 2). Because Eq. (2) is the paper's central quantitative law, the fitted parameters and Tables 2 and 3 must be reconciled; as printed, at least one of them is wrong.
[Section 3.2, Figures 3-4] No out-of-sample validation or model-selection evidence supports the 'universal rule' claim. The exponential is fit and evaluated on the same two curves (20 NLP points and 10 RL points, with RL standard deviations growing to about 47.6 at p = 0.9). A two-parameter saturating curve will fit many monotone sequences, and R² = 0.956 on ten noisy points does not discriminate the exponential from other concave alternatives. The paper should report parameter uncertainties, compare against at least a power-law or logistic alternative, and ideally test on held-out tasks or domains before claiming a universal law.
[Section 5.2, Figure 11 and Conclusions] The claim that 'the number of samples required increases exponentially with the data corruption level' is not supported by any fitted exponential curve, error bar, or statistical test in the paper. Figure 11 shows raw performance curves and a dashed benchmark line, but no quantitative relation between corruption level and required sample size is derived. This conclusion should be either derived explicitly or stated as a qualitative observation.

minor comments (5)

[Section 3.3, Eq. (4)] The caption to Eq. (4) says 'b is pattern appearance rate', but no b appears in the equation; the intended symbol is presumably λ.
[Section 5.2] The sentence 'for ex = 1 + x + ...' should be 'for e^x = 1 + x + ...'; the Taylor-series argument by itself does not establish exponential growth of required data.
[Table A1] The text says eight GLUE tasks are used, but the finetuning sequence in Table A1 lists nine entries including WNLI; the list should be reconciled.
[Section 4.2 and Figure 7] The fitted decision boundaries (logistic for NLP-SL, exponential for Signal-RL) are described qualitatively, but the fitting procedure, parameter values, and goodness-of-fit are not reported, and the 68%/95% confidence bands are mentioned only in the figure caption.
[Overall] There are numerous typos and inconsistent terms: 'model socre', 'noise-insensitvie', 'impuation', 'missing ratio' versus 'corruption ratio', and section headings with stray spaces such as 'T rade-offs'. A careful copyedit would improve readability.

Circularity Check

2 steps flagged · score 6.0 of 10

The universal exponential law's derivation restates the fitted curve: Eq. 4 is the derivative of Eq. 2, and lambda is the fitted decay rate renamed as a pattern appearance rate.

self definitional [Section 3.3, Equation 4 (explanation of Equation 2)]
"Suppose that model socre S is proportional to the number of patterns identified. Then, as x increases, the rate of S is proportional to λe−λx. This leads to Eqn 4 where a is coefficient for this linear relation. Its solution corresponds to Eqn 2. It can be shown that λe−λx = λ(a − S), so Eqn 4 actually describes a dynamic system where the rate of change in performance depends on the difference between the system’s current performance and its limit."

Equation 4 is dS/dx = aλe^{−λx}, which is exactly the derivative of the target curve Equation 2, S = a(1−e^{−λx}), with S(0)=0. Thus 'its solution corresponds to Eqn 2' is a mathematical identity, not an independent derivation. The intermediate assumption that 'the rate of S is proportional to λe^{−λx}' postulates the same exponential kernel that was fitted to the data; it assumes the exponential CDF rather than deriving it from an independent mechanism. The supporting identity 'λe^{−λx} = λ(a−S)' is also algebraically false for a≠1, since the correct relation is λ(a−S)=aλe^{−λx}. The Section 3.3 explanation therefore reduces to restating the fitted exponential in Poisson vocabulary.
fitted input called prediction [Section 3.2 (fitted parameters) and Section 3.3 (pattern appearance rate)]
"The fitted parameters for the NLP-SL experiment are: a = 0.475, λ = 3.517, where λ represents the decay rate that controls the curve’s steepness. [...] Similarly, for the Signal-RL experiment, the fitted curve parameters are: a = 395.8, λ = 7.493. [...] The probability of failing to recover such a pattern with corruption level p is e−λx, where λ is the pattern appearance rate given no corruption."

