Pith. sign in

REVIEW 3 major objections 7 minor 68 references

A Model-agnostic Strategy to Mitigate Embedding Degradation in Personalized Federated Recommendation

T0 review · 3 major / 7 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A model-agnostic plug-in, PLGC, is claimed to curb embedding degradation in federated recommendation by dynamically mixing local and global item embeddings and decorrelating embedding dimensions, reporting consistent HR@10/NDCG@10 gains of

desk verdict Useful empirical plug-in for FedRec with consistent gains, but the NTK-based theory does not hold up — treat it as a strong empirical paper in need of a reframed derivation. read the letter →

arxiv 2508.19591 v1 pith:GZBIZPPX submitted 2025-08-27 cs.IR cs.DC

Jiakui Shen , Yunqi Mi , Guoshuai Zhao , Jialie Shen , Xueming Qian This is my paper
classification cs.IRcs.DC
keywords federatedrecommendationpersonalizedlearningembeddingdegradationdimensionalcollapseneuraltangentkernelcontrastiveitemlocal-globalcollaboration
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

The paper tries to establish that a single model-agnostic plug-in can fix two failure modes that plague federated recommender systems: item embeddings that do not fit a user's preferences, and embeddings whose dimensions collapse into a low-rank space. The proposed strategy, PLGC, mixes a frozen global item-embedding table with a user's local table at every forward pass, using a convergence-based weight derived from embedding-table traces, and adds a contrastive loss that pushes embedding dimensions to be independent. The authors report that wrapping six existing federated recommenders with PLGC improves Hit Ratio and NDCG by 9.33% to 27.48% across five real-world datasets, with no added communication overhead. If correct, existing federated recommenders can adopt the strategy without redesigning their aggregation or privacy protocols.

What carries the argument

The central object is the personalized item embedding table Q(n) = λc C(n) + λg G, formed at forward inference by a convex combination of the local item embedding table C(n) and the frozen global table G. Its mixing coefficients come from a trace ratio, λc = tr(C(n)C(n)ᵀ) / (tr(C(n)C(n)ᵀ) + tr(GGᵀ)), which the paper justifies via an NTK-style convergence-rate approximation: the trace of the Gram matrix is taken to measure how fast gradient descent reduces the local error. The second load-bearing component is the embedding redundancy reduction (eRR) loss: a projector/predictor pair maps both views into a representation space, a dimensional correlation matrix H is built between local and globa

What would settle it

For a set of clients, compute the trace-based mixing coefficient λc from Eq. (15) and compare it against the mixing coefficient that actually minimizes that client's validation loss over a sweep of possible λ values. If λc does not track the empirical optimal coefficient, or if replacing λc with a fixed coefficient preserves the reported accuracy gains, then the NTK grounding of the dynamic mixing is not doing the claimed work and the improvement must be attributed to the eRR loss alone.

Watch

Extended reading notes

Core claim

PLGC targets two intertwined causes of embedding degradation in federated recommendation: suboptimal personalization and dimensional collapse. For personalization, it keeps the global item embedding table G frozen on each client and generates a personalized item embedding table Q(n) = λc C(n) + λg G during local forward inference, with λc and λg computed from the ratio of the traces of the local and global embedding tables. The paper interprets this trace ratio through a neural-tangent-kernel-style argument as measuring which table converges faster on the user's data, so the mixture adapts per user and per epoch. For dimensional collapse, it adds an embedding redundancy reduction loss that b

Load-bearing premise

The load-bearing premise is that the trace of an item-embedding covariance matrix measures how fast local gradient descent converges, so the ratio of traces correctly decides how much to trust local versus global item embeddings; the paper asserts this identification rather than proving it.

Editorial extensions

If this is right

  • Existing federated recommenders can be upgraded by inserting PLGC without changing the server-side aggregation rule or the communication protocol, since the contrastive network is kept local and the global table is frozen during local training.
  • Personalization becomes dynamic rather than fixed: the mixing coefficient is recomputed each local epoch from current embedding traces, so the balance between local and global information changes as the local model converges.
  • The method is designed to work across matrix-factorization, neural, additive-personalization, and graph-based FedRec backbones, so a single plug-in can be used to upgrade diverse architectures.
  • If dimensional collapse propagates from local clients to the aggregated global item table as the paper argues, reducing local collapse should also keep the shared global embedding healthier over training rounds.
  • Users with sparse interaction histories, who show the largest information-abundance degradation, are the ones expected to benefit most from global information being mixed into their personalized table.

