Pith. sign in

REVIEW 4 major objections 7 minor 61 references

Towards Transfer-Efficient Multi-modal Sequential Recommendation with State Space Duality

T0 review · 4 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read One architecture uses state-space decay and shared projections to make multi-modal sequential recommendation transfer about ten times faster.

desk verdict A promising but not-yet-reproducible multi-modal sequential recommender whose core fusion equation feeds a complex signal into real-valued operations without a projection rule; the empirical work is otherwise solid and the fix looks mechanical. read the letter →

arxiv 2506.02916 v5 pith:6GCIRHRP submitted 2025-06-03 cs.IR

Hao Fan , Qingyang Liu , Hongjiu Liu , Yanrong Hu , Kai Fang This is my paper · ORCID
classification cs.IR
keywords multi-modalsequentialrecommendationtransferlearningstatespacedualitymodelstime-awaremodelingFourierfilteringcross-modalalignmentcross-entropyobjective
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 show that transferable multi-modal sequential recommendation can be made dramatically cheaper by replacing complex training objectives with algebraic constraints inside the model. Its proposed framework, MMM4Rec, uses State Space Duality (SSD)—a structured sequence model whose state-decay mask acts like linear attention biased toward recent items—to align image and text features and then fuse them temporally. Both pre-training and fine-tuning use only a plain cross-entropy loss, with no contrastive or auxiliary losses. On five Amazon domains, the authors report state-of-the-art retrieval and roughly 10 times faster average convergence when transferring to large downstream datasets. The argument is that SR-compatible structural constraints, not optimization gymnastics, are what make multi-modal transfer efficient.

What carries the argument

The load-bearing object is the Time-aware Cross SSD (TiCoSSD) module—a reconfiguration of State Space Duality (SSD), a sequence model equivalent to linear attention with a state-decaying mask, in which the structured mask is built from inter-item time differences, the C matrix comes from the visual sequence, the B and X matrices come from the text sequence, and a pair of Fourier-domain filters (an adaptive per-modality filter and a shared learnable filter) merge the two modalities' time signals before the mask is applied. A companion Time-aware SSD (TiSSD) with shared weights performs sequence-level alignment before fusion. These modules together impose the algebraic constraints: temporal decay for recency bias, a refined time-aware mask to keep early critical items, and shared projections for cross-modal semantic consistency.

What would settle it

In the released code, inspect the tensor feeding the Cross-SSD equation: if the complex-valued fused signal from the inverse FFT is never converted to a real value before being used in the time-difference equations, the central fusion equation is not reproducible; additionally, re-running fine-tuning from the same pretrained checkpoint while replacing that fused signal with its real part would change the convergence curve if the Fourier fusion is genuinely load-bearing.

Watch

Extended reading notes

Core claim

MMM4Rec's central claim is that a two-stage alignment-then-fusion design, constrained by weight sharing and SSD's temporal-decay mask, is enough to make multi-modal sequential recommenders transfer efficiently. In the alignment stage, visual and textual sequences pass through the same time-aware SSD module with shared projection weights, pushing both modalities into a common recommendation semantic space. In the fusion stage, a Cross-SSD block derives the query-like matrix C from one modality and the key/value-like matrices B and X from the other, while a dual-channel Fourier filter combines the two modalities' time-difference signals into one fused signal that enters the mask. The same cross-entropy loss is used for pre-training and fine-tuning, and the authors report that this recipe beats prior transferable recommenders in retrieval and reaches strong downstream performance in a fraction of the epochs.

Load-bearing premise

The entire fusion stage depends on treating the complex-valued cross-modal time signal from the inverse Fourier transform as a real time-difference input to the state-space mask, but the paper states no rule for extracting that real signal.

Editorial extensions

If this is right

  • The model reaches state-of-the-art retrieval on the five downstream domains, with the largest reported gain being a 31.78% NDCG@10 improvement on the Pantry domain.
  • Fine-tuning converges about 10 times faster on average than the two main transferable baselines, requiring fewer epochs and less time per epoch on item-rich domains.
  • Transfer works with a single cross-entropy objective in both phases, so downstream adaptation needs no contrastive sampling or auxiliary losses.
  • The ID-removed variant stays competitive, meaning the learned multi-modal sequence representation itself carries transferable signal without item identifiers.
  • Ablation results tie every component—time awareness, shared-weight alignment, adaptive filter, learnable filter—to the reported performance, since removing any one degrades the metrics.

