Pith. sign in

REVIEW 3 major objections 6 minor 33 references

Future Sight and Tough Fights: Revolutionizing Sequential Recommendation with FENRec

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims FENRec improves sequential recommendation by about 6.16% on average across four benchmark datasets by replacing binary targets with time-dependent soft labels drawn from future interactions and by injecting enduring hard…

desk verdict FENRec is a plausible incremental trick for contrastive SR, but the paper never shows that its soft-label window is capped at the training split, so the headline 6.16% gain may be train/test leakage; referees should ask for the code before believing the empirical claim. read the letter →

arxiv 2412.11589 v4 pith:SPSZ4NEL submitted 2024-12-16 cs.IR

Yu-Hsuan Huang , Ling Lo , Hongxia Xie , Hong-Han Shuai , Wen-Huang Cheng This is my paper · ORCID
classification cs.IR
keywords sequentialrecommendationcontrastivelearningsoftlabelshardnegativeminingdatasparsityfutureinteractionutilizationtop-kranking
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 claims that data sparsity in sequential recommendation can be attacked more effectively by two cheap changes to contrastive training. Instead of rewarding only the exact next item, FENRec assigns decaying probabilities to the next two future interactions, turning upcoming behavior into soft labels. Instead of relying only on whatever in-batch negatives happen to appear, it manufactures enduring hard negatives by mixing each anchor with its negatives and reweighting the loss toward them. On Amazon Sports, Beauty, Toys, and Yelp, the combined method beats fourteen published baselines, with a claimed average improvement of 6.16% across Hit Ratio and NDCG at K=5, 10, and 20. A thoughtful reader would care because both components replace existing loss terms directly and can be dropped into contrastive sequential recommenders without new modules.

What carries the argument

The two load-bearing mechanisms are Time-Dependent Soft Labeling (Eqs. 6-7), which replaces the one-hot next-item target with a geometric distribution over the next up-to-two future items, and Enduring Hard Negatives Incorporation (Eqs. 9-10), which constructs synthetic negatives from normalized anchor and in-batch negative representations as $\tilde{h}_i = h_i/\|h_i\|_2$ and $h^-_{i,n} = \lambda\tilde{h}_i + (1-\lambda)\tilde{n}$, rescaled to the negative's norm. The paper proves a lemma that this construction keeps the synthetic negative closer in inner product to the anchor than the original negative, so it remains hard as training progresses. The final contrastive loss (Eq. 14) upweights these samples with a $\tanh$ similarity gate controlled by $\tau_2$ and a margin $m$, while `stop gradient` prevents the anchor from being pushed away from a vector that contains itself. Because these losses replace the standard $\mathcal{L}_{rec}$ and $\mathcal{L}_{cl}$, FENRec can be bolted onto any contrastive sequential recommendation framework.

What would settle it

Re-run the main table giving all fourteen baselines the same per-dataset hyperparameter search budget as FENRec, with FENRec's own hyperparameters fixed before seeing test metrics; if the average improvement falls to the noise level, the central state-of-the-art claim is falsified.

Watch

Extended reading notes

Core claim

FENRec establishes that future interactions, normally discarded at training time, can be converted into time-dependent soft labels: for a subsequence ending at position t, the items at positions t+1 and t+2 receive probabilities that decay geometrically with a smoothing hyperparameter gamma. Coupled with this, the method generates enduring hard negatives by normalizing anchor and negative representations, mixing them with a weight lambda, renormalizing, and rescaling to the negative's norm; a lemma shows the mixed vector has a strictly higher dot product with the anchor than the original negative does, so the negatives stay challenging late in training. The contrastive loss then upweights these hard negatives through a tanh-based similarity reweighting with a margin, and a stop-gradient prevents the anchor from repelling a vector that contains itself. The paper reports that this recipe beats all baselines on all four datasets across every metric, and that removing either component degrades performance, with the soft-label term having the larger effect.