Reading between the lines

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

  • The NTK grounding is likely the softest link: the paper replaces the NTK Gram matrix of training samples with the covariance of the item-embedding table to compute traces, so a sympathetic reading treats the trace ratio as a convergence-motivated heuristic unless that identification is proved.
  • A cheap mechanism test would replace λc from Eq. (15) with a fixed 0.5 mix; if performance barely changes, the reported gains are carried mainly by the eRR loss, not by convergence-based personalization.
  • The eRR loss is structurally a Barlow-Twins-style decorrelation objective; an extension would swap the global embedding view for other anchors (e.g., a batch mean or a momentum table) to test whether the local-global pairing specifically drives the gain.
  • Stratifying the reported gains by user interaction count would test the paper's own story: gains should concentrate on sparse users, and if they instead concentrate on heavy users, the personalization mechanism is not doing what the convergence argument claims.
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

3 major / 7 minor

Summary. This paper proposes Personalized Local-Global Collaboration (PLGC), a model-agnostic plug-in for cross-user federated recommender systems. PLGC targets two failure modes: suboptimal personalization and dimensional collapse. It freezes the global item embedding table G on each client and forms a personalized table Q(n)=λc C(n)+λg G during forward inference, with λc computed from the Frobenius norms of the local and global tables through an NTK-inspired trace argument (Eqs. 7–16). It also adds an Embedding Redundancy Reduction (eRR) loss, a Barlow-Twins-style feature-wise correlation objective (Eqs. 18–19), using G and C(n) as two views. Experiments on five datasets and six FedRec backbones report consistent HR@10/NDCG@10 improvements, ablations for both components, comparisons with alternate collapse-mitigation losses, hyperparameter studies, and convergence curves.

Significance. The empirical contribution is potentially useful: PLGC is simple, model-agnostic, adds no communication overhead, and the reported gains are consistent across six backbones and five datasets. The ablation study supports the utility of both dLGM and eRR, and the comparison with AU, DuoRec, and FedDecorr indicates that the eRR loss is competitive in this setting. However, the paper's stated theoretical grounding—that λc is an NTK-based convergence-optimal mixing ratio—is not established by the present derivation. The method may still work as a heuristic; the claimed 'NTK strategy' and 'optimal personalized embedding table' need to be either proven or explicitly downgraded. No code or reproducibility artifact is mentioned.

major comments (3)
  1. [§IV.B.2–3, Eqs. (7)–(16)] The NTK derivation conflates two different Gram matrices. In Eq. (7), G(e) is the gradient Gram matrix of the prediction function over k training samples, with entries Φ_i^T Φ_j, where Φ_i=∇_θ f(x_i). In §IV.B.3, G_ij is defined from item embedding vectors and Eq. (14) computes tr(G)=||E||_F^2. These are not the same object. For a FedMF-style predictor f_i=p^T q_i, the NTK trace is Σ_{i∈D_n}(||q_i||^2+||p||^2), whereas ||C(n)||_F^2 sums over all M item rows and omits p. The two quantities can rank local and global tables differently, so Eq. (15) does not follow from Eq. (13). The paper needs either a proof that the embedding-table trace is the relevant NTK trace (or a valid approximation), or the NTK-based personalization claim must be removed/reframed.
  2. [§IV.B.2, Eqs. (7)–(13)] Even if the Gram matrix were the correct NTK object, the contraction argument requires assumptions not satisfied here. Eq. (7)–(13) is the linearized squared-loss analysis: the transition from Eq. (12) to Eq. (13) uses (v_i^T ξ)^2 ≈ ||ξ||^2/k for all i, i.e., the stated uniform error assumption; this is not justified for a recommender's BCE loss (Eq. 21), and no lazy-training/constant-G argument is given for the finite models used. Moreover, Q(n) in Eq. (16) is recomputed as C(n) updates, so the prediction function itself changes during the trajectory. The convergence-rate interpretation of λ therefore remains an unvalidated heuristic.
  3. [Algorithm 1 vs. Eq. (6)] There is a discrepancy between the conceptual update in Eq. (6), where Q is formed after a local update Δw_c, and Algorithm 1, where Q is generated once per local epoch and then held fixed across all batches while C(n) is updated. If the implementation follows Algorithm 1, the 'dynamic' mixing is epoch-level, not step-level; if it follows Eq. (6), the pseudocode is incomplete. This should be clarified because it affects reproducibility and the interpretation of the convergence experiments.
minor comments (7)
  1. [Eq. (15)] Missing closing parenthesis in the denominator: should be λc = tr(G(n))/(tr(G(n)) + tr(Gg)). Also, 'trace of a determinant' should be 'trace of a matrix'.
  2. [Eq. (14)] The notation G_ij := e_i e_j^T is ambiguous; use e_i^T e_j (or ⟨e_i, e_j⟩) for the Gram matrix entry.
  3. [Algorithm 1, line 11] The variable B is used both for the batch collection ('Split D_n into batches B') and for the batch size ('of size B'). Rename one of them.
  4. [§V-D] The text says 'study the necessity of each component in RecDCL'; this should be PLGC.
  5. [Abstract / Table II] The claimed improvement range '9.33% to 27.48%' is ambiguous: these are dataset-averaged improvements from the Avg rows of Table II, not bounds across all reported cells. Clarify the wording.
  6. [Table II / §V-A.5] The paper reports averages of five experiments but no standard deviations or significance tests. Given the large improvements this is not disqualifying, but reporting variance would strengthen the empirical claims.
  7. [Throughout] Minor typos: 'personlization', 'Stand for local optimization', 'addictive' (should be 'additive' in §V-A.2), and 'solid personalization' in Fig. 1 are unclear. Also, 'optimal personalized item embedding table' is too strong unless optimality is proven.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; the NTK-to-trace substitution is an asserted assumption, not a self-referential reduction.

