Pith. sign in

REVIEW 4 major objections 4 minor 38 references

Adaptively-weighted Nearest Neighbors for Matrix Completion

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 An adaptively weighted nearest-neighbor method that sets its own weights and neighborhood by convex optimization matches the radius-tuned estimator's error rate without cross-validation.

desk verdict A neat algorithmic idea undermined by a dropped 4σ² term in the main proof; the claimed rates do not follow as written. read the letter →

arxiv 2505.09612 v1 pith:FWM3NOFK submitted 2025-05-14 stat.ML cs.LGmath.STstat.MEstat.TH

Tathagata Sadhukhan , Manit Paul , Raaz Dwivedi This is my paper · ORCID
classification stat.MLcs.LGmath.STstat.MEstat.TH MSC 62G0562G2062D10
keywords matrixcompletionweightednearestneighborsadaptiveweightsbias-variancetrade-offHöldersmoothfactormodelminimaxnonparametricratemissingdatanoisevarianceestimation
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

Row-nearest-neighbor methods for matrix completion have always needed an external choice: the search radius and the weights given to the neighboring rows, usually selected by cross-validation. This paper proposes AWNN, which sets both by minimizing a convex objective that mirrors the bias–variance decomposition of the estimator's row-wise error, with the noise variance estimated internally by a fixed-point iteration. The authors claim two things: AWNN's row-wise mean-squared-error bound is never worse than the unweighted nearest-neighbor estimator's bound, even when that estimator's radius is optimally tuned, and under a Hölder-smooth factor model with missing entries AWNN attains the rate $O(n^{-2\lambda/(d_1+2\lambda)})$ without any hyperparameter selection. If these claims are right, nearest-neighbor matrix completion becomes a turnkey procedure for recommender systems and panel-data counterfactual prediction, where the choice of radius and weights has been the standing obstacle.

What carries the argument

The engine of the argument is Proposition 1's decomposition of the row-wise MSE into a bias term, a weighted average of row distances, and a variance term, the noise level times $\|\mathbf{w}\|_2^2$; this decomposition licenses the convex surrogate $2\log(2m/\delta)\hat{\sigma}^2\|\mathbf{w}\|_2^2 + \sum_{i'}w_{i'}\hat{\rho}_{i',i}$ as the objective to minimize. The geometry comes from the noise-floor-subtracted distance $\rho_{i',i} = (1/m)\sum_j(\theta_{i,j}-\theta_{i',j})^2 - 2\sigma^2$, which lets the optimization recognize the self row in the noiseless limit. Proposition 2 solves the program in closed form as a soft-thresholded weighting over an automatically chosen neighbor set $R(i)$, and the fixed-point update of $\hat{\sigma}^2$ (Algorithm 1) keeps the whole procedure a function of the observed matrix alone. Lemma 1's comparison against the unweighted neighbor estimator is what converts the optimized bound into a parameter-free rate: because AWNN's bound dominates the best radius-tuned bound, the tuned estimator's rate is inherited by AWNN.

What would settle it

One algebraic check in Appendix S.1 settles the central claim: expanding the bias term of Proposition 1 directly, the quantity the proof equates with $2\sum_{i'}w_{i'}\rho_{i',i}$ is in fact $2\sum_{i'}w_{i'}\rho_{i',i} + 4\sigma^2$, so the displayed MSE upper bound omits a positive $4\sigma^2$ term and can be smaller than the true error; if that is so, Lemma 1 and the rate corollaries do not follow from the given proof. A reader can also check empirically whether AWNN's realized row MSE stays at or below that of the cross-validated unweighted RowNN on the Hölder-factor experiments of Section 4 at low signal-to-noise.