Load-bearing premise

The claimed 6.16% average improvement presupposes a fair comparison: baseline hyperparameters were taken from their original papers or public implementations while FENRec's were tuned per dataset, so an equal tuning budget could shrink the gap.

Editorial extensions

If this is right

  • Dropping FENRec's two loss terms into an existing contrastive sequential recommender, demonstrated on ICSRec, CL4SRec, and DuoRec, improves next-item ranking by roughly 1.7% to 12.3% relative depending on dataset and metric.
  • The gains are larger at K=20 than at K=5, so the method improves ranking quality deeper into the recommendation list, not just at the very top.
  • Users with very few interactions also benefit: FENRec outperforms every baseline in every interaction-length group, including cold-start users, on all four datasets.
  • The ablation study implies the soft-label component contributes more than the hard-negative component, and the two are synergistic because soft labels mitigate the false-negative risk that hard negatives introduce.

Reading between the lines

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

  • Because the soft-label term adds no extra module and depends only on the sequence itself, the same future-labeling trick should transfer to non-contrastive sequence encoders; if that holds, FENRec's contribution is not limited to contrastive frameworks.
  • The geometric decay controlled by gamma is effectively a label-smoothing schedule over future positions, which predicts that the optimal gamma should shift with dataset sparsity and average sequence length; this is testable across additional datasets.
  • The enduring-hard-negative guarantee is only about cosine closeness of a single mixed vector, and in-batch mixing can amplify false negatives as batch size grows, so the reported optimal batch size of 256 may not transfer to datasets with strong item co-occurrence.
  • The paper's sensitivity analysis shows lambda matters little for final performance, which suggests most of the observed gain comes from the time-dependent soft-label head rather than from the specific form of the hard-negative mixing.
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 / 6 minor

Summary. The paper proposes FENRec, an add-on for contrastive sequential recommendation. It replaces binary next-item labels with time-dependent soft labels built from the next two future interactions in the user's chronological sequence, and it augments the contrastive loss with enduring hard negatives obtained by mixing the anchor with in-batch negatives, using stop-gradient and a focal-style upweighting term. The method is evaluated by inserting it into ICSRec, CL4SRec, and DuoRec on Amazon Sports/Beauty/Toys and Yelp, and the paper reports an average relative improvement of 6.16% over baselines, together with ablations and compatibility experiments.

Significance. The two proposed ingredients are simple and potentially portable, and the geometric claim that a mixed negative has higher inner product with the anchor than the original negative is proved in Appendix Lemma 2 and is correct. The compatibility study in Table 2 is a genuine strength: it shows the components can be wrapped around existing contrastive SR frameworks rather than requiring a new backbone. However, the central empirical claim rests on a training/evaluation protocol that is not fully specified. As written, the soft-label construction in Eq. (6) appears to use held-out validation and test items as training labels, which would invalidate the reported state-of-the-art results and the ablation conclusions. The baseline tuning asymmetry further weakens the quantitative comparison, and the broken code link prevents independent verification of the implementation.