full rationale

The paper's central derivation chain is not circular. The personalized mixing coefficient λc is computed directly from the current embedding tables via Eqs. (14)-(15) and then evaluated on held-out interactions; it is not fitted to the reported HR/NDCG values, and no parameter estimated on the evaluation target is renamed as a prediction. The eRR objective (Eq. 19) is a redundancy-reduction loss over two views (local and global) with no label-based target, and improvements are measured by external ranking metrics. The only load-bearing theoretical step—identifying tr(EE^T) in Eq. (14) with the NTK Gram-matrix trace in Eq. (7)—is an asserted substitution rather than a derivation of the empirical result from itself; this is a soundness/correctness weakness, not a circular reduction. The paper contains no self-citation chain: the NTK, IA, and dimensional-collapse references are external prior work, and the 'first in FedRec' claim is a novelty assertion, not an imported uniqueness theorem. Therefore, no step in the claimed derivation is equivalent to its own inputs by construction.

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

No new physical or mathematical objects are postulated. Q(n) and eRR are internal method components, not independent entities with external falsifiable handles.

free parameters (4)
  • beta (eRR trade-off) = tuned on validation, range 0.1 to 1.0
    Balances recommendation loss Lrec and redundancy reduction loss LeRR in Eq. 22; Fig. 7 shows sensitivity varies by backbone.
  • gamma (inter-correlation coefficient) = tuned on validation, range 0.001 to 0.1
    Weights the off-diagonal penalty in LeRR in Eq. 19; Fig. 8 shows low sensitivity.
  • embedding dimension d = 32
    Chosen by hand for all experiments; embedding capacity affects the trace-based mixing coefficients and the redundancy loss.
  • learning rate and ExponentialLR decay = picked via grid search
    Local optimizer hyperparameters are chosen per dataset and backbone, but exact values are not reported.
assumptions (4)
  • ad hoc to paper The error vector xi is uniformly distributed in R^k.
    Stated in Section IV.B.2 to convert Eq. 11 into Eq. 13; without this assumption the trace cannot replace the eigenvalue-weighted sum.
  • ad hoc to paper The trace of the item embedding table, tr(EE^T), is the relevant NTK Gram matrix trace for the client's convergence rate.
    Eq. 14 computes tr(G) from embedding columns, but no proof links this to the gradient Gram matrix in Eq. 7.
  • domain assumption Pushing the cross-view correlation matrix toward the identity improves recommendation embedding utility.
    Adopted from Barlow Twins [61]; the FedRec-specific benefit is assumed and only indirectly validated by ablations.
  • domain assumption Frozen global table G prevents information abundance loss, while updating G during backward would degrade it.
    The paper asserts this based on Fig. 3 but does not test the alternative of updating G locally.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Model-agnostic Strategy to Mitigate Embedding Degradation in Personalized Federated Recommendation." pith.science (2026). https://pith.science/paper/GZBIZPPX

