Pith. sign in

REVIEW 4 major objections 5 minor 66 references

NDCG-Consistent Softmax Approximation with Accelerated Convergence

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Taylor expansion of the softmax loss at zero yields two quadratic ranking losses that inherit softmax's DCG-consistency, can be trained with closed-form alternating least squares updates, and match or beat softmax on three recommendation…

desk verdict A solid theoretical bridge from softmax to weighted squared losses, but the experiments evaluate a tuned variant rather than the exact derived objective. read the letter →

arxiv 2506.09454 v1 pith:2MW2QXRP submitted 2025-06-11 cs.LG

Yuanhao Pu , Defu Lian , Xiaolong Chen , Xu Huang , Jin Chen , Enhong Chen This is my paper
classification cs.LG
keywords similaritylearningrankingsoftmaxapproximationBayesconsistencyDCGalternatingleastsquaresweightedsquaredlossmatrixfactorization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that the softmax ranking loss can be replaced by a quadratic surrogate without losing its ranking guarantees. Taylor-expanding the softmax and sampled-softmax losses at zero produces two new losses, the squared $RG^2$ and the interaction-preserving $RG^\times$, which are non-sampling, admit closed-form alternating least squares updates, and are provably consistent with the DCG ranking metric (discounted cumulative gain) because they inherit the softmax's Bregman-divergence structure. Optimizing these losses with ALS on a matrix-factorization backbone reaches or beats softmax's MRR/NDCG on MovieLens, Amazon Electronics, and Steam in fewer epochs, while staying close on a link-prediction dataset. If the claims hold, large-scale ranking and similarity-learning systems can keep softmax-like ranking quality while replacing stochastic gradient training with faster, deterministic block updates, and the heuristic weights of weighted squared losses get a theoretical explanation.

What carries the argument

The engine is the second-order Taylor expansion of the softmax loss at the all-zero score vector, which turns the nonlinear softmax into a quadratic form. The resulting loss is reorganized as a weighted squared loss, $RG^2 \propto \sum_{x,y} |I_x| (o^{(x)}_y + 1 - r_{x,y} N/|I_x|)^2$, with $RG^\times$ subtracting the mean-interaction term $\frac{1}{N}(1_N^T o^{(x)})^2$. Consistency is carried by the Bregman-divergence representation $L = D_\phi(g(o^{(x)}), \eta)$ with generating functions $\phi_2(o) = |I_x|\|o\|^2$ and $\phi_\times(o) = |I_x| o^T (I - \frac{1}{N}1_N1_N^T)o$ and inverse-order-preserving links $g_2(o) = 2|I_x|o$, $g_\times(o) = 2|I_x|(I - \frac{1}{N}1_N1_N^T)o$; this is what transfers softmax's DCG-consistency to the quadratic losses. Finally, the quadratic structure gives closed-form ALS updates in which each block update solves a small linear system, which is why training converges faster than SGD on softmax.

What would settle it

Record the distribution of predicted scores during ALS training on any of the four datasets: if many scores move well beyond the region where the Taylor remainder is negligible (roughly $|o| > 1$), then the quadratic surrogate is no longer approximating softmax and $RG^\times$ should visibly diverge from softmax's ranking. A second check is to train on synthetic data whose true relevance matrix has rank far above 64, where the DCG-consistency guarantee should fail.

Watch

Extended reading notes

Core claim

The central claim is that the softmax loss can be approximated by a quadratic loss without sacrificing its ranking semantics. Expanding $L(o^{(x)}) = -\log(\mathrm{softmax}(o^{(x)})_y)$ at $o=0$ yields, up to constants, $-o^{(x)}_y + \frac{1}{2N}\|o^{(x)} + 1_N\|^2 - \frac{1}{2N^2}(1_N^T o^{(x)})^2$; keeping the first two terms gives the $RG^2$ loss and keeping the full quadratic gives $RG^\times$. By rewriting these as weighted squared losses and identifying them as Bregman divergences with inverse-order-preserving link functions, the paper proves both are DCG-consistent, just like softmax, and shows $RG^2$ upper-bounds softmax under a single-click assumption. Because the losses are quadratic, a matrix-factorization instantiation can be optimized with alternating least squares, whose updates have closed forms and converge linearly rather than at SGD's sublinear rate. The empirical claim, on four datasets, is that $RG^2$ and $RG^\times$ match or beat softmax ranking quality on three recommendation datasets and are only slightly behind on the link-prediction set while converging substantially faster.

Load-bearing premise

The formal guarantees presume both that the true relevance probabilities can be represented by the low-rank matrix-factorization model used (rank 64 in the experiments) and that the model's predicted scores stay close enough to zero for the Taylor expansion to hold.