major comments (3)
  1. [Method, Eq. (6); Appendix, Implementation Details] The soft-label construction is not described as being restricted to the training portion of each user sequence. Eq. (6) assigns probability mass to positions |S_us|+1 through min(|S_us|+3, |S_u|+1) of the full chronological sequence S_u, and the paper never states that these future positions are masked with respect to the train/validation/test split. Under the standard leave-last-two protocol used by the baselines, a subsequence ending at position T-2 receives soft-label mass on items T-1 and T, which are exactly the validation and test items on which FENRec is later ranked. The abstract's code link is corrupted ("zurlthttps://github.com/uikdwnd/FENRecu"), so the reader cannot check the released implementation. As written, the 6.16% average improvement and the Table 3 ablations do not demonstrate generalization; they may reflect training on the evaluation labels. Please specify and implement the masking of soft-label positions relative to the split and re-run all experiments, or provide explicit evidence that no held-out item ever appears in a soft-label window.
  2. [Appendix, Implementation Details] The claim that baselines are reported "under its optimal settings" is not supported by the described procedure. Baseline hyperparameters are taken from the original papers or public implementations, while FENRec's own hyperparameters (γ, λ, τ2, m, α, number of future items, and warm-up length) are tuned per dataset in the main text and in Appendix Figure G. Without a comparable per-dataset tuning protocol for the baselines, the reported relative gains (for example, 10-12% on Yelp) are not a fair state-of-the-art comparison, and the abstract's average improvement figure may be inflated. Please document the baseline tuning procedure or revise the optimality claim.
  3. [Table E and Comparison to SOTA (main text)] The main text states that FENRec "outperforms all other models across all metrics," but the significance annotations in Table E show that several best results are not marked with an asterisk (e.g., Beauty HIT@20 and NDCG@20; Toys HIT@5 and NDCG@5). The Appendix itself says improvements are significant "across most metrics." The abstract and main-text claim should be qualified to match the statistical evidence, unless the missing markers are an error.
minor comments (6)
  1. [Abstract] The code link is malformed: "zurlthttps://github.com/uikdwnd/FENRecu" should be "https://github.com/uikdwnd/FENRec".
  2. [Appendix, Lemma 1 proof] In case 1(b), the sentence "θxz = θxy + π = 0" is a typo; it should state that θxz = 0 and θxy = π.
  3. [Related Work] The first paragraph of Related Work refers to "CL4Rec," while the rest of the paper uses "CL4SRec"; please unify the naming.
  4. [Figure 5 caption] "Comparsion" should be "Comparison".
  5. [Appendix, Implementation Details] The phrase "noise-based negative sampling method on sentence representations" is vague; please specify how this method is adapted to sequential item representations and whether it is applied identically to all compared methods.
  6. [Table 3] The ablation table reports only a single run configuration per component removal; adding significance tests or repeated-run variability would help support the claim that the performance drops are not due to noise.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FENRec's components are defined independently of its evaluation targets, and the hardness property is a mathematical consequence of the mixing definition.

full rationale

The derivation chain is self-contained: Eq. (6)-(7) define time-dependent soft labels from logged future positions in the user sequence, Eq. (8) uses them in the cross-entropy loss, Eqs. (9)-(10) define enduring hard negatives by mixing anchor and negative representations, Eqs. (11)-(15) incorporate them into the contrastive loss, and Eq. (16) combines the objectives. None of these steps fits a parameter to the evaluation metric and then renames that fit as a prediction. Lemma 2's conclusion that the constructed hard negative has higher dot-product similarity to the anchor than the original negative follows directly from the mixing construction (a convex combination is directionally closer to the anchor); this is a mathematical property of the definition, not a circular claim about performance. Hyperparameters such as gamma, lambda, tau_2, mu, m, and the number of future items are tuned on validation and reported with significance tests against baselines, so the SOTA claim is an empirical result rather than a definitional one. The baseline numbers come from public implementations and original papers, and there is no load-bearing self-citation or imported uniqueness theorem. The only caveat is that the paper does not explicitly state whether the S_u used in Eq. (6) is restricted to the training split; under a common leave-last-two protocol, soft-label windows for some subsequences could include validation/test positions, which would be test-set leakage rather than circularity. Because the split is not specified in the text, that reduction cannot be exhibited from the paper itself, so it is not scored as circular under the hard rules.

