Pith. sign in

REVIEW 3 major objections 7 minor 45 references

Successive Point-of-Interest Recommendation with Local Differential Privacy

T0 review · 3 major / 7 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Jointly factorizing user-POI visit counts and noisy POI-to-POI transitions improves next-POI recommendation accuracy under local differential privacy.

desk verdict SPIREL is a useful engineering contribution in private successive POI recommendation, but the Step 1 privacy analysis is wrong: the one-hot transition report is 2ε1-LDP, not ε1-LDP, so the paper's equal-ε comparison to SD-PL is not apples-to-apples and the 'stronger privacy' claim is unsupported. read the letter →

arxiv 1908.09485 v2 pith:LTFLBB6I submitted 2019-08-26 cs.IR cs.CRcs.CY

Jong Seon Kim , Jong Wook Kim , Yon Dohn Chung This is my paper · ORCID
classification cs.IRcs.CRcs.CY
keywords successivePOIrecommendationlocaldifferentialprivacymatrixfactorizationtransferlearningtransitionpatternrandomizedresponseimplicitfeedbacklocation
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

SPIREL is a framework for recommending a user's next point of interest without a server ever seeing raw check-ins or the user's current location. The paper's central claim is that jointly factorizing the user-POI matrix of visit counts and a POI-POI matrix of noisy transition patterns produces better top-k next-POI recommendations than a private single-domain matrix-factorization recommender. The argument is that the shared POI latent matrix lets population-level movement trends, collected under local differential privacy, compensate for the extreme sparsity of individual visit histories. Experiments on four public datasets show the private cross-domain method beating the private single-domain baseline by roughly 41.6% in hit rate and 38.8% in mean reciprocal rank, and even beating the non-private single-domain method on the sparsest dataset.

What carries the argument

The central object is the shared POI latent matrix $V$, factorized in both the user-POI term and the POI-POI term. The POI-POI matrix $Q$ is built by having each user sample one consecutive transition, randomize it with optimized randomized response, and letting the server debias the aggregate; the result is turned into a confidence score $1+\sigma(Q[i][j])$ so that negative debiased counts remain usable. The user profile vectors are updated locally by ALS using a precomputed $V(V^\top V+\lambda I)^{-1}$, while the POI profiles are updated by the server with SGD on aggregated perturbed gradients, with users partitioned so that each participates in one learning iteration. The mechanism that carries the argument is the ranking score $u_i^\top v_k + v_j^\top v_k$: the first term is personal preference, and the second term transfers the noisy aggregate transition knowledge into each recommendation.

What would settle it

Train SPIREL with the same privacy budget and pipeline but replace the true transition data in $Q$ with random or shuffled transitions; if hit rate and mean reciprocal rank stay the same, the transition term is contributing nothing and the reported gains come from elsewhere. A second check is to remove the $v_j^\top v_k$ term from the ranking score while still training on $Q$; if performance does not drop, the transfer mechanism itself is not the carrier of the improvement.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a single POI latent matrix $V$ can carry both static and dynamic knowledge. The objective function $\mathcal{L}_{\mathrm{SPIREL}} = \|P - U V^\top\|^2 + \|Q - V V^\top\|^2 + \lambda(\|U\|^2 + \|V\|^2)$ forces $V$ to explain both the users' visit-count preferences and the first-order Markov transition patterns between consecutive POIs, where $Q$ is built from one perturbed transition sampled per user and mapped through a sigmoid confidence score. With the learned factors, the preference of user $i$ currently at POI $j$ for next POI $k$ is $u_i^\top v_k + v_j^\top v_k$, so the ranking can be computed locally. The authors claim this integrated learning is what makes the private framework outperform the single-domain private baseline by on average 41.56% in HR@k and 38.84% in MRR across the four datasets, and the whole pipeline is $\epsilon$-LDP by sequential composition because each user sends only perturbed bits and one perturbed gradient entry per iteration.

Load-bearing premise

Everything rests on the empirical assumption that a noisy, user-agnostic aggregate of one sampled transition per user carries information about an individual's next POI choice, so that sharing $V$ between the two factorizations improves the ranking $u_i^\top v_k + v_j^\top v_k$.