Editorial extensions

If this is right

  • Because $RG^2$ and $RG^\times$ are quadratic, they can be optimized with ALS at per-iteration cost $O(T(|D|K^2 + (M+N)K^3))$, so softmax-style ranking can scale to very large item corpora without negative sampling.
  • Under the single-click assumption, $RG^2$ is an upper bound on softmax loss at their optima, so optimizing $RG^2$ also controls softmax behavior; this gives a principled explanation for why weighted squared losses such as WRMF work well for ranking.
  • The sampled-softmax-derived weights show that increasing the number of negative samples in sampled softmax is equivalent to adjusting positive and negative weights in a non-sampling squared loss, formally connecting the sampling and non-sampling paradigms.
  • The consistency and generalization results imply that, whenever the Bayes-optimal scoring function lies inside the rank-$K$ hypothesis space, optimizing $RG$ losses recovers the optimal DCG ranking rather than merely a proxy for it.
  • On the reported datasets, $RG^2$ and $RG^\times$ match or exceed softmax on all recommendation benchmarks while the ALS optimization reaches a given ranking metric far earlier in wall-clock time.

Reading between the lines

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

  • The same Taylor-at-zero trick could be applied to other normalizing losses, such as sparsemax or hierarchical softmax, to obtain quadratic surrogates with their own consistency properties; the paper does not explore this.
  • Because the Taylor expansion is centered at zero and justified by regularization, the method implicitly predicts that strong $L^2$ regularization is part of the approximation: without it, RG losses should drift away from softmax behavior, which is a testable prediction.
  • The derived weight formula suggests a principled way to set weighted-squared-loss hyperparameters from a target negative-sampling budget rather than by heuristic tuning, which the experiments still do via $\alpha$ and $\beta$.
  • The generalization bound weakens as the embedding dimension $K$ grows, so on genuinely high-rank data the practical advantage of RG losses may come from the optimization speedup rather than from the DCG-consistency theorem.
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.

Referee Report

4 major / 5 minor

Summary. The paper proposes two squared-form surrogate losses, RG² and RG×, obtained by a second-order Taylor expansion of the softmax and sampled-softmax losses at the zero logit vector. The authors show that these losses can be written as weighted squared losses with weights determined by |I_x| and the sampling size n, thereby connecting non-sampling losses with sampling-based softmax variants. They prove DCG-consistency of the proposed losses via a Bregman-divergence argument, provide a generalization bound for the RG losses instantiated with matrix factorization, and state an ALS convergence-rate comparison. Experiments on MovieLens, Amazon Electronics, Steam, and Wikipedia compare RG² and RG× against BPR, BCE, SSM, Sparsemax, UIB, SML, WRMF, and softmax, reporting that the RG losses match or exceed softmax ranking performance with substantially faster convergence under ALS. The main theoretical claims rely on the assumption that the Bayes-optimal scoring function lies in the low-rank hypothesis space, and the appendix states that the experimental implementation replaces the derived weights W and V by tuned hyperparameters α and β.

Significance. If the claims are fully substantiated, the paper offers a valuable conceptual bridge: it derives WSL-type weights from a Taylor expansion of the softmax loss, provides a theoretical explanation for the empirical success of weighted squared losses in ranking, and connects sampling and non-sampling loss families through the sampling size n. The DCG-consistency framework and the generalization bound are useful additions to the ranking theory literature, and the experiments cover diverse datasets and baselines. However, the current empirical validation is weakened by the replacement of the derived weights with tuned hyperparameters, so the reported gains do not directly test the theoretically motivated objective. The formal consistency and generalization results also depend on assumptions (low-rank Bayes optimality, accuracy of the Taylor expansion) that are not validated on the trained models. These gaps are fixable but require substantive additional work, so the paper is not yet ready for acceptance.