Watch

Extended reading notes

Core claim

The paper's central claim is that the bias–variance trade-off in row-nearest-neighbor regression can be resolved by optimization instead of tuning. AWNN estimates the noise variance $\sigma^2$ internally by a fixed-point iteration, computes empirical row distances $\hat{\rho}_{i',i}$, then for each target row solves the convex program $\min_{\mathbf{w}: w_{i'} \ge 0,\ \sum_{i'}w_{i'}=1} 2\log(2m/\delta)\hat{\sigma}^2\|\mathbf{w}\|_2^2 + \sum_{i'}w_{i'}\hat{\rho}_{i',i}$. Proposition 2 gives the closed-form solution: within an automatically selected neighbor set $R(i)$, row $i'$ receives weight $1/|R(i)| - (\hat{\rho}_{i',i}-\bar{\rho}_i)/(4\log(2m/\delta)\sigma^2)$, so closer rows are up-weighted and distant rows are excluded. The authors prove (Lemma 1) that the resulting row-wise mean-squared-error bound is no larger than that of the unweighted nearest-neighbor estimator under any radius, and (Corollary 2) that under a Hölder-smooth factor model with MCAR missingness the row-wise MSE decays as $O(n^{-2\lambda/(d_1+2\lambda)})$ — the same rate as the optimally tuned neighbor rule — with no hyperparameter selected by the user.

Load-bearing premise

Every stated error bound rests on one algebraic step in the proof of Proposition 1 — rewriting the weighted mean of squared row differences in terms of the paper's row-distance measure — that is asserted without derivation; if that step is off by even a constant, the mean-squared-error bound and everything built on it (Lemma 1, Theorem 2, both rate corollaries) no longer follow, and the claimed missing-data rate additionally assumes the unproved negligibility condition (9).

Editorial extensions

If this is right

  • Nearest-neighbor matrix completion no longer needs cross-validation or a grid search over the radius $\eta$: the user supplies the observed matrix and a confidence level, and the weights and neighbor sets come out of the closed-form solution.
  • The guarantees do not require the ground-truth matrix to be low rank or sparse — only that rows and columns follow a Hölder-smooth factor model — so the method is claimed to stay reliable where ERM-based completion degrades, including missing-not-at-random patterns, where the authors state Theorem 2 applies unchanged.
  • In the noiseless limit the self-tuning weights collapse to the optimal extreme: full weight on the target row when self-neighbors are allowed, and on the nearest distinct row otherwise, which is the best any estimator could do without extra information.
  • Under MCAR missingness, AWNN is claimed to match the row-wise MSE rate of the radius-optimized unweighted nearest-neighbor estimator, so automatic weighting costs nothing asymptotically while removing the tuning burden (Corollary 2).
  • Because the weights have a closed form, the only iterative part is a scalar fixed-point update of the noise variance; the experiments report that this internal estimate lands on the true variance, so the fully automatic version matches the oracle version of the algorithm.

Reading between the lines

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

  • The weight formula is a shrinkage rule — uniform $1/K$ minus a distance deviation scaled by the noise level — so the same convex surrogate could plausibly be applied to kernel regression, where bandwidth choice has resisted a parameter-free resolution; that extension is not in the paper.
  • If the internal noise-variance estimate is as accurate as the Section 4 experiments suggest, AWNN recovers the optimal weighting without knowing $\sigma^2$, making it a candidate drop-in replacement for cross-validated neighbor methods in recommender and panel-data pipelines.
  • The analysis is row-symmetric, so the same weight-optimization could be run on columns or on two-sided neighbor sets; transferring the dominance lemma to those estimators is a natural next test.
  • The paper states that Theorem 2 covers missing-not-at-random patterns without modification, but its experiments vary only MCAR missingness; simulating missingness correlated with the signal would test whether the bias condition (9) stays negligible in that regime.
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 manuscript proposes AWNN, a row-nearest-neighbor matrix completion method in which the weights and the effective neighborhood are selected by minimizing a penalized objective that combines a variance proxy with estimated centered row distances, using a plug-in estimate of the noise variance. The paper claims finite-sample data-dependent row-MSE upper bounds (Prop. 1, Thms. 1 and 2), a dominance result over optimally tuned unweighted row nearest neighbors (Lem. 1), and an adaptive rate O(n^{-2\lambda/(d_1+2\lambda)}) under MCAR missingness without hyperparameter tuning (Cors. 1 and 2). The experiments compare AWNN with RowNN, Oracle-AWNN, and USVT on synthetic Hölder factor models.