Editorial extensions

If this is right

  • A privacy-preserving successive POI recommender can be built without a trusted server, without raw check-in collection, and without users disclosing their current location.
  • On very sparse check-in datasets, noisy population-level transition knowledge can compensate for missing individual visit history well enough to beat a non-private single-domain recommender.
  • The framework is modular in its privacy mechanisms: replacing the gradient perturbation with the piecewise mechanism yields comparable results, so future higher-accuracy LDP mechanisms can be dropped in without redesigning the model.
  • Privacy-budget allocation and POI-domain size interact: small POI domains favor spending budget on gradient perturbation, while sparse large domains favor spending it on transition patterns.
  • The server-side cost scales with $O(mn^2)$ for building the transition matrix, so the number of POIs is the main scaling bottleneck.

Reading between the lines

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

  • A natural extension is to let each user report more than one sampled transition under a split privacy budget; the current design deliberately uses one sample per user, so the marginal value of additional samples is a testable question the paper leaves open.
  • The same shared-latent construction could carry over to private sequential recommendation in other implicit-feedback domains, such as app usage or purchase sequences, where a global transition matrix can be factorized with the user-item matrix while raw events stay on-device.
  • The confidence mapping from debiased transition counts to $1+\sigma(\cdot)$ is a design choice; comparing it with logarithmic or linear confidence scores would reveal how much of the reported gain comes from the mapping rather than from sharing the latent POI matrix.
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

3 major / 7 minor

Summary. The paper proposes SPIREL, a local-differential-privacy (LDP) framework for successive point-of-interest (POI) recommendation. Each user samples one consecutive POI-POI transition from his or her check-in history, encodes it as a one-hot bit string of length n^2, and perturbs it with optimized randomized response; the server aggregates the noisy strings into a POI-POI matrix Q and converts the estimated counts into sigmoid confidence scores. SPIREL jointly factorizes the user-POI visit-count matrix P and the POI-POI matrix Q with a shared POI latent matrix V by minimizing ||P - U V^T||^2 + ||Q - V V^T||^2 plus regularization. Users update their latent vectors locally via ALS, while each user contributes one perturbed gradient coordinate for one randomly selected POI; the server aggregates these noisy gradients and updates V with Adam. The next-POI score is u_i^T v_k + v_j^T v_k, so users do not need to disclose their current location. Experiments on Gowalla, Taxi, Yelp, and Foursquare compare SPIREL with a single-domain baseline, a private single-domain baseline (SD-PL), a non-private cross-domain version, a piecewise-mechanism variant, and item-based collaborative filtering, using HR@k and MRR.

Significance. If the privacy accounting is corrected, the transfer-learning idea is a plausible contribution: SPIREL appears to be the first LDP successive-POI recommender that combines user-POI and POI-POI relationships, and it has the practical advantage that users can receive recommendations without sending their current location. The paper gives a complete algorithmic description and evaluates on four public datasets, and the private-vs-private comparison consistently favors SPIREL. However, the central privacy analysis currently overstates the guarantee: the one-hot transition report is 2*epsilon1-LDP, not epsilon1-LDP, so the total privacy cost is epsilon+epsilon1 rather than epsilon. The experimental comparison to SD-PL at equal nominal epsilon is therefore inequitable, and the headline claim of "stronger privacy protection" is not supported. The manuscript is not circular: Q is constructed from training check-ins, which is standard training data usage, and the cited LDP mechanisms are by other groups.