@misc{pith2026250819591,
  author       = {Pith},
  title        = {Pith review of: A Model-agnostic Strategy to Mitigate Embedding Degradation in Personalized Federated Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GZBIZPPX}},
  note         = {Machine review of arXiv:2508.19591}
}
read the original abstract

Centralized recommender systems encounter privacy leakage due to the need to collect user behavior and other private data. Hence, federated recommender systems (FedRec) have become a promising approach with an aggregated global model on the server. However, this distributed training paradigm suffers from embedding degradation caused by suboptimal personalization and dimensional collapse, due to the existence of sparse interactions and heterogeneous preferences. To this end, we propose a novel model-agnostic strategy for FedRec to strengthen the personalized embedding utility, which is called Personalized Local-Global Collaboration (PLGC). It is the first research in federated recommendation to alleviate the dimensional collapse issue. Particularly, we incorporate the frozen global item embedding table into local devices. Based on a Neural Tangent Kernel strategy that dynamically balances local and global information, PLGC optimizes personalized representations during forward inference, ultimately converging to user-specific preferences. Additionally, PLGC carries on a contrastive objective function to reduce embedding redundancy by dissolving dependencies between dimensions, thereby improving the backward representation learning process. We introduce PLGC as a model-agnostic personalized training strategy for federated recommendations that can be applied to existing baselines to alleviate embedding degradation. Extensive experiments on five real-world datasets have demonstrated the effectiveness and adaptability of PLGC, which outperforms various baseline algorithms.

Figures

Figures reproduced from arXiv: 2508.19591 by the authors.

Figure 1
Figure 1. Comparison of three different architectures of the [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of personalized local-global collaboration in federated recommendation. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. IA of item embedding table in FedRec, which is [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The workflow of the Contrastive Learning network. these methods are not designed for FedRec tasks. Interaction isolation restricts the appliance of contrastive learning methods [23], [27] based on InfoNCE [60]. Barlow Twins (BT) [61] achieves invariance by generating t…
Figure 5
Figure 5. Figure 5: IA of item embedding table in PLGC, where [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Analysis of dimensional collapse mitigation of PLGC with SVD (a, e) and t-SNE visualization (b-d, f-h). The SVD [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Influence of different β on Douban-Book. (a) FedMF w/Ours (b) FedRAP w/Ours [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Influence of different γ on Douban-Book. small value of γ provides a strong enough signal to the model to significantly reduce the inter-correlations (e.g. Hij , i ̸= j). Once the dimensions are reasonably decorrelated, increasing γ further forces the off-diagonal elem…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

68 extracted references · 65 canonical work pages

  1. [1]

    Daisyrec 2.0: Benchmarking recommendation for rigorous evaluation,

    Z. Sun, H. Fang, J. Yang, X. Qu, H. Liu, D. Yu, Y .-S. Ong, and J. Zhang, “Daisyrec 2.0: Benchmarking recommendation for rigorous evaluation,” TPAMI, 2023

    work page 2023

  2. [2]

    Adaptive fair representation learning for personalized fairness in recommendations via information align- ment,

    X. Zhu, L. Zhang, and N. Yang, “Adaptive fair representation learning for personalized fairness in recommendations via information align- ment,” in SIGIR, 2024

    work page 2024

  3. [3]

    Mitigating confounding bias in practical recommender systems with partially inaccessible exposure status,

    T. Cao, Q. Xu, Z. Yang, and Q. Huang, “Mitigating confounding bias in practical recommender systems with partially inaccessible exposure status,” TPAMI, 2024

    work page 2024

  4. [4]

    A survey on federated recommendation systems,

    Z. Sun, Y . Xu, Y . Liu, W. He, L. Kong, F. Wu, Y . Jiang, and L. Cui, “A survey on federated recommendation systems,” TNNLS, 2024

    work page 2024

  5. [5]

    Horizontal federated recommender system: A survey,

    L. Wang, H. Zhou, Y . Bao, X. Yan, G. Shen, and X. Kong, “Horizontal federated recommender system: A survey,” ACM Computing Surveys , 2024

    work page 2024

  6. [6]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in AISTATS, 2017

    work page 2017

  7. [7]