Significance. If the theoretical claims were valid, AWNN would be a genuinely useful contribution: a tuning-free nearest-neighbor matrix completion method with closed-form weights, no low-rank or sparsity assumptions, and rate-adaptive guarantees. The algorithmic idea is appealing, the weight computation is simple, and the reported synthetic results are encouraging. However, the proof of the central results contains several load-bearing errors, including an incorrect substitution of Eq. (3), a confusion between centered and uncentered distance in the proof of Thm. 2, and an internally inconsistent argument in Lem. 1. As a consequence, the headline rate claims are not established by the arguments given.

major comments (4)
  1. [S.1 (Prop. 1), Eq. (3)] The proof of Prop. 1 substitutes Eq. (3) incorrectly. It writes 2/m \sum_{i'} w_{i'} \sum_j (\theta_{i,j}-\theta_{i',j})^2 = 2 \sum_{i'} w_{i'} \rho_{i',i}, but Eq. (3) defines \rho_{i',i} = (1/m)\sum_j(\theta_{i,j}-\theta_{i',j})^2 - 2\sigma^2, so the correct identity is 2\sum_{i'}w_{i'}\rho_{i',i} + 4\sigma^2, using \sum_{i'}w_{i'}=1. The stated upper bound in Prop. 1 therefore omits a 4\sigma^2 term and is not a valid upper bound; it can even be negative when the estimated centered distances are negative. Since Prop. 1 is invoked in Thm. 1, Lem. 1, Cor. 1, Thm. 2, and Cor. 2, the chain of rate claims is not established. In particular, inserting the missing 4\sigma^2 into the display preceding Cor. 2 and substituting \eta=O(n^{-\lambda/(d_1+2\lambda)}) yields an O(1) rather than O(n^{-2\lambda/(d_1+2\lambda)}) bound. This is not a cosmetic constant: because \hat{\rho}_{i',i} is centered at -2\sigma^2, a small threshold on \hat{\rho} does not by itself control the uncentered signal-distance term appearing in the row MSE.
  2. [S.7 (Thm. 2)] The proof of Thm. 2 defines \rho^*_{i',i}=E[(\theta_{i,j}-\theta_{i',j})^2] and later claims that |\rho_{i',i}-\rho^*_{i',i}| is bounded by the vanishing concentration term 8M^2\sqrt{\log(2/\delta)}(\|\hat{w}_{i',\cdot}\|_2/\|\hat{w}_{i',\cdot}\|_1). But \rho_{i',i} is defined in Eq. (3) as the same average minus 2\sigma^2, i.e. \rho_{i',i}=\rho^*_{i',i}-2\sigma^2 up to sampling fluctuations. The difference is therefore of constant order \sigma^2 and cannot be made o(1) by concentration. This invalidates the bias bound in Thm. 2 and hence the derivation of Cor. 2 in the missing-data setting.
  3. [S.4 (Lem. 1)] The proof of Lem. 1 states that AWNN's selected neighborhood size satisfies K_i = \arg\max_{K\in(0,K_{\max}]} MSE(K); since AWNN minimizes the objective in Eq. (4), the correct statement would be K_i = \arg\min_K MSE(K), and the inequality MSE(K_i)\le MSE(K) used in Eq. (E.12) only follows for a minimizer. As written, the proof of the central dominance lemma is internally inconsistent. Moreover, Lemma 1 compares deterministic upper bounds, not actual row-wise MSE or risk; the sentence that AWNN 'guarantees a lower row wise mean squared error than the best-performing unweighted row-nearest-neighbour estimator' overstates what a bound-to-bound comparison establishes.
  4. [Section 3.2, Eq. (9)] Cor. 2's rate is conditional on Eq. (9), an o() negligibility assumption involving the estimated weights, which is asserted without proof. The surrounding text says the term is 'generally o(...)', but no argument is supplied. Under the stated goal of guarantees 'under minimal assumptions', an unverified data-dependent condition that is needed for the main completion rate is a substantial gap; even if the earlier algebra were corrected, Cor. 2 would only be a conditional statement.
minor comments (4)
  1. [Algorithms 1 and 2] The theory assumes \sigma^2 is known, while Algorithm 1 initializes \hat{\sigma}^2 as Var/10 and uses a fixed-point iteration; no convergence analysis or consistency proof is given for this iteration, so the practical algorithm and the analyzed procedure are not fully matched.
  2. [Lem. 2] The statement of Lem. 2 uses '4\log(2m\delta)\sigma^2' where '4\log(2m/\delta)\sigma^2' is evidently intended; this appears in several places and should be corrected.
  3. [Algorithm 2, line 1] Algorithm 2 explicitly sets the diagonal of the estimated distance matrix to 0, but under Eq. (3) the self-distance is -2\sigma^2; this discrepancy between the analyzed objective and the implemented algorithm is not discussed.
  4. [Figure 1] The axis labels of Figure 1 appear garbled (e.g., '2 5', '2 3', '2 1'), and the reported decay slopes are embedded in the legend; a table or clearer axis labels would improve readability.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: the 'lower MSE' guarantee in Lemma 1 is a bound-to-bound comparison inherited from the algorithm's own minimization objective; the rate claim itself rests on independent unweighted-NN rates, so circularity is partial.