major comments (3)
  1. [Section IV-F.1, Algorithms 1-2] The claim that the one-hot ORR reporting step satisfies epsilon1-LDP is incorrect. For the ORR mechanism of Algorithm 1, p = e^epsilon1 / (1 + e^epsilon1) and q = 1 / (1 + e^epsilon1). For two distinct transition patterns a and b, consider an output in which bit a is 1 and bit b is 0. The likelihood ratio is Pr[output | input = a] / Pr[output | input = b] = [p(1 - q)] / [q(1 - p)] = e^(2*epsilon1). Changing the sampled transition flips two coordinates of the one-hot n^2-bit vector, each contributing a factor e^epsilon1. Therefore Algorithm 2 is 2*epsilon1-LDP, not epsilon1-LDP. Combined with the gradient perturbation in Step 3, the full protocol guarantees 2*epsilon1 + epsilon2-LDP, which equals epsilon + epsilon1-LDP under the stated split epsilon1 + epsilon2 = epsilon. This is a formal error in the core privacy analysis, not a wording issue, and it invalidates the abstract's claim of "stronger privacy protection" relative to SD-PL at the same nominal epsilon.
  2. [Section V-B.1, Figs. 5-6] Because of the privacy-accounting error above, the reported average improvements of 41.56% in HR@k and 38.84% in MRR of SPIREL over SD-PL are not a clean test of the modeling contribution: SPIREL at nominal epsilon consumes epsilon + epsilon1 of true privacy budget while SD-PL consumes epsilon. The authors should either correct the transition-reporting mechanism to genuinely achieve epsilon1-LDP (for example, by using generalized randomized response over the n^2-sized domain) or rerun SD-PL with budget epsilon + epsilon1, and then compare the methods at equal true privacy loss. Without this, the observed gains may partly reflect additional privacy budget rather than the benefit of transferring the POI-POI knowledge.
  3. [Section IV-A and Eq. (5)] The transfer-learning assumption is load-bearing: the paper provides no theoretical argument that the global, user-independent transition-frequency matrix Q, estimated from one sampled transition per user, carries information that improves an individual's next-POI ranking u_i^T v_k + v_j^T v_k. The entire advantage over SD-PL rests on this empirical assumption. I recommend adding an ablation in which Q is replaced by a non-informative or random matrix, and comparing with non-private sequential recommenders (e.g., FPMC or PRME) to calibrate whether the benefit is real and practically meaningful. The current evaluation only measures SPIREL against self-created baselines and one private baseline, which limits external validity.
minor comments (7)
  1. [Section V-B.2 and Fig. 12 captions] There are typos: "Foursuqare" in Section V-B.2 should be "Foursquare," and the caption of Figure 12 contains "T raining time" and "Mean traning time."
  2. [Section IV-E] The phrase "matrix conversion" should be "matrix inversion" when referring to (V^T V + lambda I)^{-1}.
  3. [Section II-A, Eq. (2)] Equation (2) uses the notation "v_i" in the gradient with respect to the user profile vector; this should be v_j to match the sum over items, and the subscript in the gradient notation should be consistent.
  4. [Section IV-F.1] The sentence "each bit is relevant to the independent transition pattern" is misleading: independence of the identities of the patterns does not imply independence of the privacy loss when the input is a one-hot vector. Section IV-F.1 should be revised along with the corrected privacy analysis.
  5. [Section V-A.4 and Fig. 9] The default equal privacy-budget allocation 5:5 is used for the headline comparisons, but Figure 9 shows that the best allocation ratio is dataset-dependent. The authors should either justify the fixed default or report results at each dataset's optimal ratio.
  6. [Section V-B, Figs. 5-6] Although the experiments are averaged over 50 runs, Figures 5 and 6 show only point values without error bars or standard deviations; adding these would strengthen the claim that the differences between methods are meaningful.
  7. [Section VI, reference [45]] The text cites "Gao et al." for the method that uses Factorization Machines, but reference [45] lists C. Chen, J. Zhou, B. Wu, et al.; the citation should be corrected to match the reference entry.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: SPIREL's ranking is learned from training data and evaluated on held-out last check-ins, with no parameter fit to the test labels.

full rationale

The claimed derivation chain is not circular. The auxiliary POI-POI matrix Q is constructed from each user's sampled transition pattern (Algorithm 2), perturbed under ORR and aggregated (Algorithm 3); the user-POI matrix P is constructed from visit counts. The objective function (Eq. 5) is a design choice that jointly factorizes these two training-derived matrices, and the ranking score (Eq. 12) is the inner-product composition of the learned latent factors. No parameter in this chain is fitted to the held-out last check-ins used for HR@k and MRR: the held-out labels enter only at evaluation, and the model parameters are learned from the remaining check-ins. The cited LDP mechanisms ([10], [13], [25], [30]) are by other research groups, so there is no load-bearing self-citation. The privacy analysis's claim that Step 1 is ε1-LDP (Section IV-F.1) may be incorrect—for a one-hot bit string the ORR likelihood ratio is e^{2ε1} per pair of outputs—but that is a privacy-accounting and correctness issue, not a circularity of the recommendation derivation. The performance gains are therefore empirically contingent results, not consequences of the inputs by construction.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the transfer-learning premise that a noisy aggregate transition matrix helps individual next-POI ranking, plus the standard correctness of the LDP mechanisms and MF optimization. There are six free parameters, most of which are chosen empirically without sensitivity analysis.