Assumptions & free parameters 8 free parameters · 4 assumptions · 0 invented entities

No new entities. The enduring hard negatives are linear combinations of existing anchor and negative representations, not new model components or physical quantities. The method relies on standard contrastive learning assumptions, on the validity of using future interactions as training-only soft labels, and on the correctness of the geometric lemma.

free parameters (8)
  • γ (soft label smoothing exponent) = tuned in {0.1,0.2,0.3,0.4,0.5}; best 0.2 or 0.3
    Controls the concentration of time-dependent soft labels in Eq. (7); tuned per dataset on validation.
  • λ (hard negative mixing proportion) = tuned in {0.1,...,0.5}; described as insensitive
    Controls anchor proportion in the generated enduring hard negative in Eq. (10).
  • τ2 (tanh scaling) = tuned in {8, 10}
    Scales the similarity measure tanh in the focal contrastive loss, Eq. (13).
  • µ (enduring hard negative weight) = 0.1
    Weights the enduring hard negatives in the contrastive loss, Eq. (14).
  • m (focal offset) = 0.2
    Offset in the hard negative upweighting loss, Eq. (13), from focal InfoNCE.
  • α (contrastive loss weight) = not specified in main text
    Balances L'_{rec} and the contrastive losses in Eq. (16); assumed default or tuned in appendix.
  • number of future items in soft labels = 2 beyond the immediate next item
    Tuned per dataset (Appendix Fig. Gb); more distant items add noise.
  • warm-up epochs before hard negatives = 20
    Hard negatives are incorporated only after 20 epochs of training.
assumptions (4)
  • domain assumption Contrastive learning in SR improves representation quality and can be combined with the recommendation loss via multi-task learning.
    Borrowed from CL4SRec, DuoRec, ICLRec, and ICSRec (Preliminaries and Eq. (16)); not proven in this paper.
  • domain assumption The user's second and third future interactions are valid soft-label training targets for every prefix subsequence.
    Used in Eq. (6)-(7); if future behavior is noisy or non-stationary, the soft labels inject noise rather than signal.
  • domain assumption The noise-based negative sampling method from Zhou et al. 2022 applies unchanged to the SR representation space used here.
    Invoked in Implementation Details without specifying the procedure; the comparison fairness depends on applying it identically to all methods.
  • standard math Lemma 1 and Lemma 2 are correct: mixing a normalized anchor with a normalized negative yields a vector with higher cosine similarity to the anchor.
    Appendix, used to justify the enduring hard negatives in Eq. (10) and (14).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Future Sight and Tough Fights: Revolutionizing Sequential Recommendation with FENRec." pith.science (2026). https://pith.science/paper/SPSZ4NEL

@misc{pith2026241211589,
  author       = {Pith},
  title        = {Pith review of: Future Sight and Tough Fights: Revolutionizing Sequential Recommendation with FENRec},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SPSZ4NEL}},
  note         = {Machine review of arXiv:2412.11589}
}
read the original abstract

Sequential recommendation (SR) systems predict user preferences by analyzing time-ordered interaction sequences. A common challenge for SR is data sparsity, as users typically interact with only a limited number of items. While contrastive learning has been employed in previous approaches to address the challenges, these methods often adopt binary labels, missing finer patterns and overlooking detailed information in subsequent behaviors of users. Additionally, they rely on random sampling to select negatives in contrastive learning, which may not yield sufficiently hard negatives during later training stages. In this paper, we propose Future data utilization with Enduring Negatives for contrastive learning in sequential Recommendation (FENRec). Our approach aims to leverage future data with time-dependent soft labels and generate enduring hard negatives from existing data, thereby enhancing the effectiveness in tackling data sparsity. Experiment results demonstrate our state-of-the-art performance across four benchmark datasets, with an average improvement of 6.16\% across all metrics.

Figures

Figures reproduced from arXiv: 2412.11589 by the authors.