  1. self definitional [Section 3.1, Lemma 1; algorithm objective Eq. (4); proof in App. S.4]
    "Lemma 1. The A WNNalgorithm guarantees a lower row wise mean squared error than the best-performing unweighted row-nearest-neighbour estimator, even when the latter’s search radiusη2 is optimally tuned. BAW_i + VAW_i ≤ min_{η>0}{BUW_i(η2) + VUW_i(η2)}."

    The inequality actually proved in App. S.4 compares the Prop. 1 / Thm. 1 upper bounds, not the true row MSEs. Eq. (4) defines the AWNN weights as the minimizer of exactly the Prop. 1 upper-bound expression (with σ² in place of σ̂²). Since the unweighted uniform weights are a feasible point of that minimization, the bound-to-bound inequality is a direct consequence of the definition of argmin. Thus the advertised 'guarantees a lower row wise mean squared error' is not an independent statistical dominance result; it is a restatement of the fact that a minimum is no larger than any feasible value. The actual MSE dominance does not follow from this bound comparison.

full rationale

The main advertised finite-sample claim, Lemma 1, is a bound-to-bound comparison: the proof in App. S.4 establishes BAW_i + VAW_i ≤ min_η {BUW_i(η²)+VUW_i(η²)}, while the statement claims a lower row-wise MSE. Since Eq. (4) defines the AWNN weights as the minimizer of the very upper bound from Prop. 1, the inequality is essentially the definition of argmin (the unweighted uniform weights are feasible). Hence the 'guarantee' is partly self-referential: the method is constructed to minimize the bound that the theorem then reports. This is not a full circularity because the bound itself is derived from the data model, and the final rate is obtained by comparison with the independently known unweighted NN minimax rate; however, the stated MSE-dominance claim is not proved by the argument. Separately, the paper's rate proof in App. S.5 imports Lemma 2 from [SPD24] to lower-bound the unweighted NN count. This is a self-citation by the same three authors, but the lemma is a parameter-free statement about unweighted NN counts under stated assumptions and does not assume the AWNN result, so under the review rules it counts as independent support and is not scored as circular. Finally, I note a non-circular correctness defect: Appendix S.1's proof of Prop. 1 substitutes Eq. (3) with 2/m Σ w Σ(θ_i,j−θ_i',j)² = 2Σ w ρ_{i',i}, dropping a +4σ² term (since ρ_{i',i} = (1/m)Σ(θ_i,j−θ_i',j)² − 2σ² and Σw=1). The corrected identity has +4σ², which propagates through Thm. 1, Lem. 1, Cor. 1, Thm. 2, and Cor. 2 and invalidates the stated O(n^{-2λ/(d1+2λ)}) rate. This is an algebraic error, not circularity, but it means the advertised rate is not established by the manuscript's own derivation.