free parameters (6)
  • privacy budget allocation ratio epsilon1:epsilon2 = 1:1 by default; tuned to 9:1 or 1:9 per dataset in Section V-B3
    The split between transition-pattern perturbation and gradient perturbation is a modeling choice that materially changes HR@5; optimal ratio varies by dataset.
  • learning rate gamma = 0.01 for SPIREL/SPIREL-PM/CD, 0.001 for SD/SD-PL
    Different learning rates per method were used; no sensitivity analysis is given for the baselines.
  • maximum iterations = 20 by default, tested 5-40
    Iterations consume privacy budget; performance peaks at different iteration counts per dataset.
  • profile dimension d = 40
    Set to 40 for all methods without ablation.
  • clipping bound for gradients = [-1, 1]
    Gradient values are manually clipped to this range before perturbation; the choice affects bias and variance but is not analyzed.
  • regularization lambda = 1e-4
    Set in Section V-A4; no sensitivity study.
assumptions (5)
  • domain assumption The next POI depends only on the current POI (first-order Markov chain).
    Section IV-A builds the POI-POI matrix from two-hop transitions only.
  • domain assumption A single shared latent POI matrix V can bridge the user-POI and POI-POI domains.
    Eq. 5 assumes V factorizes both P and Q.
  • domain assumption An observed transition, even if fake due to noise, indicates 'more preferable' than non-observed feedback.
    Section IV-B2 follows [28] on implicit feedback.
  • standard math The closed-form ALS updates and SGD are standard and converge for the fixed-point of the objective.
    Used in Section IV-C to derive Eqs. 6 and 7.
  • domain assumption The ORR scheme [25] and the Nguyen et al. mechanism [10] satisfy the stated epsilon-LDP guarantees.
    Section IV-F relies on these guarantees; the paper does not re-prove them.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Successive Point-of-Interest Recommendation with Local Differential Privacy." pith.science (2026). https://pith.science/paper/LTFLBB6I

@misc{pith2026190809485,
  author       = {Pith},
  title        = {Pith review of: Successive Point-of-Interest Recommendation with Local Differential Privacy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LTFLBB6I}},
  note         = {Machine review of arXiv:1908.09485}
}
read the original abstract

A point-of-interest (POI) recommendation system performs an important role in location-based services because it can help people to explore new locations and promote advertisers to launch advertisements at appropriate locations. The existing POI recommendation systems require raw check-in history of users, which might cause location privacy violations. Although there have been several matrix factorization (MF) based privacy-preserving recommendation systems, they can only focus on user-POI relationships without considering the human movements in check-in history. To tackle this problem, we design a successive POI recommendation framework with local differential privacy, named SPIREL. SPIREL uses two types of information derived from the check-in history as input for the factorization: a transition pattern between two POIs and the visit counts of POIs. We propose a novel objective function for learning the user-POI and POI-POI relationships simultaneously. We further integrate local differential privacy mechanisms in our proposed framework to prevent potential location privacy breaches. Experiments using four public datasets demonstrate that SPIREL achieves better POI recommendation quality while accomplishing stronger privacy protection.

Figures

Figures reproduced from arXiv: 1908.09485 by the authors.

Figure 1
Figure 1. FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3 [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: FIGURE 4 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: FIGURE 5 [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: FIGURE 6 [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 10
Figure 10. Figure 10: Moreover, we illustrate the root mean square error [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 7
Figure 7. Figure 7: FIGURE 7 [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: FIGURE 8 [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: FIGURE 9 [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: FIGURE 10 [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: FIGURE 11 [PITH_FULL_IMAGE:figures/full_fig_p014_11.png]
Figure 12
Figure 12. Figure 12: FIGURE 12 [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 42 canonical work pages

  1. [1]

    Where you like to go next: Successive point-of-interest recommendation,

    C. Cheng, H. Yang, M. R. Lyu, and I. King, “Where you like to go next: Successive point-of-interest recommendation,” in IJCAI, pp. 2605–2611, 2013

    work page 2013

  2. [2]

    Personalized point-of-interest recommendation by mining users’ preference transition,