major comments (4)
  1. [Section 5 and Appendix B.2] The experiments evaluate a different objective from the one whose theory is established. Appendix B.2 states that 'the coefficients W and V in equations are replaced by hyperparameters α, β,' while Eq. (32) defines W and V as |I_x|, |I_x|/N, or w_xy. The DCG-consistency result in Theorem 12 and the Bregman argument in Section 4.2 depend on these exact weights: the link functions g2 and g× are inverse order-preserving only for uniform weighting, and the paper itself warns that assigning different weights to positive and negative samples 'may disrupt this property and hence affect DCG-consistency.' Consequently, Tables 3-4 and Figure 1 validate a tuned weighted-squared-loss family rather than the proposed RG²/RG× objectives. If α and β are tuned separately for positives and negatives, the consistency proof no longer applies; if they are uniform constants, the claimed benefit of the derived weights is untested. The central empirical claim must be re-run with the exact derived W and V, or the paper must explicitly reframe the experiments as evaluating a separate tuned surrogate family.
  2. [Section 4.2 and Proposition 15] The consistency and generalization results require f_B(x,y)=E[r|x,y] to lie in the hypothesis space H, as stated after Eq. (25). For rank-64 matrix factorization on real-world data with the sparsity reported in Table 5, this is not plausible. Under misspecification, the bound in Lemma 9 gains an approximation error term involving inf_{f∈H} Σ(f-f_B)², which is neither bounded nor discussed. Therefore the formal claim that the trained MF model is NDCG/DCG-consistent is not established. The authors should either add a misspecification analysis or explicitly list this as a limitation of the theoretical results.
  3. [Section 4.1 and Appendix A.3] The Taylor expansion at o0=0 is the only derivation of RG² and RG×, but its validity for the logit magnitudes reached by ALS is not established. Section 4.1 justifies the expansion by noting that the optimum is 'typically constrained close to the zero vector due to regularization,' yet the closed-form ALS updates in Eqs. (36)-(39) can produce logits far from zero, and Proposition 7's proof in Appendix A.3 only checks a Hessian bound under a 'single-click behavior' condition that is not formalized or verified on the trained models. Without a quantitative bound on the Taylor remainder along the optimization trajectory, the claim that RG losses are efficient approximations of softmax is not rigorously supported.
  4. [Section 4.5 and Appendix A.7] The convergence rate claim for ALS on the RG objectives is not proven for the specific losses in Eq. (31). Theorem 26 (Jain et al.) requires µ-marginal strong convexity, L-marginal strong smoothness, and M-robust bistability; the text asserts that these hold 'due to their quadratic structures' but does not verify them, and Eq. (31) is only marginally quadratic, not jointly. Robust bistability in particular is nontrivial for a quartic objective. The convergence-rate table is therefore a comparison of textbook rates rather than a guarantee for the proposed method. Please provide the verification or state the conditions under which the cited theorem applies.
minor comments (5)
  1. [Section 5.3] The text refers to 'RG 1' in the description of Figure 1, but RG1 is never defined; this should be RG² or RG×.
  2. [Abstract and Section 4.2] The abstract and title say NDCG-consistency, but Definition 8 and the proofs in Section 4.2 concern DCG. Remark 22 discusses normalization, but the abstract and title should be aligned with the actual theorem statements.
  3. [Appendix B.3] The sentence 'the negative sampling numbers are 10, 10, 1 and 1, 100, 50 for all recommendation datasets' is garbled and should be restated with the intended per-dataset values.
  4. [Equation (33) and Appendix A.5] The generalization bound derivation contains inconsistent constants: the proof replaces the factor (32eB)^d with (16eL)^d without comment, and the final displayed bound should be reconciled with the δ-solving step.
  5. [Section 5.2, Findings 2] The phrase 'a 3.3% average relative decrease compared with SM' is unclear; please specify the dataset and whether a decrease is a desirable outcome or not.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Taylor-derived RG losses and their DCG-consistency proofs are self-contained; the only self-citation is incidental.

full rationale

The central derivation is a Taylor expansion of the softmax and sampled-softmax losses at o0=0 (Eqs. 12-18), which is a direct algebraic transformation of the softmax definition, not an assumption of the conclusion. The DCG-consistency of RG2/RGx (Theorem 12, Corollary 11, and the Bregman-divergence representation in Eqs. 29-30) is established via the standard squared-loss-to-DCG bound of Lemma 9, following Cossock and Zhang, with externally cited results from Ravikumar et al. and Yang and Koyejo; it does not depend on the authors' own prior work. The generalization bound (Theorem 14) uses Srebro et al.'s pseudo-dimension bound and Anthony et al.'s covering-number argument, and the ALS convergence analysis (Section 4.5) cites Jain et al.; all are independent, external mathematical results. The paper's only self-citation, Pu et al. (2024), is used for an optional simplification ("as we did in Pu et al. (2024)") and for an alternative optimizer, and is not load-bearing for the main consistency or approximation claims. One experimental-validity caveat should be flagged: Appendix B.2 states "Considering the operability of experiments, the coefficients W and V in equations are replaced by hyperparameters α, β," so Tables 3-4 and Figure 1 evaluate a hyperparameter-tuned member of the weighted-squared-loss family rather than the exact RG2/RGx objectives whose DCG-consistency is proved. This weakens the empirical link between theory and experiments, but it is not circular reasoning: the theory is proved independently of the tuned coefficients, and the experiments do not feed back into the derivation. Therefore, no circular step is exhibited, and the circularity score is 0.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central theoretical claims rest on assumptions that are either unverified (Taylor accuracy, f_B in H, ALS conditions) or only partially proven (single-click upper bound). The experiments use tuned weights rather than the derived weights, adding a layer of fitting between theory and validation.