    Federated learning for generalization, robustness, fairness: A survey and benchmark,

    W. Huang, M. Ye, Z. Shi, G. Wan, H. Li, B. Du, and Q. Yang, “Federated learning for generalization, robustness, fairness: A survey and benchmark,” TPAMI, 2024

    work page 2024

  8. [8]

    Federated collaborative filtering for privacy-preserving personalized recommendation system,

    M. Ammad-Ud-Din, E. Ivannikova, S. A. Khan, W. Oyomno, Q. Fu, K. E. Tan, and A. Flanagan, “Federated collaborative filtering for privacy-preserving personalized recommendation system,” arXiv preprint arXiv:1901.09888, 2019

    arXiv 1901

Show all 68 references
  1. [9]

    Fedgnn: Federated graph neural network for privacy-preserving recommendation,

    C. Wu, F. Wu, Y . Cao, Y . Huang, and X. Xie, “Fedgnn: Federated graph neural network for privacy-preserving recommendation,” in KDD, 2021

    2021

  2. [10]

    Lightfr: Lightweight federated recommendation with privacy-preserving matrix factorization,

    H. Zhang, F. Luo, J. Wu, X. He, and Y . Li, “Lightfr: Lightweight federated recommendation with privacy-preserving matrix factorization,” TOIS, 2023

    2023

  3. [11]

    Personalized latent structure learning for recommendation,

    S. Zhang, F. Feng, K. Kuang, W. Zhang, Z. Zhao, H. Yang, T.-S. Chua, and F. Wu, “Personalized latent structure learning for recommendation,” TPMAI, 2023. 12

    2023

  4. [12]

    A survey on accuracy-oriented neural recommendation: From collaborative filtering to information-rich recommendation,

    L. Wu, X. He, X. Wang, K. Zhang, and M. Wang, “A survey on accuracy-oriented neural recommendation: From collaborative filtering to information-rich recommendation,” TKDE, 2023

    2023

  5. [13]

    Language representations can be what recommenders need: Findings and potentials,

    L. Sheng, A. Zhang, Y . Zhang, Y . Chen, X. Wang, and T.-S. Chua, “Language representations can be what recommenders need: Findings and potentials,” in ICLR, 2025

    2025

  6. [14]

    Secure federated matrix factorization,

    D. Chai, L. Wang, K. Chen, and Q. Yang, “Secure federated matrix factorization,” IEEE Intelligent Systems , 2021

    2021

  7. [15]

    Federated neural collaborative filter- ing,

    V . Perifanis and P. S. Efraimidis, “Federated neural collaborative filter- ing,” KBS, 2022

    2022

  8. [16]

    Dual personalization on federated recommendation,

    C. Zhang, G. Long, T. Zhou, P. Yan, Z. Zhang, C. Zhang, and B. Yang, “Dual personalization on federated recommendation,” in IJCAI, 2023

    2023

  9. [17]

    Marking the pace: A blockchain-enhanced privacy-traceable strategy for federated recom- mender systems,

    Z. Cai, T. Tang, S. Yu, Y . Xiao, and F. Xia, “Marking the pace: A blockchain-enhanced privacy-traceable strategy for federated recom- mender systems,” IOTJ, 2024

    2024

  10. [18]

    Personalized federated recommendation via joint representation learning, user clustering, and model adaptation,

    S. Luo, Y . Xiao, and L. Song, “Personalized federated recommendation via joint representation learning, user clustering, and model adaptation,” in CIKM, 2022

    2022

  11. [19]

    Cluster-driven personalized federated recommendation with interest-aware graph convolution network for multimedia,

    X. Mao, Y . Liu, L. Qi, L. Duan, X. Xu, X. Zhang, W. Dou, A. Beheshti, and X. Zhou, “Cluster-driven personalized federated recommendation with interest-aware graph convolution network for multimedia,” in ACM MM, 2024

    2024

  12. [20]

    Federated recommendation with additive personalization,

    Z. Li, G. Long, and T. Zhou, “Federated recommendation with additive personalization,” in ICLR, 2024

    2024

  13. [21]

    Gpfedrec: Graph-guided personalization for federated recommendation,

    C. Zhang, G. Long, T. Zhou, Z. Zhang, P. Yan, and B. Yang, “Gpfedrec: Graph-guided personalization for federated recommendation,” in KDD, 2024