    X. Liu, Y . Liu, K. Aberer, and C. Miao, “Personalized point-of-interest recommendation by mining users’ preference transition,” in ACM CIKM, pp. 733–738, 2013

    work page 2013

  3. [3]

    Geomf: joint geographical modeling and matrix factorization for point-of-interest recom- mendation,

    D. Lian, C. Zhao, X. Xie, G. Sun, E. Chen, and Y . Rui, “Geomf: joint geographical modeling and matrix factorization for point-of-interest recom- mendation,” in ACM SIGKDD, pp. 831–840, 2014

    work page 2014

  4. [4]

    Content-aware hierar- chical point-of-interest embedding model for successive POI recommenda- tion.,

    B. Chang, Y . Park, D. Park, S. Kim, and J. Kang, “Content-aware hierar- chical point-of-interest embedding model for successive POI recommenda- tion.,” in IJCAI, pp. 3301–3307, 2018

    work page 2018

  5. [5]

    Unique in the crowd: The privacy bounds of human mobility,

    Y .-A. De Montjoye, C. A. Hidalgo, M. Verleysen, and V . D. Blondel, “Unique in the crowd: The privacy bounds of human mobility,” Scientific reports, vol. 3, pp. 1–5, 2013

    work page 2013

  6. [6]

    Too much information: assessing privacy risks of contact trace data disclosure on people with covid-19 in south korea,

    G. Jung, H. Lee, A. Kim, and U. Lee, “Too much information: assessing privacy risks of contact trace data disclosure on people with covid-19 in south korea,” Frontiers in public health, vol. 8, pp. 1–13, 2020

    work page 2020

  7. [7]

    Rappor: Randomized aggregat- able privacy-preserving ordinal response,

    Ú. Erlingsson, V . Pihur, and A. Korolova, “Rappor: Randomized aggregat- able privacy-preserving ordinal response,” in ACM CCS, pp. 1054–1067, 2014

    work page 2014

  8. [8]

    Learning with privacy at scale,

    Differential Privacy Team, “Learning with privacy at scale,” Apple Machine Learning Journal, vol. 1, no. 8, pp. 1–25, 2017

    work page 2017

Show all 45 references
  1. [9]

    Collecting telemetry data privately,

    B. Ding, J. Kulkarni, and S. Yekhanin, “Collecting telemetry data privately,” in NIPS, pp. 3571–3580, 2017

    2017

  2. [10]

    Collecting and analyzing data from smart device users with local differential privacy,

    T. T. Nguyên, X. Xiao, Y . Yang, S. C. Hui, H. Shin, and J. Shin, “Collecting and analyzing data from smart device users with local differential privacy,” arXiv preprint arXiv:1606.05053, 2016

    2016 arXiv

  3. [11]

    Differentially private matrix factorization,

    J. Hua, C. Xia, and S. Zhong, “Differentially private matrix factorization,” in IJCAI, pp. 1763–1770, 2015

    2015

  4. [12]

    Probabilistic ma- trix factorization with personalized differential privacy,

    S. Zhang, L. Liu, Z. Chen, and H. Zhong, “Probabilistic ma- trix factorization with personalized differential privacy,” arXiv preprint arXiv:1810.08509, 2018

    2018 arXiv

  5. [13]

    Privacy enhanced matrix factor- ization for recommendation with local differential privacy,

    H. Shin, S. Kim, J. Shin, and X. Xiao, “Privacy enhanced matrix factor- ization for recommendation with local differential privacy,” IEEE TKDE, vol. 30, no. 9, pp. 1770–1782, 2018

    2018

  6. [14]

    Matrix factorization techniques for recommender systems,

    Y . Koren, R. Bell, and C. V olinsky, “Matrix factorization techniques for recommender systems,” Computer, no. 8, pp. 30–37, 2009

    2009

  7. [15]

    Faster and cheaper: Parallelizing large-scale matrix factorization on gpus,

    W. Tan, L. Cao, and L. Fong, “Faster and cheaper: Parallelizing large-scale matrix factorization on gpus,” in HPDC, pp. 219–230, 2016. VOLUME x, 20xx 15 J.S.Kim et al.: Successive Point-of-Interest Recommendation with Local Differential Privacy

    2016

  8. [16]