free parameters (2)
  • alpha and beta (loss weight hyperparameters) = tuned per dataset from {0, 1, 0.5, 0.1, 0.05, 0.01, 0.005, 0.001}
    In Appendix B.2, the theoretically derived weights W and V in Eq. (31) are replaced by tuned hyperparameters alpha and beta for experimental operability. The loss actually optimized in experiments is not exactly the derived RG loss, and the reported gains could depend on this tuning.
  • lambda (regularization coefficient) = tuned from {0, 0.1, 0.01, 0.001} for WRMF and {0, 1, 0.5, 0.1, 0.05, 0.01, 0.005, 0.001} for RG
    Regularization strength is a standard free parameter; it affects how close the solution stays to o=0, which the Taylor expansion relies on.
assumptions (4)
  • domain assumption The Taylor expansion of softmax loss at o0=0 is a sufficiently accurate approximation across the score range seen in training.
    Section 4.1 states o0=0 is chosen because regularization keeps the optimal solution near zero, but no bound on the approximation error over the optimization trajectory is provided; ranking logits in matrix factorization often grow away from zero.
  • domain assumption The Bayes-optimal scoring function f_B(x,y)=E[r|x,y] lies within the low-rank hypothesis space H.
    Section 4.2 assumes f_B is in H after Eq. 25 to make the DCG-consistency proof apply. With embedding dimension K=64 on sparse real datasets this is generally false, so the consistency guarantee does not directly cover the trained model.
  • ad hoc to paper Single-click behavior assumption: each context has at most one relevant positive item.
    Proposition 7, that softmax loss is upper bounded by RG, relies on a sufficient condition p_max <= 1/N + sum p_j^2. The proof in Appendix A.3 argues this holds for uniform or highly concentrated distributions and explicitly invokes the single-click behavior assumption, but it does not cover intermediate distributions in general.
  • ad hoc to paper The RG losses satisfy mu-marginal strong convexity, L-marginal strong smoothness, and M-robust bistability needed for ALS linear convergence.
    Appendix A.7 asserts the RG losses have these properties 'due to their quadratic structures' without verifying the conditions for the actual weighted loss, which may have zero or small weights on negative samples.

how reviews work

0 comments
Cite this review

Pith. "Pith review of NDCG-Consistent Softmax Approximation with Accelerated Convergence." pith.science (2026). https://pith.science/paper/2MW2QXRP

@misc{pith2026250609454,
  author       = {Pith},
  title        = {Pith review of: NDCG-Consistent Softmax Approximation with Accelerated Convergence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2MW2QXRP}},
  note         = {Machine review of arXiv:2506.09454}
}
abstract

Ranking tasks constitute fundamental components of extreme similarity learning frameworks, where extremely large corpora of objects are modeled through relative similarity relationships adhering to predefined ordinal structures. Among various ranking surrogates, Softmax (SM) Loss has been widely adopted due to its natural capability to handle listwise ranking via global negative comparisons, along with its flexibility across diverse application scenarios. However, despite its effectiveness, SM Loss often suffers from significant computational overhead and scalability limitations when applied to large-scale object spaces. To address this challenge, we propose novel loss formulations that align directly with ranking metrics: the Ranking-Generalizable \textbf{squared} (RG$^2$) Loss and the Ranking-Generalizable interactive (RG$^\times$) Loss, both derived through Taylor expansions of the SM Loss. Notably, RG$^2$ reveals the intrinsic mechanisms underlying weighted squared losses (WSL) in ranking methods and uncovers fundamental connections between sampling-based and non-sampling-based loss paradigms. Furthermore, we integrate the proposed RG losses with the highly efficient Alternating Least Squares (ALS) optimization method, providing both generalization guarantees and convergence rate analyses. Empirical evaluations on real-world datasets demonstrate that our approach achieves comparable or superior ranking performance relative to SM Loss, while significantly accelerating convergence. This framework offers the similarity learning community both theoretical insights and practically efficient tools, with methodologies applicable to a broad range of tasks where balancing ranking quality and computational efficiency is essential.

Figures

Figures reproduced from arXiv: 2506.09454 by the authors.