    2024

  14. [22]

    Disentangled representation learning for recommendation,

    X. Wang, H. Chen, Y . Zhou, J. Ma, and W. Zhu, “Disentangled representation learning for recommendation,” TPAMI, 2023

    2023

  15. [23]

    Are graph augmentations necessary? simple graph contrastive learning for recommendation,

    J. Yu, H. Yin, X. Xia, T. Chen, L. Cui, and Q. V . H. Nguyen, “Are graph augmentations necessary? simple graph contrastive learning for recommendation,” in SIGIR, 2022

    2022

  16. [24]

    Contrastive learning for representation degeneration problem in sequential recommendation,

    R. Qiu, Z. Huang, H. Yin, and Z. Wang, “Contrastive learning for representation degeneration problem in sequential recommendation,” in WSDM, 2022

    2022

  17. [25]

    Recdcl: Dual contrastive learning for recommendation,

    D. Zhang, Y . Geng, W. Gong, Z. Qi, Z. Chen, X. Tang, Y . Shan, Y . Dong, and J. Tang, “Recdcl: Dual contrastive learning for recommendation,” in WWW, 2024

    2024

  18. [26]

    Understand- ing and mitigating dimensional collapse in federated learning,

    Y . Shi, J. Liang, W. Zhang, C. Xue, V . Y . Tan, and S. Bai, “Understand- ing and mitigating dimensional collapse in federated learning,” TPAMI, 2023

    2023

  19. [27]

    Relaxed contrastive learning for federated learning,

    S. Seo, J. Kim, G. Kim, and B. Han, “Relaxed contrastive learning for federated learning,” in CVPR, 2024

    2024

  20. [28]

    Fedloge: Joint local and generic federated learning under long-tailed data,

    Z. Xiao, Z. Chen, L. Liu, Y . FENG, J. T. Zhou, J. Wu, W. Liu, H. H. Yang, and Z. Liu, “Fedloge: Joint local and generic federated learning under long-tailed data,” in ICLR, 2024

    2024

  21. [29]

    Mitigating the popularity bias of graph collaborative filtering: A dimensional collapse perspective,

    Y . Zhang, H. Zhu, Z. Song, P. Koniusz, I. King et al., “Mitigating the popularity bias of graph collaborative filtering: A dimensional collapse perspective,” in NeurIPS, 2023

    2023

  22. [30]

    Lower bounds and optimal algorithms for personalized federated learning,

    F. Hanzely, S. Hanzely, S. Horv ´ath, and P. Richt ´arik, “Lower bounds and optimal algorithms for personalized federated learning,” in NeurIPS, 2020

    2020

  23. [31]

    Personalized federated learning under mixture of distribu- tions,

    Y . Wu, S. Zhang, W. Yu, Y . Liu, Q. Gu, D. Zhou, H. Chen, and W. Cheng, “Personalized federated learning under mixture of distribu- tions,” in ICML, 2023

    2023

  24. [33]

    Personalized federated learning on non-iid data via group-based meta-learning,

    L. Yang, J. Huang, W. Lin, and J. Cao, “Personalized federated learning on non-iid data via group-based meta-learning,” TKDD, 2023

    2023

  25. [34]

    Fedbabu: Toward enhanced representation for federated image classification,

    J. Oh, S. Kim, and S.-Y . Yun, “Fedbabu: Toward enhanced representation for federated image classification,” in ICLR, 2022

    2022

  26. [35]

    Fedpop: A bayesian approach for personalised federated learning,

    N. Y . Kotelevskii, M. V ono, A. Durmus, and E. Moulines, “Fedpop: A bayesian approach for personalised federated learning,” in NeurIPS, 2022

    2022

  27. [36]

    A statistical framework for personalized federated learning and estimation: Theory, algorithms, and privacy,

    K. Ozkara, A. M. Girgis, D. Data, and S. N. Diggavi, “A statistical framework for personalized federated learning and estimation: Theory, algorithms, and privacy,” in ICLR, 2023

    2023

  28. [37]

    Adaptive personalized federated learning,

    Y . Deng, M. M. Kamani, and M. Mahdavi, “Adaptive personalized federated learning,” arXiv preprint arXiv:2003.13461 , 2020

    2003 arXiv

  29. [38]

    Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach,

    A. Fallah, A. Mokhtari, and A. Ozdaglar, “Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach,” in NeurIPS, 2020

    2020

  30. [39]

    Exploiting shared representations for personalized federated learning,

    L. Collins, H. Hassani, A. Mokhtari, and S. Shakkottai, “Exploiting shared representations for personalized federated learning,” in ICML, 2021

    2021

  31. [40]

    Heterogeneous personalized federated learning by local-global updates mixing via convergence rate,

    M. Jiang, A. Le, X. Li, and Q. Dou, “Heterogeneous personalized federated learning by local-global updates mixing via convergence rate,” in ICLR, 2024

    2024

  32. [41]

    Refer: Retrieval-enhanced vertical federated recommenda- tion for full set user benefit,

    W. Li, Z. Wang, J. Wang, S.-T. Xia, J. Zhu, M. Chen, J. Fan, J. Cheng, and J. Lei, “Refer: Retrieval-enhanced vertical federated recommenda- tion for full set user benefit,” in SIGIR, 2024

    2024

  33. [42]

    Towards efficient communication and secure federated recommendation system via low-rank training,

    N.-H. Nguyen, T.-A. Nguyen, T. Nguyen, V . T. Hoang, D. D. Le, and K.-S. Wong, “Towards efficient communication and secure federated recommendation system via low-rank training,” in WWW, 2024

    2024

  34. [43]

    Fine-grained preference-aware personalized federated poi recommendation with data sparsity,

    X. Zhang, Z. Ye, J. Lu, F. Zhuang, Y . Zheng, and D. Yu, “Fine-grained preference-aware personalized federated poi recommendation with data sparsity,” in SIGIR, 2023

    2023

  35. [44]

    Privacy-preserving sequential recommendation with collaborative confusion,

    W. Wang, Y . Lin, P. Ren, Z. Chen, T. Mine, J. Zhao, Q. Zhao, M. Zhang, X. Ben, and Y . Li, “Privacy-preserving sequential recommendation with collaborative confusion,” TOIS, 2025

    2025

  36. [45]

    Hfsa: A semi-asynchronous hierarchical federated recommendation system in smart city,

    Y . Li, H. Yu, Y . Zeng, and Q. Pan, “Hfsa: A semi-asynchronous hierarchical federated recommendation system in smart city,” IOTJ, 2023

    2023

  37. [46]

    Fedcore: Federated learning for cross-organization recommen- dation ecosystem,

    Z. Li, X. Wu, W. Pan, Y . Ding, Z. Wu, S. Tan, Q. Xu, Q. Yang, and Z. Ming, “Fedcore: Federated learning for cross-organization recommen- dation ecosystem,” TKDE, 2024

    2024

  38. [47]

    Federated graph learning for cross-domain recommenda- tion,

    Z. Yang, Z. Peng, Z. Wang, J. Qi, C. Chen, W. Pan, C. Wen, C. Wang, and X. Fan, “Federated graph learning for cross-domain recommenda- tion,” arXiv preprint arXiv:2410.08249 , 2024

    2024 arXiv

  39. [48]

    FedCSR: A federated framework for multi-platform cross-domain sequential recommendation with dual contrastive learning,

    D. Zheng, H. Zhang, J. Zhai, L. Zhong, L. Wang, J. Feng, X. Liao, Y . Tian, N. Xiao, and Q. Liao, “FedCSR: A federated framework for multi-platform cross-domain sequential recommendation with dual contrastive learning,” in COLING, 2025

    2025

  40. [49]

    Not one less: Exploring interplay between user profiles and items in untargeted attacks against federated recommendation,

    Y . Hao, X. Chen, X. Lyu, J. Liu, Y . Zhu, Z. Wan, S. Mauw, and W. Wang, “Not one less: Exploring interplay between user profiles and items in untargeted attacks against federated recommendation,” in CCS, 2024

    2024

  41. [50]

    Defending against membership inference attack for counterfactual federated recommendation with differentially private representation learning,

    X. Liu, Y . Chen, and S. Pang, “Defending against membership inference attack for counterfactual federated recommendation with differentially private representation learning,” TIFS, 2024

    2024

  42. [51]

    Federated recommender system based on diffusion augmentation and guided denoising,