Reading between the lines

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

  • If the convergence-speed result generalizes beyond these five domains, the design suggests that architectural priors such as temporal decay and weight tying could substitute for elaborate pretraining objectives in other transferable recommender systems.
  • The Fourier-domain fusion of time-difference signals is a modular idea that could be lifted into other state-space or linear-attention recommenders that already use time-dependent masks.
  • A natural next test is whether the same alignment-then-fusion recipe extends to more than two modalities, since the paper only combines image and text.
  • One could probe the boundary of the claim by applying MMM4Rec to domains with very short interaction sequences, where SSD's recency bias provides less advantage and the Fourier fusion must carry more of the load.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The paper proposes MMM4Rec, a multi-modal sequential recommendation framework built on State Space Duality (SSD) / Mamba-style temporal modeling. The method consists of (i) frozen pretrained image/text encoders with lightweight adapters, (ii) sequence-level cross-modal alignment through weight-shared Time-aware SSD (TiSSD) modules, and (iii) a cross-modal fusion stage (TiCoSSD) that combines dual-channel Fourier filtering with a structurally adapted SSD. Both pre-training and fine-tuning use a simple cross-entropy objective with in-batch and full-corpus negative sampling, respectively. The central empirical claim is that MMM4Rec achieves state-of-the-art retrieval performance on five downstream Amazon domains and converges about 10x faster than the transferable baselines MMSRec and MISSRec, with ablations supporting the contributions of the time-aware, weight-sharing, and Fourier-filtering components. The paper also reports robustness in ID-removed and modality-missing settings.

Significance. If the mathematical definitions are corrected, the paper would offer a substantial empirical study: it compares against 14 baselines across 10 Amazon domains, includes both with-ID and ID-removed settings, provides ablation studies, full convergence curves, and an implementation link. The design goal of avoiding contrastive auxiliary losses and using a single cross-entropy objective is a noteworthy practical strength, and the convergence-speed results, if reproducible, would be valuable for transfer-efficient recommendation. However, the core TiCoSSD fusion module is not mathematically well-defined as written: a complex-valued signal from the inverse FFT is substituted into real-valued time-gap operations, and eq. (18) is dimensionally inconsistent. Because this module produces the user representation used in both losses, the headline results cannot currently be traced to a reproducible model. The issues are local and fixable, but they are load-bearing for the paper's central claim.

major comments (4)
  1. [§3.4.1–3.4.2, Eq. (16)] The output bD_f of the inverse FFT is defined in C^L, but in §3.4.2 it is substituted for the real time-difference vector D of §3.3.1, where it feeds Softplus in eq. (6), mask construction in eqs. (7)–(8), and the SSD output in eq. (9). LayerNorm and FFN in eq. (19) are then applied to the resulting M. No real-part, magnitude, or learned projection is specified, and none of these operations is defined for complex tensors as written. Please state the exact real-extraction rule and verify that all subsequent operations are real-valued.
  2. [§3.4.2, Eq. (18)] With Δ defined as an L-vector in eq. (17), the expression Δ̂^T X has shape 1×N, not L×N, so the equation is dimensionally inconsistent even if all quantities are real. Please clarify whether the intended operation is elementwise gating (for example Δ̂[:, None] * X) or a different row-wise scaling, and correct the formula accordingly.
  3. [§3.5.3, Eq. (22)] The fine-tuning loss denominator sums exp(<u_j, i_j> / τ) over j=1..|I|; full-corpus ranking for user k should instead sum exp(<u_k, i> / τ) over candidate items i ∈ I. As written, the denominator pairs item j with user j and is not a proper softmax over candidates for u_k. This typo changes the training objective and must be fixed.
  4. [§3.2.3] The optional item modality bias is implemented as two matrices E_t, E_v ∈ R^{|I|×N} added to item features; this is exactly an ID embedding table of size |I|×N per modality, so the statement that the module 'bypasses the dimensionality explosion of explicit ID features' is not supported. Please revise this claim or clarify what dimensionality is actually being bypassed.
minor comments (7)
  1. [§3.3.1, Eq. (5)] D is defined as [0, d_1, ..., d_{T-1}] with d_l = t_{l+1} - t_l for l ∈ [1, T]; a sequence of length L has only L-1 gaps, so the index range should be stated consistently as l ∈ [1, L-1] to avoid an off-by-one ambiguity.
  2. [§3.3.1, Eq. (4)] The notation Q_t = P_t * ω uses Q_t and P_t that are not defined; please define the causal convolution input and output variables explicitly.
  3. [§4.2 heading] The heading 'Comparasion' is a typo and should read 'Comparison'.
  4. [Table 1] In the Movies row, the number of items is printed as '59.203' with a period instead of a comma; please correct the formatting.
  5. [Fig. 2] The annotations 'Score =2.4%' and 'Score =3.6%' in the caption are not explained in the text; please clarify what these percentage values represent.
  6. [Tables 2 and 3] Since §3.2.3 describes the item modality bias as optional, please state explicitly whether the main results in Tables 2 and 3 use the bias or the ID-removed variant, and how the switch between the two is handled in the reported numbers.
  7. [References] Reference [7] is the authors' own TiM4Rec paper; the temporal backbone is therefore not independently introduced in this manuscript and should be attributed more prominently in §3.3.1.