Assumptions & free parameters 1 free parameters · 8 assumptions · 0 invented entities

The central claim rests on five structural assumptions plus an unproved negligibility condition (9), and the method introduces one data-fitted quantity, bsigma^2, whose consistency is not shown. No new physical or model entities are postulated.

free parameters (1)
  • noise variance sigma^2 (estimated as bsigma^2) = fixed-point iteration initialized at Var(X)/10
    The algorithm replaces the sigma^2 used in Proposition 1 and Theorem 1 with a data-driven fixed-point estimate; the theory assumes sigma^2 is known, so the practical procedure introduces an unanalyzed fitted quantity.
assumptions (8)
  • domain assumption Assumption 1: estimated row distances brho concentrate around true distances rho within zeta_{i',i}(m,delta) with probability 1-delta.
    The size of zeta is never computed for the algorithm's actual distance estimator, and Corollary 1's rate is only nontrivial in the regime n = O(max zeta^{-(d1+2lambda)/(2lambda)}).
  • domain assumption Assumption 2: noise terms are i.i.d. centered sub-gaussian with variance sigma^2.
    Used for the Hoeffding and sub-gaussian tail bounds in Proposition 1 and Theorem 2.
  • domain assumption Assumption 3: ground truth follows a nonlinear factor model theta_{i,j} = f(u_i,v_j) with f a (lambda,L) Hoelder function.
    Needed to translate distance concentration into a lower bound on the number of neighbors and to obtain the rate n^{-2lambda/(d1+2lambda)}.
  • domain assumption Assumption 4: row latent factors u_i are i.i.d. uniform on [0,1]^{d1}.
    Used with Lemma 2 in [SPD24] to lower bound the number of neighbors in a radius-eta ball.
  • domain assumption Assumption 5: MCAR missingness, with indicators i.i.d. Bernoulli(p), independent of latent factors and noise.
    Used for the clean rate statement in Corollary 2; the authors state Theorem 2 itself applies to general missingness, but the corollary's rate relies on MCAR.
  • ad hoc to paper Equation (9): the distance-estimation error term is o of the average bias term.
    This negligibility condition is unproved and is invoked to convert the unwieldy Theorem 2 bound into the clean rate of Corollary 2.
  • standard math Lemma 2 from [SPD24] lower-bounds the number of nearest neighbors k_eta^2 under the uniform latent-factor model.
    Self-cited lemma used in the proofs of Corollaries 1 and 2; it is a covering or volume argument but is not proved in this paper and is not machine-checked.
  • standard math Result 1: weighted Azuma martingale concentration bound.
    Used in the proof of Theorem 2; stated without proof or citation, though standard versions exist.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adaptively-weighted Nearest Neighbors for Matrix Completion." pith.science (2026). https://pith.science/paper/FWM3NOFK

@misc{pith2026250509612,
  author       = {Pith},
  title        = {Pith review of: Adaptively-weighted Nearest Neighbors for Matrix Completion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FWM3NOFK}},
  note         = {Machine review of arXiv:2505.09612}
}
read the original abstract