Figure 1
Figure 1. Comparisons of convergence speed on all datasets. [PITH_FULL_IMAGE:figures/full_fig_p022_1.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 59 canonical work pages

  1. [1]

    Backpropagation and stochastic gradient descent method

    Shun-ichi Amari. Backpropagation and stochastic gradient descent method. Neurocomputing, 5 0 (4-5): 0 185--196, 1993

    work page 1993

  2. [2]

    Neural network learning: Theoretical foundations, volume 9

    Martin Anthony, Peter L Bartlett, Peter L Bartlett, et al. Neural network learning: Theoretical foundations, volume 9. cambridge university press Cambridge, 1999

    1999

  3. [3]

    H-consistency bounds for surrogate loss minimizers

    Pranjal Awasthi, Anqi Mao, Mehryar Mohri, and Yutao Zhong. H-consistency bounds for surrogate loss minimizers. In International Conference on Machine Learning, pages 1117--1174. PMLR, 2022 a

    work page 2022

  4. [4]

    Multi-class h -consistency bounds

    Pranjal Awasthi, Anqi Mao, Mehryar Mohri, and Yutao Zhong. Multi-class h -consistency bounds. Advances in neural information processing systems, 35: 0 782--795, 2022 b

    2022

  5. [5]

    Convexity, classification, and risk bounds

    Peter L Bartlett, Michael I Jordan, and Jon D McAuliffe. Convexity, classification, and risk bounds. Journal of the American Statistical Association, 101 0 (473): 0 138--156, 2006

    2006

  6. [6]

    A generic coordinate descent framework for learning from implicit feedback

    Immanuel Bayer, Xiangnan He, Bhargav Kanagal, and Steffen Rendle. A generic coordinate descent framework for learning from implicit feedback. In Proceedings of the 26th international conference on world wide web, pages 1341--1350, 2017

    work page 2017

  7. [7]

    Adaptive importance sampling to accelerate training of a neural probabilistic language model

    Yoshua Bengio and Jean-S \'e bastien Sen \'e cal. Adaptive importance sampling to accelerate training of a neural probabilistic language model. IEEE Transactions on Neural Networks, 19 0 (4): 0 713--722, 2008

    work page 2008

  8. [8]

    Optimization methods for large-scale machine learning

    L \'e on Bottou, Frank E Curtis, and Jorge Nocedal. Optimization methods for large-scale machine learning. SIAM Review, 60 0 (2): 0 223--311, 2018

    work page 2018

Show all 66 references
  1. [9]

    An alternative cross entropy loss for learning-to-rank

    Sebastian Bruch. An alternative cross entropy loss for learning-to-rank. In Proceedings of the web conference 2021, pages 118--126, 2021

    2021

  2. [10]

    An analysis of the softmax cross entropy loss for learning-to-rank with binary relevance

    Sebastian Bruch, Xuanhui Wang, Michael Bendersky, and Marc Najork. An analysis of the softmax cross entropy loss for learning-to-rank with binary relevance. In Proceedings of the 2019 ACM SIGIR international conference on theory of information retrieval, pages 75--78, 2019

    2019

  3. [11]

    Christopher J.C. Burges. From ranknet to lambdarank to lambdamart: An overview. In Learning to Rank Challenge, 2010

    2010

  4. [12]

    Efficient neural matrix factorization without sampling for recommendation

    Chong Chen, Min Zhang, Yongfeng Zhang, Yiqun Liu, and Shaoping Ma. Efficient neural matrix factorization without sampling for recommendation. ACM Transactions on Information Systems (TOIS), 38 0 (2): 0 1--28, 2020 a

    2020

  5. [13]

    Revisiting negative sampling vs

    Chong Chen, Weizhi Ma, Min Zhang, Chenyang Wang, Yiqun Liu, and Shaoping Ma. Revisiting negative sampling vs. non-sampling in implicit recommendation. ACM Transactions on Information Systems, 41 0 (1): 0 1--25, 2023

    2023

  6. [14]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. ICML, pages 1597--1607, 2020 b

    2020

  7. [15]

    Learning a similarity metric discriminatively, with application to face verification

    Sumit Chopra, Raia Hadsell, and Yann LeCun. Learning a similarity metric discriminatively, with application to face verification. In CVPR, volume 1, pages 539--546, 2005

    2005

  8. [16]

    Subset ranking using regression

    David Cossock and Tong Zhang. Subset ranking using regression. In Learning Theory: 19th Annual Conference on Learning Theory, COLT 2006, Pittsburgh, PA, USA, June 22-25, 2006. Proceedings 19, pages 605--619. Springer, 2006

    2006

  9. [17]

    Deep neural networks for youtube recommendations

    Paul Covington, Jay Adams, and Emre Sargin. Deep neural networks for youtube recommendations. In Proceedings of the 10th ACM conference on recommender systems, pages 191--198, 2016