The λ used in the Poisson 'explanation' is not measured independently from pattern counts or held-out data; it is the decay constant obtained by fitting Equation 2 to the same model scores that the explanation is supposed to account for. Renaming this fitted constant 'the pattern appearance rate' does not add information. The conclusion that 'RL tasks are more sensitive to data corruption' is likewise a comparison of two fitted λ values, not a prediction from an independent mechanism. No out-of-sample task or independent estimate of pattern appearance is provided, so the explanatory mechanism carries no evidential content beyond the fit.

full rationale

I identify two linked circular steps. First, Section 3.3's differential equation is the derivative of the target equation, so solving it returns Equation 2 by construction; the Poisson preamble does not supply independent content because it already posits an e^{−λx} failure probability, which is the same exponential form that was fitted to the data. Second, λ is fitted to the same data and then relabelled as 'pattern appearance rate'; the claimed sensitivity difference between RL and NLP tasks is a comparison of two fitted constants rather than an out-of-sample prediction. There is no self-citation chain, no imported uniqueness theorem, and the experimental measurements themselves appear genuine. However, the central 'universal rule' is not validated against held-out tasks, alternative functional forms, or a null model. In addition, the reported Signal-RL fit is internally inconsistent with the paper's own baseline: with a=395.8 and λ=7.493, Equation 2 gives S0≈395.6 at p=0, while Table 3 reports the p=0 mean score as 409.86; the stated formula a=S0/(1−e^{−λ}) would require a≈410.1. This inconsistency further weakens the claim but is a correctness issue rather than circularity. Because the derivation and the explanatory parameter reduce to the fitted curve, I score 6 rather than 0-2.

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

The paper introduces no new entities. The main burden is on the two fitted parameters (a, λ) and the Poisson/linearity assumptions used to 'explain' the exponential curve, which are not independently validated.

free parameters (4)

a (amplitude of exponential fit) = 0.475 (NLP-SL), 395.8 (Signal-RL)
Fitted to model scores at different corruption levels; defined as a = S0/(1-e^{-λ}), essentially rescaling the curve to pass through the clean-data score.
λ (decay rate) = 3.517 (NLP-SL), 7.493 (Signal-RL)
Fitted to model scores; interpreted as 'pattern appearance rate' but not measured independently. Governs steepness of the exponential curve.
Imputation noise level q = varied 0 to 1
Controlled parameter for artificial imputation method, not fitted, but used to evaluate trade-offs. Not a free parameter in the sense of fitting, but it is a tunable knob.
Subset ratio for BERT imputation = 0.1
Chosen for computational speed, not justified, and affects the BERT imputation results.

assumptions (3)

domain assumption Pattern discovery is a rare event, so the binomial distribution approximates Poisson (Section 3.3).
The derivation of Equation 2 assumes that each pattern has small probability of being observed per sample and the number of patterns is large, which is not verified for the actual datasets.
ad hoc to paper Model score S is proportional to the number of identified patterns.
This linearity assumption is stated without evidence and is needed to get the exponential differential equation. It is plausible but not derived.
domain assumption The corruption process affects each pattern independently with equal probability p.
The experiments corrupt words or state elements independently, but patterns in language or traffic may be correlated. The paper acknowledges this in future work but uses independence in the model.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Navigating Data Corruption in Machine Learning: Balancing Quality, Quantity, and Imputation Strategies." pith.science (2026). https://pith.science/paper/6EVA74DG

@misc{pith2026241218296,
  author       = {Pith},
  title        = {Pith review of: Navigating Data Corruption in Machine Learning: Balancing Quality, Quantity, and Imputation Strategies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6EVA74DG}},
  note         = {Machine review of arXiv:2412.18296}
}

read the original abstract

Data corruption, including missing and noisy data, poses significant challenges in real-world machine learning. This study investigates the effects of data corruption on model performance and explores strategies to mitigate these effects through two experimental setups: supervised learning with NLP tasks (NLP-SL) and deep reinforcement learning for traffic signal optimization (Signal-RL). We analyze the relationship between data corruption levels and model performance, evaluate the effectiveness of data imputation methods, and assess the utility of enlarging datasets to address data corruption. Our results show that model performance under data corruption follows a diminishing return curve, modeled by the exponential function. Missing data, while detrimental, is less harmful than noisy data, which causes severe performance degradation and training instability, particularly in sequential decision-making tasks like Signal-RL. Imputation strategies involve a trade-off: they recover missing information but may introduce noise. Their effectiveness depends on imputation accuracy and corruption ratio. We identify distinct regions in the imputation advantage heatmap, including an "imputation advantageous corner" and an "imputation disadvantageous edge" and classify tasks as "noise-sensitive" or "noise-insensitive" based on their decision boundaries. Furthermore, we find that increasing dataset size mitigates but cannot fully overcome the effects of data corruption. The marginal utility of additional data diminishes as corruption increases. An empirical rule emerges: approximately 30% of the data is critical for determining performance, while the remaining 70% has minimal impact. These findings provide actionable insights into data preprocessing, imputation strategies, and data collection practices, guiding the development of robust machine learning systems in noisy environments.

Figures

Figures reproduced from arXiv: 2412.18296 by the authors.