In this technical note, we introduce and analyze AWNN: an adaptively weighted nearest neighbor method for performing matrix completion. Nearest neighbor (NN) methods are widely used in missing data problems across multiple disciplines such as in recommender systems and for performing counterfactual inference in panel data settings. Prior works have shown that in addition to being very intuitive and easy to implement, NN methods enjoy nice theoretical guarantees. However, the performance of majority of the NN methods rely on the appropriate choice of the radii and the weights assigned to each member in the nearest neighbor set and despite several works on nearest neighbor methods in the past two decades, there does not exist a systematic approach of choosing the radii and the weights without relying on methods like cross-validation. AWNN addresses this challenge by judiciously balancing the bias variance trade off inherent in weighted nearest-neighbor regression. We provide theoretical guarantees for the proposed method under minimal assumptions and support the theory via synthetic experiments.

Figures

Figures reproduced from arXiv: 2505.09612 by the authors.

Figure 1
Figure 1. MSE of AWNN and the benchmarks as a function of number of rows n(= m). Results are averaged across 10 runs where signals were generated from Lipschitz function (f with λ = 1) with latent variables’ dimension d = 2 and SNR = 1. Experimental setup We work with square matrices i.e., n = m to keep the demonstration of empirical results straightforward. RowNN’s hyperparameter η is tuned via cross-validation with the hype… view at source ↗
Figure 2
Figure 2. Variation of AWNN’s MSE behaviour with changing smoothness levels (λ) of signals {θi,j}i,j∈[n]×[m] in synthetic data. Results are averaged across 10 runs where signals were generated from Lipschitz function (f with λ ∈ {0.5, 0.75, 1}) with latent variables’ dimension d = 2. Top and bottom row correspond to MCAR and no missingness setup respectively while left and right column correspond to SNR of 10 and 2 respective… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 30 canonical work pages

  1. [1]

    Synthetic difference in differences

    Dmitry Arkhangelsky, Susan Athey, David A Hirshberg, Guido W Imbens, and Stefan Wager. Synthetic difference in differences. Technical report, National Bureau of Economic Research, 2019

    work page 2019

  2. [2]

    Abadie, A

    A. Abadie, A. Diamond, and J. Hainmueller. Synthetic control methods for comparative case studies: Estimating the effect of california's tobacco control program. Journal of the American Statistical Association , 2010

    work page 2010

  3. [3]

    Causal matrix completion

    Anish Agarwal, Munther Dahleh, Devavrat Shah, and Dennis Shen. Causal matrix completion. arXiv preprint arXiv:2109.15154 , 2021

    arXiv 2021

  4. [4]

    Abadie and J

    A. Abadie and J. Gardeazabal. The economic costs of conflict: A case study of the basque country. American Economic Review , 2003

    work page 2003

  5. [5]

    Design-based analysis in difference-in-differences settings with staggered adoption

    Susan Athey and Guido W Imbens. Design-based analysis in difference-in-differences settings with staggered adoption. Journal of Econometrics , 226(1):62--79, 2022

    work page 2022

  6. [6]

    In retrospect: Book of optics, 2015

    Jim Al-Khalili. In retrospect: Book of optics, 2015

    work page 2015

  7. [7]

    Synthetic interventions

    Anish Agarwal, Devavrat Shah, and Dennis Shen. Synthetic interventions. arXiv preprint arXiv:2006.07691 , 2020

    arXiv 2006

  8. [8]

    Matrix completion with data-dependent missingness probabilities

    Sohom Bhattacharya and Sourav Chatterjee. Matrix completion with data-dependent missingness probabilities. IEEE Transactions on Information Theory , 68(10):6762--6773, 2022

    work page 2022

Show all 38 references
  1. [9]

    On hoeffding’s inequalities

    Vidmantas Bentkus. On hoeffding’s inequalities. The Annals of Probability , 2004

    2004

  2. [10]

    Fast exact matrix completion: A unified optimization framework for matrix completion

    Dimitris Bertsimas and Michael Lingzhi Li. Fast exact matrix completion: A unified optimization framework for matrix completion. Journal of Machine Learning Research , 21(231):1--43, 2020