    2016

  10. [18]

    Saga: A fast incremental gradient method with support for non-strongly convex composite objectives

    Aaron Defazio, Francis Bach, and Simon Lacoste-Julien. Saga: A fast incremental gradient method with support for non-strongly convex composite objectives. In Advances in Neural Information Processing Systems (NeurIPS), volume 27, page 1646–1654, 2014

    2014

  11. [19]

    Inexact newton methods

    Ron S Dembo, Stanley C Eisenstat, and Trond Steihaug. Inexact newton methods. SIAM Journal on Numerical Analysis, 19 0 (2): 0 400--408, 1982

    1982

  12. [20]

    Simcse: Simple contrastive learning of sentence embeddings

    Tianyu Gao, Xingcheng Yao, and Danqi Chen. Simcse: Simple contrastive learning of sentence embeddings. In EMNLP, pages 6894--6910, 2021

    2021

  13. [21]

    Stochastic first-and zeroth-order methods for nonconvex stochastic programming

    Saeed Ghadimi and Guanghui Lan. Stochastic first-and zeroth-order methods for nonconvex stochastic programming. SIAM Journal on Optimization, 23 0 (4): 0 2341--2368, 2013

    2013

  14. [22]

    Noise-contrastive estimation: A new estimation principle for unnormalized statistical models

    Michael Gutmann and Aapo Hyv \"a rinen. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proceedings of the thirteenth international conference on artificial intelligence and statistics, pages 297--304. JMLR Workshop and Conferen...

    2010

  15. [23]

    Neural collaborative filtering

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. Neural collaborative filtering. In Proceedings of the 26th international conference on world wide web, pages 173--182, 2017

    2017

  16. [24]

    Collaborative filtering for implicit feedback datasets

    Yifan Hu, Yehuda Koren, and Chris Volinsky. Collaborative filtering for implicit feedback datasets. In 2008 Eighth IEEE international conference on data mining, pages 263--272. Ieee, 2008

    2008

  17. [25]

    Cooperative retriever and ranker in deep recommenders

    Xu Huang, Defu Lian, Jin Chen, Liu Zheng, Xing Xie, and Enhong Chen. Cooperative retriever and ranker in deep recommenders. In Proceedings of the ACM Web Conference 2023, pages 1150--1161, 2023

    2023

  18. [26]

    Peter J. Huber. Robust Statistics. Wiley, 2nd edition, 2004

    2004

  19. [27]

    Non-convex optimization for machine learning

    Prateek Jain, Purushottam Kar, et al. Non-convex optimization for machine learning. Foundations and Trends in Machine Learning , 10 0 (3-4): 0 142--363, 2017

    2017

  20. [28]

    Accelerating stochastic gradient descent using predictive variance reduction

    Rie Johnson and Tong Zhang. Accelerating stochastic gradient descent using predictive variance reduction. In Advances in Neural Information Processing Systems (NeurIPS), volume 26, page 315–323, 2013

    2013

  21. [29]

    Self-attentive sequential recommendation

    Wang-Cheng Kang and Julian McAuley. Self-attentive sequential recommendation. In 2018 IEEE international conference on data mining (ICDM), pages 197--206. IEEE, 2018

    2018

  22. [30]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  23. [31]

    Factorization meets the neighborhood: a multifaceted collaborative filtering model

    Yehuda Koren. Factorization meets the neighborhood: a multifaceted collaborative filtering model. In Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 426--434, 2008

    2008

  24. [32]

    Matrix factorization techniques for recommender systems

    Yehuda Koren, Robert Bell, and Chris Volinsky. Matrix factorization techniques for recommender systems. Computer, 42 0 (8): 0 30--37, 2009

    2009

  25. [33]

    Efficient training on very large corpora via gramian estimation

    Walid Krichene, Nicolas Mayoraz, Steffen Rendle, Li Zhang, Xinyang Yi, Lichan Hong, Ed Chi, and John Anderson. Efficient training on very large corpora via gramian estimation. In International Conference on Learning Representations, 2018

    2018

  26. [34]

    Symmetric metric learning with adaptive margin for recommendation

    Mingming Li, Shuai Zhang, Fuqing Zhu, Wanhui Qian, Liangjun Zang, Jizhong Han, and Songlin Hu. Symmetric metric learning with adaptive margin for recommendation. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 4634--4641, 2020

    2020

  27. [35]

    Geomf: joint geographical modeling and matrix factorization for point-of-interest recommendation

    Defu Lian, Cong Zhao, Xing Xie, Guangzhong Sun, Enhong Chen, and Yong Rui. Geomf: joint geographical modeling and matrix factorization for point-of-interest recommendation. In Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, p...