**Figure 2.** Model convergence in two experiments. (a) NLP-SL experiment: The [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** Model performance across varying data missing ratios. The [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Figures from the paper (8 more)

**Figure 4.** Figure 4: Relationship between model performance and data corruption ratio, fitted by Equation [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

**Figure 5.** Figure 5: Signal-RL model performance under varying data corruption ratios. (a) Noise insertion: [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

**Figure 6.** Figure 6: Training instability under noise-insertion corruption. The [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

**Figure 7.** Figure 7: Heatmap of imputation advantage. The x-axis denotes the data missing ratio p, and the y-axis represents the imputation noise level q. The black dashed line indicates the decision boundary separating regions where imputation is beneficial from those where it is detrimen…

**Figure 8.** Figure 8: Illustration of imputation advantage pattern. The [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]

**Figure 9.** Figure 9: Illustration of context-filling imputation in the Signal-RL experiment. This represents [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]

**Figure 10.** Figure 10: Effectiveness of alternative imputation methods. The [PITH_FULL_IMAGE:figures/full_fig_p018_10.png]

**Figure 11.** Figure 11: Effectiveness of enlarging training dataset. (a) NLP-SL. The [PITH_FULL_IMAGE:figures/full_fig_p019_11.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Theory Foundation of Physics-Enhanced Residual Learning
cs.LG 2025-08 reject novelty 4.0 of 10

A set of conditional bounds shows PERL's advantages follow from assumed smaller Lipschitz constant and loss ceiling, without proving those assumptions or connecting them correctly to neural network training.

Reference graph

Works this paper leans on

43 extracted references · 34 canonical work pages · cited by 1 Pith paper

[1]

K.; Stirling, W

Moon, T. K.; Stirling, W. C. Mathematical Methods and Algorithms for Signal Processing ; Prentice Hall: Upper Saddle River, NJ, 2000; ISBN 0-201-36186-8

work page 2000
[2]

M.; Nasrabadi, N

Bishop, C. M.; Nasrabadi, N. M. Pattern Recognition and Machine Learning ; Springer: New York, 2006

work page 2006
[3]

Zhou, Y.; Aryal, S.; Bouadjenek, M. R. Review for Handling Missing Data with Special Missing Mechanism. arXiv 2024, arXiv:2404.04905

arXiv 2024
[4]

A Survey on Missing Data in Machine Learning

Emmanuel, T.; Maupong, T.; Mpoeleng, D.; Semong, T.; Mphago, B.; Tabona, O. A Survey on Missing Data in Machine Learning. J. Big Data 2021, 8, 1–37

work page 2021
[5]

Compression of GNSS Data with the Aim of Speeding Up Communication to Autonomous Vehicles

Rakhmanov, A.; Wiseman, Y. Compression of GNSS Data with the Aim of Speeding Up Communication to Autonomous Vehicles. Remote Sens. 2023, 15, 2165

work page 2023
[6]

An Introduction to Probability Theory and Its Applications , 3rd ed., Vol

Feller, W. An Introduction to Probability Theory and Its Applications , 3rd ed., Vol. 1; Wiley: New York, 1991

work page 1991
[7]

Rubin, D. B. Inference and Missing Data. Biometrika 1976, 63, 581–592

work page 1976
[8]

Little, R. J. A.; Rubin, D. B. Statistical Analysis with Missing Data, 3rd ed.; Wiley: Hoboken, NJ, 2019

work page 2019

Show all 43 references

[9]

L.; Graham, J

Schafer, J. L.; Graham, J. W. Missing Data: Our View of the State of the Art. Psychol. Methods 2002, 7, 147–177

2002
[10]

Little, R. J. A. Missing-Data Adjustments in Large Surveys. J. Bus. Econ. Stat. 1988, 6, 287–296

1988
[11]

Rubin, D. B. Multiple Imputation for Nonresponse in Surveys ; Wiley: New York, 1987

1987
[12]

Enders, C. K. A Primer on Maximum Likelihood Algorithms Available for Use with Missing Data. Struct. Equ. Modeling 2001, 8, 128–141

2001
[13]

P.; Laird, N

Dempster, A. P.; Laird, N. M.; Rubin, D. B. Maximum Likelihood from Incomplete Data via the EM Algorithm. J. R. Stat. Soc. Series B 1977, 39, 1–22

1977
[14]

Troyanskaya, O.; Cantor, M.; Sherlock, G.; Brown, P.; Hastie, T.; Tibshirani, R.; Botstein, D.; Altman, R. B. Missing Value Estimation Methods for DNA Microarrays.Bioinformatics 2001, 17, 520–525