Figure 1
Figure 1. An illustration of binary labels compared to the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The distribution of similarity between samples and [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The framework of our method, FENRec, the user sequences will first be split into subsequences and encoded into [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of model performance across different [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Comparsion of ℓuniform and performance. Model Discriminative Capability Analysis. In this sec￾tion, we evaluate the model’s discriminative capability us￾ing item uniformity, which indicates the model’s ability to differentiate between semantically similar items. Lower …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 27 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    M.; Navimipour, N

    Alamdari, P. M.; Navimipour, N. J.; Hosseinzadeh, M.; Safaei, A. A.; and Darwesh, A. 2020. A systematic study on the recommender systems in the E-commerce. Ieee Access, 8: 115694--115716

    work page 2020

  4. [4]

    Chen, Y.; Liu, Z.; Li, J.; McAuley, J.; and Xiong, C. 2022. Intent contrastive learning for sequential recommendation. In Proceedings of the ACM Web Conference, 2172--2182

    work page 2022

  5. [5]

    Cheng, M.; Yuan, F.; Liu, Q.; Ge, S.; Li, Z.; Yu, R.; Lian, D.; Yuan, S.; and Chen, E. 2021. Learning recommender systems with implicit feedback via soft target enhancement. In Proceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, 575--584

    work page 2021

  6. [6]

    S.; Liu, Y.; Liu, G.; and Zhao, L

    Du, H.; Shi, H.; Zhao, P.; Wang, D.; Sheng, V. S.; Liu, Y.; Liu, G.; and Zhao, L. 2022. Contrastive learning with bidirectional transformers for sequential recommendation. In Proceedings of the ACM International Conference on Information and Knowledge Management, 396--405

    work page 2022

  7. [7]

    Hidasi, B.; Karatzoglou, A.; Baltrunas, L.; and Tikk, D. 2016. Session-based Recommendations with Recurrent Neural Networks. In Bengio, Y.; and LeCun, Y., eds., International Conference on Learning Representations

    work page 2016

  8. [8]

    Hou, P.; and Li, X. 2023. Improving Contrastive Learning of Sentence Embeddings with Focal I nfo NCE . In Findings of the Association for Computational Linguistics: EMNLP 2023

    work page 2023

Show all 33 references
  1. [9]

    B.; Pion, N.; Weinzaepfel, P.; and Larlus, D

    Kalantidis, Y.; Sariyildiz, M. B.; Pion, N.; Weinzaepfel, P.; and Larlus, D. 2020. Hard negative mixing for contrastive learning. Advances in neural information processing systems, 33: 21798--21809

    2020

  2. [10]

    Kang, W.-C.; and McAuley, J. 2018. Self-attentive sequential recommendation. In IEEE International Conference on Data Mining, 197--206. IEEE

    2018

  3. [11]

    Krichene, W.; and Rendle, S. 2020. On sampled metrics for item recommendation. In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 1748--1757

    2020

  4. [12]

    S.; McAuley, J.; and Xiong, C

    Liu, Z.; Chen, Y.; Li, J.; Yu, P. S.; McAuley, J.; and Xiong, C. 2021. Contrastive self-supervised sequential recommendation with robust augmentation. arXiv preprint arXiv:2108.06479

    2021 arXiv

  5. [13]

    Lu, J.; Wu, D.; Mao, M.; Wang, W.; and Zhang, G. 2015. Recommender system application developments: a survey. Decision support systems, 12--32

    2015

  6. [14]

    Qin, X.; Yuan, H.; Zhao, P.; Liu, G.; Zhuang, F.; and Sheng, V. S. 2024. Intent Contrastive Learning with Cross Subsequences for Sequential Recommendation. In Proceedings of the ACM International Conference on Web Search and Data Mining, 548--556

    2024

  7. [15]

    Qiu, R.; Huang, Z.; Yin, H.; and Wang, Z. 2022. Contrastive learning for representation degeneration problem in sequential recommendation. In Proceedings of the ACM international conference on web search and data mining, 813--823

    2022

  8. [16]

    Ren, X.; Xia, L.; Yang, Y.; Wei, W.; Wang, T.; Cai, X.; and Huang, C. 2024. Sslrec: A self-supervised learning framework for recommendation. In Proceedings of the ACM International Conference on Web Search and Data Mining, 567--575

    2024

  9. [17]

    Robinson, J.; Chuang, C.-Y.; Sra, S.; and Jegelka, S. 2021. Contrastive Learning with Hard Negative Samples. In International Conference on Learning Representations