    Cross domain recommender sys- tems: A systematic literature review,

    M. M. Khan, R. Ibrahim, and I. Ghani, “Cross domain recommender sys- tems: A systematic literature review,” ACM Computing Surveys (CSUR), vol. 50, no. 3, pp. 1–34, 2017

    2017

  9. [17]

    Transfer learning in collaborative filtering for sparsity reduction,

    W. Pan, E. W. Xiang, N. N. Liu, and Q. Yang, “Transfer learning in collaborative filtering for sparsity reduction,” in AAAI, pp. 230–235, 2010

    2010

  10. [18]

    Transfer learning to predict missing ratings via heterogeneous user feedbacks,

    W. Pan, N. N. Liu, E. W. Xiang, and Q. Yang, “Transfer learning to predict missing ratings via heterogeneous user feedbacks,” in AAAI, pp. 2318– 2323, 2011

    2011

  11. [19]

    Transfer learning in heterogeneous collaborative filtering domains,

    W. Pan and Q. Yang, “Transfer learning in heterogeneous collaborative filtering domains,” Artificial intelligence, vol. 197, pp. 39–55, 2013

    2013

  12. [20]

    Local privacy and statistical minimax rates,

    J. C. Duchi, M. I. Jordan, and M. J. Wainwright, “Local privacy and statistical minimax rates,” in IEEE FOCS, pp. 429–438, 2013

    2013

  13. [21]

    Calibrating noise to sensitivity in private data analysis,

    C. Dwork, F. McSherry, K. Nissim, and A. Smith, “Calibrating noise to sensitivity in private data analysis,” in Theory of cryptography conference, pp. 265–284, 2006

    2006

  14. [22]

    Privacy integrated queries: an extensible platform for privacy-preserving data analysis,

    F. D. McSherry, “Privacy integrated queries: an extensible platform for privacy-preserving data analysis,” in ACM SIGMOD, pp. 19–30, 2009

    2009

  15. [23]

    Deep learning with differential privacy,

    M. Abadi, A. Chu, I. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang, “Deep learning with differential privacy,” in ACM CCS, pp. 308–318, 2016

    2016

  16. [24]

    Randomized response: A survey technique for eliminating evasive answer bias,

    S. L. Warner, “Randomized response: A survey technique for eliminating evasive answer bias,” Journal of the American Statistical Association, vol. 60, no. 309, pp. 63–69, 1965

    1965

  17. [25]

    Locally differentially private protocols for frequency estimation,

    T. Wang, J. Blocki, N. Li, and S. Jha, “Locally differentially private protocols for frequency estimation,” in USENIX Security Symposium, pp. 729–745, 2017

    2017

  18. [26]

    Hadamard response: Estimating distributions privately, efficiently, and with little communication,

    J. Acharya, Z. Sun, and H. Zhang, “Hadamard response: Estimating distributions privately, efficiently, and with little communication,” in PMLR, pp. 1120–1129, 2019

    2019

  19. [27]

    Learning new words from keystroke data with local differential privacy,

    S. Kim, H. Shin, C. H. Baek, S. Kim, and J. Shin, “Learning new words from keystroke data with local differential privacy,” IEEE TKDE, vol. 32, no. 7, pp. 479–491, 2018

    2018

  20. [28]

    Bpr: Bayesian personalized ranking from implicit feedback,

    S. Rendle, C. Freudenthaler, Z. Gantner, and L. Schmidt-Thieme, “Bpr: Bayesian personalized ranking from implicit feedback,” arXiv preprint arXiv:1205.2618, 2012

    2012 arXiv

  21. [29]

    Collaborative filtering for implicit feedback datasets,

    Y . Hu, Y . Koren, and C. V olinsky, “Collaborative filtering for implicit feedback datasets,” in IEEE ICDM, pp. 263–272, 2008

    2008

  22. [30]

    Collecting and analyzing multidimensional data with local differential privacy,

    N. Wang, X. Xiao, Y . Yang, J. Zhao, S. C. Hui, H. Shin, J. Shin, and G. Yu, “Collecting and analyzing multidimensional data with local differential privacy,” in IEEE ICDE, pp. 638–649, 2019

    2019

  23. [31]

    Privacy aware learning,

    J. C. Duchi, M. I. Jordan, and M. J. Wainwright, “Privacy aware learning,” Journal of the ACM (JACM), vol. 61, no. 6, pp. 1–57, 2014