2001
[15]

H.; Olshen, R

Breiman, L.; Friedman, J. H.; Olshen, R. A.; Stone, C. J. Classification and Regression Trees; Wadsworth International Group: Belmont, CA, 1984

1984
[16]

Random Forests

Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. 24

2001
[17]

Extracting and Composing Robust Features with Denoising Autoencoders

Vincent, P.; Larochelle, H.; Bengio, Y.; Manzagol, P.-A. Extracting and Composing Robust Features with Denoising Autoencoders. In Proceedings of the 25th International Conference on Machine Learning ; 2008; pp. 1096–1103

2008
[18]

Generative Adversarial Nets

Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; Bengio, Y. Generative Adversarial Nets. In Advances in Neural Information Processing Systems; 2014; Vol. 27, pp. 2672–2680

2014
[19]

EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks

Wei, J.; Zou, K. EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks. arXiv preprint arXiv:1901.11196 2019

1901 arXiv
[20]

A Survey of Data Augmen- tation Approaches for NLP

Feng, S.; Gangal, V.; Wei, J.; Chandar, S.; Reddy, S.; Diab, M. A Survey of Data Augmen- tation Approaches for NLP. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021; 2021; pp. 968–988

2021
[21]

Missing Token Imputation Using Masked Language Models

Yuan, J.; Wang, R.; Zhang, Y. Missing Token Imputation Using Masked Language Models. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing; 2021; pp. 1234–1240

2021
[22]

Imputing Missing Sentences with Generative Adversarial Net- works

Li, Y.; Guo, Y.; Li, D.; Li, Z. Imputing Missing Sentences with Generative Adversarial Net- works. In Proceedings of the AAAI Conference on Artificial Intelligence ; 2020; Vol. 34, pp. 8470–8477

2020
[23]

Missing Data Imputation: Focusing on Single Imputation

Zhang, Z. Missing Data Imputation: Focusing on Single Imputation. Ann. Transl. Med. 2016, 4

2016
[24]

Learning from Noisy Labels with Deep Neural Networks: A Survey

Song, H.; Kim, M.; Park, D.; Shin, Y.; Lee, J.-G. Learning from Noisy Labels with Deep Neural Networks: A Survey. IEEE Trans. Neural Netw. Learn. Syst. 2022, 34, 8135–8153

2022
[25]

Recurrent Neural Networks for Multivariate Time Series with Missing Values

Che, Z.; Purushotham, S.; Cho, K.; Sontag, D.; Liu, Y. Recurrent Neural Networks for Multivariate Time Series with Missing Values. Sci. Rep. 2018, 8, 6085

2018
[26]

GAIN: Missing Data Imputation Using Generative Adver- sarial Nets

Yoon, J.; Jordon, J.; Schaar, M. GAIN: Missing Data Imputation Using Generative Adver- sarial Nets. In International Conference on Machine Learning ; 2018; pp. 5689–5698

2018
[27]

Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels

Han, B.; Yao, Q.; Yu, X.; Niu, G.; Xu, M.; Hu, W.; Tsang, I.; Sugiyama, M. Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels. Adv. Neural Inf. Process. Syst. 2018, 31

2018
[28]

Deep Learning Is Robust to Massive Label Noise

Rolnick, D. Deep Learning Is Robust to Massive Label Noise. arXiv preprint arXiv:1705.10694 2017

2017 arXiv
[29]

J.; Shlens, J.; Szegedy, C

Goodfellow, I. J.; Shlens, J.; Szegedy, C. Explaining and Harnessing Adversarial Examples. arXiv preprint arXiv:1412.6572 2014

2014 arXiv
[30]

BERT: Pre-training of Deep Bidirectional Transformers for Language Understand- ing

Devlin, J. BERT: Pre-training of Deep Bidirectional Transformers for Language Understand- ing. arXiv preprint arXiv:1810.04805 2018

2018 arXiv
[31]

Brown, T. B. Language Models Are Few-shot Learners. arXiv preprint arXiv:2005.14165 2020. 25

2005 arXiv
[32]

M.; Gebru, T.; McMillan-Major, A.; Shmitchell, S

Bender, E. M.; Gebru, T.; McMillan-Major, A.; Shmitchell, S. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency; 2021; pp. 610–623

2021
[33]

SimCSE: Simple Contrastive Learning of Sentence Embeddings

Gao, T.; Yao, X.; Chen, D. SimCSE: Simple Contrastive Learning of Sentence Embeddings. arXiv preprint arXiv:2104.08821 2021

2021 arXiv
[34]