    2014

  28. [36]

    Recstudio: Towards a highly-modularized recommender system

    Defu Lian, Xu Huang, Xiaolong Chen, Jin Chen, Xingmei Wang, Yankai Wang, Haoran Jin, Rui Fan, Zheng Liu, Le Wu, et al. Recstudio: Towards a highly-modularized recommender system. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Infor...

    2023

  29. [37]

    Variational autoencoders for collaborative filtering

    Dawen Liang, Rahul G Krishnan, Matthew D Hoffman, and Tony Jebara. Variational autoencoders for collaborative filtering. In Proceedings of the 2018 world wide web conference, pages 689--698, 2018

    2018

  30. [38]

    Consistency versus realizable h-consistency for multiclass classification

    Phil Long and Rocco Servedio. Consistency versus realizable h-consistency for multiclass classification. In International conference on machine learning, pages 801--809. PMLR, 2013

    2013

  31. [39]

    Support and centrality: Learning weights for knowledge graph embedding models

    Gengchen Mai, Krzysztof Janowicz, and Bo Yan. Support and centrality: Learning weights for knowledge graph embedding models. In Knowledge Engineering and Knowledge Management: 21st International Conference, EKAW 2018, Nancy, France, November 12-16, 2018, Proceedings 21, pages ...

    2018

  32. [40]

    From softmax to sparsemax: A sparse model of attention and multi-label classification

    Andre Martins and Ramon Astudillo. From softmax to sparsemax: A sparse model of attention and multi-label classification. In International conference on machine learning, pages 1614--1623. PMLR, 2016

    2016

  33. [41]

    Multilabel reductions: what is my loss optimising? Advances in Neural Information Processing Systems, 32, 2019

    Aditya K Menon, Ankit Singh Rawat, Sashank Reddi, and Sanjiv Kumar. Multilabel reductions: what is my loss optimising? Advances in Neural Information Processing Systems, 32, 2019

    2019

  34. [42]

    Distributed representations of words and phrases and their compositionality

    Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. Distributed representations of words and phrases and their compositionality. NIPS, 26, 2013

    2013

  35. [43]

    Learning-efficient yet generalizable collaborative filtering for item recommendation

    Yuanhao Pu, Xiaolong Chen, Xu Huang, Jin Chen, Defu Lian, and Enhong Chen. Learning-efficient yet generalizable collaborative filtering for item recommendation. In Forty-first International Conference on Machine Learning, 2024

    2024

  36. [44]

    On ndcg consistency of listwise ranking methods

    Pradeep Ravikumar, Ambuj Tewari, and Eunho Yang. On ndcg consistency of listwise ranking methods. In Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, pages 618--626. JMLR Workshop and Conference Proceedings, 2011

    2011

  37. [45]

    Sentence-bert: Sentence embeddings using siamese bert-networks with contrastive loss

    Nils Reimers and Iryna Gurevych. Sentence-bert: Sentence embeddings using siamese bert-networks with contrastive loss. EMNLP, pages 3982--3992, 2019

    2019

  38. [46]

    Item recommendation from implicit feedback

    Steffen Rendle. Item recommendation from implicit feedback. In Recommender Systems Handbook, pages 143--171. Springer, 2021

    2021

  39. [47]

    Improving pairwise learning for item recommendation from implicit feedback

    Steffen Rendle and Christoph Freudenthaler. Improving pairwise learning for item recommendation from implicit feedback. In Proceedings of the 7th ACM international conference on Web search and data mining, pages 273--282, 2014

    2014

  40. [48]

    Bpr: Bayesian personalized ranking from implicit feedback

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. Bpr: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618, 2012

    2012 arXiv

  41. [49]

    Revisiting the performance of ials on item recommendation benchmarks

    Steffen Rendle, Walid Krichene, Li Zhang, and Yehuda Koren. Revisiting the performance of ials on item recommendation benchmarks. In Proceedings of the 16th ACM Conference on Recommender Systems, pages 427--435, 2022

    2022

  42. [50]

    Facenet: A unified embedding for face recognition and clustering

    Florian Schroff, Dmitry Kalenichenko, and James Philbin. Facenet: A unified embedding for face recognition and clustering. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015

    2015

  43. [51]

    Bridgland

    Andrew W Senior, Richard Evans, John Jumper, James Kirkpatrick, Laurent Sifre, Tim Green, Chongli Qin, Augustin Ž \'i dek, Alexander W R Nelson, and Alex et al. Bridgland. Improved protein structure prediction using potentials from deep learning. Nature, 577 0 (7792): 0 706--710, 2020