Circularity Check

0 steps flagged · score 2.0 of 10

The paper's central transfer claims rest on external benchmark comparisons; only a minor non-load-bearing self-citation is present, and the undefined complex-valued fusion step in eqs. (16)-(18) is a correctness issue rather than circularity.

full rationale

MMM4Rec's headline results are empirical: pre-training on five Amazon domains and fine-tuning on five downstream domains, compared with 14 external baselines in Tables 2-3, with convergence speed measured as fine-tuning epochs and per-epoch time against MMSRec and MISSRec in Table 4. No fitted constant is renamed as a prediction, and no equation reduces to its own input by construction. The only same-author element is the temporal backbone: §3.3.1 states 'we adopt the Time-aware SSD proposed in TiM4Rec [7]' and cites the authors' own prior Neurocomputing paper. That is a component adopted from externally published work, not a claim derived here, and the multimodal alignment, fusion, and transfer results are evaluated independently, so the citation is not load-bearing in a circular sense. Separately, the fusion chain in §3.4 is mathematically underspecified: eq. (16) defines bD_f as a complex vector from an inverse FFT, and §3.4.2 then replaces D in section 3.3.1 with bD_f, while eqs. (5)-(6) and the subsequent Softplus, Δ, and mask constructions are real-valued; no real-extraction rule is stated, and eq. (18) also has an apparent dimension inconsistency with Δ^T X when Δ is an L-vector. This is a reproducibility or correctness defect, not a circular reduction, so it does not raise the circularity score beyond the minor same-author citation.

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

The central claim rests on a handful of hand-set hyperparameters (temperature, dropout, dimensions, layer count) and on two under-specified modeling choices: the complex-to-real time signal substitution in Cross-SSD and the handling of missing image modalities. The math background (SSD, ZOH, FFT) is standard and cited. No new physical entities are introduced.

free parameters (6)
  • temperature tau = 0.8
    Fixed globally in eqs. (21)-(22); sharpens or flattens the softmax and changes all ranking metrics; no sensitivity study.
  • dropout rate = 0.4
    Set 'to address Amazon sparsity' in §4.1.4; no tuning curve reported.
  • SSM state factor D = 64
    Mamba backbone state dimension; fixed in §4.1.4 without ablation.
  • causal convolution kernel size K = 4
    Used in eq. (4) and time-difference convolution; no sensitivity analysis.
  • feature modeling dimension N = 256
    Adapter output dimensionality for both modalities; set without ablation.
  • number of TiSSD/TiCoSSD layers = 1
    Ablation '2L' improves Office results, so the single-layer choice affects headline numbers.
assumptions (5)
  • standard math ZOH discretization and SSD semi-separable mask algebra (Dao et al. 2024, Gu and Dao 2023)
    Used in eqs. (6)-(9) without derivation; treated as background.
  • domain assumption Mamba's recency bias is a valid prior for sequential recommendation
    §2 supports it with a truncation experiment on SASRec, not on Mamba; the design assumes the bias transfers to multi-modal sequences.
  • ad hoc to paper Weight-shared TiSSD forces image and text into a common recommendation semantic space
    §3.3.2 states the constraint but provides no separate validation that shared weights alone produce semantic convergence.
  • ad hoc to paper Complex-valued inverse-FFT output can be used as a real time-difference signal
    Eq. (16) returns D_f in C^L; eq. (18) feeds it into eqs. (5)-(6) where Softplus and causal convolution expect real inputs; no real-part/magnitude rule is given.
  • ad hoc to paper Items with missing images can be processed without explicit masking or imputation
    §4.1.1 retains modality-missing items (Kindle has 0% image coverage) but §3.2.1 does not describe how the ViT encoder handles absent images.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Transfer-Efficient Multi-modal Sequential Recommendation with State Space Duality." pith.science (2026). https://pith.science/paper/6GCIRHRP