S.; Zettlemoyer, L.; Levy, O

Joshi, M.; Chen, D.; Liu, Y.; Weld, D. S.; Zettlemoyer, L.; Levy, O. SpanBERT: Improving Pre-training by Representing and Predicting Spans. Trans. Assoc. Comput. Linguist. 2020, 8, 64–77

2020
[35]

Deep Recurrent Q-learning for Partially Observable MDPs

Hausknecht, M.; Stone, P. Deep Recurrent Q-learning for Partially Observable MDPs. In2015 AAAI Fall Symposium Series ; 2015

2015
[36]

A Model-based Reinforcement Learning with Adversarial Training for Online Recommendation

Bai, X.; Guan, J.; Wang, H. A Model-based Reinforcement Learning with Adversarial Training for Online Recommendation. Adv. Neural Inf. Process. Syst. 2019, 32

2019
[37]

Reinforcement Learning with Long Short-term Memory

Bakker, B. Reinforcement Learning with Long Short-term Memory. Adv. Neural Inf. Process. Syst. 2001, 14

2001
[38]

A.; Veness, J.; Bellemare, M

Mnih, V.; Kavukcuoglu, K.; Silver, D.; Rusu, A. A.; Veness, J.; Bellemare, M. G.; Graves, A.; Riedmiller, M.; Fidjeland, A. K.; Ostrovski, G.; et al. Human-level Control Through Deep Reinforcement Learning. Nature 2015, 518, 529–533

2015
[39]

A.; Darrell, T

Pathak, D.; Agrawal, P.; Efros, A. A.; Darrell, T. Curiosity-driven Exploration by Self- supervised Prediction. In International Conference on Machine Learning ; 2017; pp. 2778– 2787

2017
[40]

E.; Stone, P

Taylor, M. E.; Stone, P. Transfer Learning for Reinforcement Learning Domains: A Survey. J. Mach. Learn. Res. 2009, 10

2009
[41]

KILT: A Benchmark for Knowledge Intensive Language Tasks

Petroni, F.; Piktus, A.; Fan, A.; Lewis, P.; Yazdani, M.; De Cao, N.; Thorne, J.; Jernite, Y.; Karpukhin, V.; Maillard, J.; et al. KILT: A Benchmark for Knowledge Intensive Language Tasks. arXiv preprint arXiv:2009.02252 2020

2009 arXiv
[42]

RoBERTa: A Robustly Optimized BERT Pretraining Approach

Liu, Y. RoBERTa: A Robustly Optimized BERT Pretraining Approach. arXiv preprint arXiv:1907.11692 2019, 364

1907 arXiv
[43]

X.; Maharjan, S

Tong, W.; Hussain, A.; Bo, W. X.; Maharjan, S. Artificial Intelligence for Vehicle-to- Everything: A Survey. IEEE Access 2019, 7, 10823–10843. 26

2019

Pith tools

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

[1] [1]

K.; Stirling, W

Moon, T. K.; Stirling, W. C. Mathematical Methods and Algorithms for Signal Processing ; Prentice Hall: Upper Saddle River, NJ, 2000; ISBN 0-201-36186-8

work page 2000

[2] [2]

M.; Nasrabadi, N

Bishop, C. M.; Nasrabadi, N. M. Pattern Recognition and Machine Learning ; Springer: New York, 2006

work page 2006

[3] [3]

Zhou, Y.; Aryal, S.; Bouadjenek, M. R. Review for Handling Missing Data with Special Missing Mechanism. arXiv 2024, arXiv:2404.04905

arXiv 2024

[4] [4]

A Survey on Missing Data in Machine Learning

Emmanuel, T.; Maupong, T.; Mpoeleng, D.; Semong, T.; Mphago, B.; Tabona, O. A Survey on Missing Data in Machine Learning. J. Big Data 2021, 8, 1–37

work page 2021

[5] [5]

Compression of GNSS Data with the Aim of Speeding Up Communication to Autonomous Vehicles

Rakhmanov, A.; Wiseman, Y. Compression of GNSS Data with the Aim of Speeding Up Communication to Autonomous Vehicles. Remote Sens. 2023, 15, 2165

work page 2023

[6] [6]

An Introduction to Probability Theory and Its Applications , 3rd ed., Vol

Feller, W. An Introduction to Probability Theory and Its Applications , 3rd ed., Vol. 1; Wiley: New York, 1991

work page 1991

[7] [7]

Rubin, D. B. Inference and Missing Data. Biometrika 1976, 63, 581–592

work page 1976

[8] [8]

Little, R. J. A.; Rubin, D. B. Statistical Analysis with Missing Data, 3rd ed.; Wiley: Hoboken, NJ, 2019

work page 2019