    2014

  24. [32]

    Efficient mini-batch training for stochastic optimization,

    M. Li, T. Zhang, Y . Chen, and A. J. Smola, “Efficient mini-batch training for stochastic optimization,” in ACM SIGKDD, pp. 661–670, 2014

    2014

  25. [33]

    Adaptive subgradient methods for online learning and stochastic optimization,

    J. Duchi, E. Hazan, and Y . Singer, “Adaptive subgradient methods for online learning and stochastic optimization,” Journal of machine learning research, vol. 12, no. Jul, pp. 2121–2159, 2011

    2011

  26. [34]

    Neural networks for machine learning lecture 6a: overview of mini-batch gradient descent,

    G. Hinton, N. Srivastava, and K. Swersky, “Neural networks for machine learning lecture 6a: overview of mini-batch gradient descent,” University of Toronto, Technical Report, vol. 14, no. 8, pp. 8–31, 2012

    2012

  27. [35]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  28. [36]

    Adapting to user interest drift for poi recommendation,

    H. Yin, X. Zhou, B. Cui, H. Wang, K. Zheng, and Q. V . H. Nguyen, “Adapting to user interest drift for poi recommendation,” IEEE TKDE, vol. 28, no. 10, pp. 2566–2581, 2016

    2016

  29. [37]

    Locally differentially private item- based collaborative filtering,

    T. Guo, J. Luo, K. Dong, and M. Yang, “Locally differentially private item- based collaborative filtering,” Information Sciences, vol. 502, pp. 220–246, 2019

    2019

  30. [38]

    Privtrie: Effective frequent term discovery under local differential privacy,

    N. Wang, X. Xiao, Y . Yang, T. D. Hoang, H. Shin, J. Shin, and G. Yu, “Privtrie: Effective frequent term discovery under local differential privacy,” in IEEE ICDE, pp. 821–832, 2018

    2018

  31. [39]

    Factorizing person- alized markov chains for next-basket recommendation,

    S. Rendle, C. Freudenthaler, and L. Schmidt-Thieme, “Factorizing person- alized markov chains for next-basket recommendation,” in WWW, pp. 811– 820, 2010

    2010

  32. [40]

    Personalized ranking metric embedding for next new poi recommendation,

    S. Feng, X. Li, Y . Zeng, G. Cong, and Y . M. Chee, “Personalized ranking metric embedding for next new poi recommendation,” in IJCAI, pp. 2069– 2075, 2015

    2015

  33. [41]

    Lglmf: local geographical based logistic matrix factor- ization model for poi recommendation,

    H. A. Rahmani, M. Aliannejadi, S. Ahmadian, M. Baratchi, M. Afsharchi, and F. Crestani, “Lglmf: local geographical based logistic matrix factor- ization model for poi recommendation,” in Asia Information Retrieval Symposium, pp. 66–78, 2019

    2019

  34. [42]

    Building a rappor with the un- known: Privacy-preserving learning of associations and data dictionaries,

    G. Fanti, V . Pihur, and Ú. Erlingsson, “Building a rappor with the un- known: Privacy-preserving learning of associations and data dictionaries,” Proceedings on Privacy Enhancing Technologies, vol. 2016, no. 3, pp. 41– 61, 2016

    2016

  35. [43]

    Differentially private sequential data publication via variable-length n-grams,

    R. Chen, G. Acs, and C. Castelluccia, “Differentially private sequential data publication via variable-length n-grams,” in ACM CCS, pp. 638–649, 2012

    2012

  36. [44]

    “When and Where Do You Want to Hide?

    M. Asada, M. Yoshikawa, and Y . Cao, ““When and Where Do You Want to Hide?”–Recommendation of Location Privacy Preferences with Local Differential Privacy,” in IFIP Annual Conference on Data and Applications Security and Privacy, pp. 164–176, 2019

    2019

  37. [45]

    Practical privacy preserving poi recommendation,

    C. Chen, J. Zhou, B. Wu, W. Fang, L. Wang, Y . Qi, and X. Zheng, “Practical privacy preserving poi recommendation,” ACM Transactions on Intelligent Systems and Technology (TIST), vol. 11, no. 5, pp. 1–20, 2020. JONG SEON KIM received his B.S. degree in computer science from Ko...

    2020

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.