    2020

  3. [11]

    Synthetic controls with staggered adoption

    Eli Ben-Michael, Avi Feller, and Jesse Rothstein. Synthetic controls with staggered adoption. Journal of the Royal Statistical Society Series B: Statistical Methodology , 84(2):351--381, 2022

    2022

  4. [12]

    Local nearest neighbour classification with applications to semi-supervised learning

    Timothy I Cannings, Thomas B Berrett, and Richard J Samworth. Local nearest neighbour classification with applications to semi-supervised learning. The Annals of Statistics , 48(3):1789--1814, 2020

    2020

  5. [13]

    Noisy matrix completion: Understanding statistical guarantees for convex relaxation via nonconvex optimization

    Yuxin Chen, Yuejie Chi, Jianqing Fan, Cong Ma, and Yuling Yan. Noisy matrix completion: Understanding statistical guarantees for convex relaxation via nonconvex optimization. SIAM journal on optimization , 30(4):3098--3121, 2020

    2020

  6. [14]

    Using side information to reliably learn low-rank matrices from missing and corrupted observations

    Kai-Yang Chiang, Inderjit S Dhillon, and Cho-Jui Hsieh. Using side information to reliably learn low-rank matrices from missing and corrupted observations. Journal of Machine Learning Research , 19(76):1--35, 2018

    2018

  7. [15]

    Nearest neighbor pattern classification

    Thomas Cover and Peter Hart. Nearest neighbor pattern classification. IEEE transactions on information theory , 13(1):21--27, 1967

    1967

  8. [16]

    Matrix estimation by universal singular value thresholding

    Sourav Chatterjee. Matrix estimation by universal singular value thresholding. The Annals of Statistics , 43(1):177--214, 2015

    2015

  9. [17]

    Matrix completion with noisy side information

    Kai-Yang Chiang, Cho-Jui Hsieh, and Inderjit S Dhillon. Matrix completion with noisy side information. Advances in neural information processing systems , 28, 2015

    2015

  10. [18]

    Counterfactual inference for sequential experiments

    Raaz Dwivedi, Katherine Tian, Sabina Tomkins, Predrag Klasnja, Susan Murphy, and Devavrat Shah. Counterfactual inference for sequential experiments. arXiv preprint arXiv:2202.06891 , 2022

    2022 arXiv

  11. [19]

    Doubly robust nearest neighbors in factor models

    Raaz Dwivedi, Katherine Tian, Sabina Tomkins, Predrag Klasnja, Susan Murphy, and Devavrat Shah. Doubly robust nearest neighbors in factor models. arXiv preprint arXiv:2211.14297 , 2022

    2022 arXiv

  12. [20]

    Weighted matrix completion and recovery with prior subspace information

    Armin Eftekhari, Dehui Yang, and Michael B Wakin. Weighted matrix completion and recovery with prior subspace information. IEEE Transactions on Information Theory , 64(6):4044--4071, 2018

    2018

  13. [21]

    Discriminatory analysis: nonparametric discrimination, small sample performance

    Evelyn Fix and Joseph Lawson Hodges. Discriminatory analysis: nonparametric discrimination, small sample performance . Air University, USAF School of Aviation Medecine, 1952

    1952

  14. [22]

    Bandwidth selection in kernel density estimation: Oracle inequalities and adaptive minimax optimality

    Alexander Goldenshluger and Oleg Lepski. Bandwidth selection in kernel density estimation: Oracle inequalities and adaptive minimax optimality. The Annals of Statistics , 39(3), June 2011

    2011

  15. [23]

    General selection rule from a family of linear estimators

    AV Goldenshluger and OV Lepski. General selection rule from a family of linear estimators. Theory of Probability & Its Applications , 57(2):209--226, 2013

    2013

  16. [24]

    Global optimality in inductive matrix completion

    Mohsen Ghassemi, Anand Sarwate, and Naveen Goela. Global optimality in inductive matrix completion. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 2226--2230. IEEE, 2018