    2020

  44. [52]

    Learning semantic representations using convolutional neural networks for web search

    Yelong Shen, Xiaodong He, Jianfeng Gao, Li Deng, and Gr \'e goire Mesnil. Learning semantic representations using convolutional neural networks for web search. In Proceedings of the 23rd international conference on world wide web, pages 373--374, 2014

    2014

  45. [53]

    Generalization error bounds for collaborative prediction with low-rank matrices

    Nathan Srebro, Noga Alon, and Tommi Jaakkola. Generalization error bounds for collaborative prediction with low-rank matrices. Advances In Neural Information Processing Systems, 17, 2004

    2004

  46. [54]

    Bert4rec: Sequential recommendation with bidirectional encoder representations from transformer

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang. Bert4rec: Sequential recommendation with bidirectional encoder representations from transformer. In Proceedings of the 28th ACM international conference on information and knowledge management, pages ...

    2019

  47. [55]

    Alternating least squares for personalized ranking

    G \'a bor Tak \'a cs and Domonkos Tikk. Alternating least squares for personalized ranking. In Proceedings of the sixth ACM conference on Recommender systems, pages 83--90, 2012

    2012

  48. [56]

    On the consistency of multiclass classification methods

    Ambuj Tewari and Peter L Bartlett. On the consistency of multiclass classification methods. Journal of Machine Learning Research, 8 0 (5), 2007

    2007

  49. [57]

    Learning fine-grained image similarity with deep ranking

    Jiang Wang, Yang Song, Thomas Leung, Chuck Rosenberg, Jinbin Wang, James Philbin, Bo Chen, and Ying Wu. Learning fine-grained image similarity with deep ranking. CVPR, pages 1386--1393, 2014

    2014

  50. [58]

    Wsabie: Scaling up to large vocabulary image annotation

    Jason Weston, Samy Bengio, and Nicolas Usunier. Wsabie: Scaling up to large vocabulary image annotation. 2011

    2011

  51. [59]

    On the effectiveness of sampled softmax loss for item recommendation

    Jiancan Wu, Xiang Wang, Xingyu Gao, Jiawei Chen, Hongcheng Fu, Tianyu Qiu, and Xiangnan He. On the effectiveness of sampled softmax loss for item recommendation. arXiv preprint arXiv:2201.02327, 2022

    2022 arXiv

  52. [60]

    A no-regret generalization of hierarchical softmax to extreme multi-label classification

    Marek Wydmuch, Kalina Jasinska, Mikhail Kuznetsov, R \'o bert Busa-Fekete, and Krzysztof Dembczynski. A no-regret generalization of hierarchical softmax to extreme multi-label classification. Advances in neural information processing systems, 31, 2018

    2018

  53. [61]

    Batch is not heavy: Learning word representations from all samples

    Xin Xin, Fajie Yuan, Xiangnan He, and Joemon M Jose. Batch is not heavy: Learning word representations from all samples. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1853--1862, 2018

    2018

  54. [62]

    On the consistency of top-k surrogate losses

    Forest Yang and Sanmi Koyejo. On the consistency of top-k surrogate losses. In International Conference on Machine Learning, pages 10727--10735. PMLR, 2020

    2020

  55. [63]

    Sampling-bias-corrected neural modeling for large corpus item recommendations

    Xinyang Yi, Ji Yang, Lichan Hong, Derek Zhiyuan Cheng, Lukasz Heldt, Aditee Kumthekar, Zhe Zhao, Li Wei, and Ed Chi. Sampling-bias-corrected neural modeling for large corpus item recommendations. In Proceedings of the 13th ACM Conference on Recommender Systems, pages 269--277, 2019

    2019

  56. [64]

    Statistical behavior and consistency of classification methods based on convex risk minimization

    Tong Zhang. Statistical behavior and consistency of classification methods based on convex risk minimization. The Annals of Statistics, 32 0 (1): 0 56--85, 2004

    2004

  57. [65]

    Optimizing top-n collaborative filtering via dynamic negative item sampling

    Weinan Zhang, Tianqi Chen, Jun Wang, and Yong Yu. Optimizing top-n collaborative filtering via dynamic negative item sampling. In Proceedings of the 36th international ACM SIGIR conference on Research and development in information retrieval, pages 785--788, 2013

    2013

  58. [66]

    Learning explicit user interest boundary for recommendation

    Jianhuan Zhuo, Qiannan Zhu, Yinliang Yue, and Yuhong Zhao. Learning explicit user interest boundary for recommendation. In Proceedings of the ACM Web Conference 2022, pages 193--202, 2022

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.