    Y . Di, H. Shi, X. Wang, R. Ma, and Y . Liu, “Federated recommender system based on diffusion augmentation and guided denoising,” TOIS, 2025

    2025

  43. [52]

    On the embedding collapse when scaling up recommendation models,

    X. Guo, J. Pan, X. Wang, B. Chen, J. Jiang, and M. Long, “On the embedding collapse when scaling up recommendation models,” in ICML, 2024

    2024

  44. [53]

    Neural tangent kernel: conver- gence and generalization in neural networks,

    A. Jacot, F. Gabriel, and C. Hongler, “Neural tangent kernel: conver- gence and generalization in neural networks,” in STOC, 2021

    2021

  45. [54]

    On exact computation with an infinitely wide neural net,

    S. Arora, S. S. Du, W. Hu, Z. Li, R. Salakhutdinov, and R. Wang, “On exact computation with an infinitely wide neural net,” in NeurIPS, 2019

    2019

  46. [55]

    Generalized leverage score sampling for neural networks,

    J. D. Lee, R. Shen, Z. Song, M. Wang, and z. Yu, “Generalized leverage score sampling for neural networks,” in NeurIPS, 2020

    2020

  47. [56]

    Gradient descent provably optimizes over-parameterized neural networks,

    S. S. Du, X. Zhai, B. Poczos, and A. Singh, “Gradient descent provably optimizes over-parameterized neural networks,” in ICLR, 2019

    2019

  48. [57]

    Dynamics of deep neural networks and neural tangent hierarchy,

    J. Huang and H.-T. Yau, “Dynamics of deep neural networks and neural tangent hierarchy,” in ICML, 2020

    2020

  49. [58]

    Over- parameterized adversarial training: An analysis overcoming the curse of,

    Y . Zhang, O. Plevrakis, S. S. Du, X. Li, Z. Song, and S. Arora, “Over- parameterized adversarial training: An analysis overcoming the curse of,” in NeurIPS, 2020

    2020

  50. [59]

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere,

    T. Wang and P. Isola, “Understanding contrastive representation learning through alignment and uniformity on the hypersphere,” in ICML, 2020

    2020

  51. [60]

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

    M. Gutmann and A. Hyv ¨arinen, “Noise-contrastive estimation: A new estimation principle for unnormalized statistical models,” in AISTATS, 2010

    2010

  52. [61]

    Barlow twins: Self-supervised learning via redundancy reduction,

    J. Zbontar, L. Jing, I. Misra, Y . LeCun, and S. Deny, “Barlow twins: Self-supervised learning via redundancy reduction,” in ICML, 2021

    2021

  53. [62]

    Zero-cl: Instance and feature decorrelation for negative-free symmetric contrastive learning,

    S. Zhang, F. Zhu, J. Yan, R. Zhao, and X. Yang, “Zero-cl: Instance and feature decorrelation for negative-free symmetric contrastive learning,” in ICLR, 2022

    2022

  54. [63]

    Cl4ctr: A contrastive learning framework for ctr prediction,

    F. Wang, Y . Wang, D. Li, H. Gu, T. Lu, P. Zhang, and N. Gu, “Cl4ctr: A contrastive learning framework for ctr prediction,” in WSDM, 2023

    2023

  55. [64]

    Neural collaborative filtering,

    X. He, L. Liao, H. Zhang, L. Nie, X. Hu, and T.-S. Chua, “Neural collaborative filtering,” in WWW, 2017

    2017

  56. [65]

    Item-based top-n recommendation algorithms,

    M. Deshpande and G. Karypis, “Item-based top-n recommendation algorithms,” TOIS, 2004

    2004

  57. [66]

    Trirank: Review-aware explainable recommendation by modeling aspects,

    X. He, T. Chen, M.-Y . Kan, and X. Chen, “Trirank: Review-aware explainable recommendation by modeling aspects,” in CIKM, 2015

    2015

  58. [67]

    A stochastic approximation method,

    H. Robbins and S. Monro, “A stochastic approximation method,” The annals of mathematical statistics , 1951

    1951

  59. [68]

    An exponential learning rate schedule for deep learning,

    Z. Li and S. Arora, “An exponential learning rate schedule for deep learning,” in ICLR, 2020

    2020

  60. [69]

    Visualizing data using t-sne,

    L. Van der Maaten and G. Hinton, “Visualizing data using t-sne,” JMLR, 2008

    2008

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.