    2018

  17. [25]

    Convex co-embedding for matrix completion with predictive side information

    Yuhong Guo. Convex co-embedding for matrix completion with predictive side information. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 31, 2017

    2017

  18. [26]

    Matrix completion and low-rank svd via fast alternating least squares

    Trevor Hastie, Rahul Mazumder, Jason D Lee, and Reza Zadeh. Matrix completion and low-rank svd via fast alternating least squares. The Journal of Machine Learning Research , 16(1):3367--3402, 2015

    2015

  19. [27]

    Provable inductive matrix completion

    Prateek Jain and Inderjit S Dhillon. Provable inductive matrix completion. arXiv preprint arXiv:1306.0626 , 2013

    2013 arXiv

  20. [28]

    Matrix factorization techniques for recommender systems

    Yehuda Koren, Robert Bell, and Chris Volinsky. Matrix factorization techniques for recommender systems. Computer , 42(8):30--37, 2009

    2009

  21. [29]

    An inequality for the analysis of variance

    Norbert Kaiblinger and Bernhard Spangl. An inequality for the analysis of variance. Mathematical Inequalities & Applications , 23(3):961--9, 2020

    2020

  22. [30]

    A sparse interactive model for matrix completion with side information

    Jin Lu, Guannan Liang, Jiangwen Sun, and Jinbo Bi. A sparse interactive model for matrix completion with side information. Advances in neural information processing systems , 29, 2016

    2016

  23. [31]

    Nearest neighbors for matrix estimation interpreted as blind regression for latent variable model

    Yihua Li, Devavrat Shah, Dogyoon Song, and Christina Lee Yu. Nearest neighbors for matrix estimation interpreted as blind regression for latent variable model. IEEE Transactions on Information Theory , 66(3):1760--1784, 2019

    2019

  24. [32]

    Trust your neighbors: A comprehensive survey of neighborhood-based methods for recommender systems

    Athanasios N Nikolakopoulos, Xia Ning, Christian Desrosiers, and George Karypis. Trust your neighbors: A comprehensive survey of neighborhood-based methods for recommender systems. Recommender systems handbook , pages 39--89, 2021

    2021

  25. [33]

    Alhazen and the nearest neighbor rule

    Marcello Pelillo. Alhazen and the nearest neighbor rule. Pattern Recognition Letters , 38:34--37, 2014

    2014

  26. [34]

    A simpler approach to matrix completion

    Benjamin Recht. A simpler approach to matrix completion. Journal of Machine Learning Research , 12(12), 2011

    2011

  27. [35]

    On adaptivity and minimax optimality of two-sided nearest neighbors

    Tathagata Sadhukhan, Manit Paul, and Raaz Dwivedi. On adaptivity and minimax optimality of two-sided nearest neighbors. arXiv preprint arXiv:2411.12965 , 2024

    2024 arXiv

  28. [36]

    Speedup matrix completion with side information: Application to multi-label learning

    Miao Xu, Rong Jin, and Zhi-Hua Zhou. Speedup matrix completion with side information: Application to multi-label learning. Advances in neural information processing systems , 26, 2013

    2013

  29. [37]

    Nonparametric matrix estimation with one-sided covariates

    Christina Lee Yu. Nonparametric matrix estimation with one-sided covariates. In 2022 IEEE International Symposium on Information Theory (ISIT) , pages 892--897. IEEE, 2022

    2022

  30. [38]

    Efficient matrix sensing using rank-1 gaussian measurements

    Kai Zhong, Prateek Jain, and Inderjit S Dhillon. Efficient matrix sensing using rank-1 gaussian measurements. In Algorithmic Learning Theory: 26th International Conference, ALT 2015, Banff, AB, Canada, October 4-6, 2015, Proceedings 26 , pages 3--18. Springer, 2015

    2015

Pith tools

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