    2021

  10. [18]

    Sharma, K.; Lee, Y.-C.; Nambi, S.; Salian, A.; Shah, S.; Kim, S.-W.; and Kumar, S. 2024. A survey of graph neural networks for social recommender systems. ACM Computing Surveys, 56(10): 1--34

    2024

  11. [19]

    Shin, Y.; Choi, J.; Wi, H.; and Park, N. 2024. An attentive inductive bias for sequential recommendation beyond the self-attention. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 8984--8992

    2024

  12. [20]

    Suh, Y.; Han, B.; Kim, W.; and Lee, K. M. 2019. Stochastic class-based hard example mining for deep metric learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7251--7259

    2019

  13. [21]

    Sun, F.; Liu, J.; Wu, J.; Pei, C.; Lin, X.; Ou, W.; and Jiang, P. 2019. BERT4Rec: Sequential recommendation with bidirectional encoder representations from transformer. In Proceedings of the ACM International Conference on Information and Knowledge Management, 1441--1450

    2019

  14. [22]

    Tang, J.; and Wang, K. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. In Proceedings of the ACM international conference on web search and data mining, 565--573

    2018

  15. [23]

    Wang, X.; He, X.; Wang, M.; Feng, F.; and Chua, T.-S. 2019. Neural graph collaborative filtering. In Proceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, 165--174

    2019

  16. [24]

    Wu, S.; Xin, X.; Ren, P.; Chen, Z.; Ma, J.; de Rijke, M.; and Ren, Z. 2023. Learning Robust Sequential Recommenders through Confident Soft Labels. arXiv preprint arXiv:2311.02446

    2023 arXiv

  17. [25]

    Xie, X.; Sun, F.; Liu, Z.; Wu, S.; Gao, J.; Zhang, J.; Ding, B.; and Cui, B. 2022. Contrastive learning for sequential recommendation. In IEEE International Conference on Data Engineering, 1259--1273. IEEE

    2022

  18. [26]

    Xuan, H.; Stylianou, A.; Liu, X.; and Pless, R. 2020. Hard negative examples are hard, but useful. In European Conference on Computer Vision, 126--142. Springer

    2020

  19. [27]

    Ye, Y.; Xia, L.; and Huang, C. 2023. Graph masked autoencoder for sequential recommendation. In Proceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, 321--330

    2023

  20. [28]

    Yu, L.; Zhang, C.; Liang, S.; and Zhang, X. 2019. Multi-order attentive ranking model for sequential recommendation. In Proceedings of the AAAI conference on artificial intelligence, 5709--5716

    2019

  21. [29]

    Yue, Z.; Wang, Y.; He, Z.; Zeng, H.; McAuley, J.; and Wang, D. 2024. Linear recurrent units for sequential recommendation. In Proceedings of the ACM International Conference on Web Search and Data Mining, 930--938

    2024

  22. [30]

    Zhan, J.; Mao, J.; Liu, Y.; Guo, J.; Zhang, M.; and Ma, S. 2021. Optimizing dense retrieval model training with hard negatives. In Proceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, 1503--1512

    2021

  23. [31]

    Zhang, Y.; Zhang, R.; Mensah, S.; Liu, X.; and Mao, Y. 2022. Unsupervised sentence representation via contrastive learning with mixing negatives. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, 11730--11738

    2022

  24. [32]

    Zhou, K.; Wang, H.; Wen, J.-r.; and Zhao, W. X. 2023. Enhancing Multi-View Smoothness for Sequential Recommendation Models. ACM Transactions on Information Systems, 41(4): 1--27

    2023

  25. [33]

    X.; and Wen, J.-R

    Zhou, K.; Zhang, B.; Zhao, W. X.; and Wen, J.-R. 2022. Debiased contrastive learning of unsupervised sentence representations. In Proceedings of Annual Meeting of the Association for Computational Linguistics

    2022

Pith tools

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