@misc{pith2026250602916,
  author       = {Pith},
  title        = {Pith review of: Towards Transfer-Efficient Multi-modal Sequential Recommendation with State Space Duality},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6GCIRHRP}},
  note         = {Machine review of arXiv:2506.02916}
}
abstract

Sequential Recommendation (SR) models infer user preferences from interaction histories. While transferable Multi-modal SR models outperform traditional ID-based approaches, existing methods struggle with slow fine-tuning convergence due to complex optimization requirements and negative transfer effects. We propose MMM4Rec (Multi-Modal Mamba for Sequential Recommendation), a novel Multi-modal SR framework that incorporates a dedicated algebraic constraint mechanism for efficient transfer learning. By combining State Space Duality (SSD)'s temporal decay properties with a globally-aware temporal modeling design, our model dynamically prioritizes key modality information, overcoming limitations of Transformer-based approaches. The framework implements a constrained two-stage process: (1) sequence-level cross-modal alignment via shared projection matrices, followed by (2) temporal fusion using our newly designed Cross-SSD module and dual-channel Fourier adaptive filtering. This architecture maintains semantic consistency while suppressing noise propagation. By incorporating algebraic structural constraints aligned with SR priors, MMM4Rec employs a simple and consistent cross-entropy objective across both pre-training and fine-tuning, enabling rapid fine-tuning convergence, substantially improving multimodal recommendation accuracy, and preserving strong transferability. Extensive experiments demonstrate MMM4Rec's state-of-the-art performance, achieving strong multi-modal retrieval capability and exhibiting 10$\times$ faster average convergence speed when transferring to large-scale downstream datasets. The implementation is available at link https://github.com/AlwaysFHao/MMM4Rec.

Figures

Figures reproduced from arXiv: 2506.02916 by the authors.

Figure 1
Figure 1. The two main research problems in multi-modal SR: (a) The alignment problem between multi-modal information [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Performance of SASRec at different truncation [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The overview of MMM4Rec . R 𝐿×𝐷 𝑡 𝑝 and image-modal feature sequence 𝑭 𝑣 = [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 7
Figure 7. Figure 7: Comparison of model convergence speed on Arts. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Comparison of model convergence speed on Office. [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 6
Figure 6. Figure 6: Comparison of model convergence speed on Instru [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 20 canonical work pages

  1. [1]

    Hangbo Bao, Wenhui Wang, Li Dong, Qiang Liu, Owais Khan Mohammed, Kriti Aggarwal, Subhojit Som, Songhao Piao, and Furu Wei. 2022. VLMo: Unified Vision-Language Pre-Training with Mixture-of-Modality-Experts. InAdvances in Neural Information Processing Systems 35: Annual Conference on Neural In- formation Processing Systems 2022, NeurIPS 2022, November 28 -...

    work page 2022

  2. [2]

    Ankur Bapna and Orhan Firat. 2019. Simple, Scalable Adaptation for Neural Machine Translation. InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, EMNLP-IJCNLP 2019, November 3-7, 2019, Kentaro Inui, Jing Jiang, Vincent Ng, and Xiaojun Wan (Eds....

    doi:10.18653/v1/d19-1165 2019

  3. [3]

    Tri Dao and Albert Gu. 2024. Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality. InForty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024. OpenReview.net, 1–31. https://openreview.net/forum?id=ztn8FCR1td

    work page 2024

  4. [4]

    Hao Ding, Yifei Ma, Anoop Deoras, Yuyang Wang, and Hao Wang. 2021. Zero- Shot Recommender Systems.CoRRabs/2105.08318 (2021). arXiv:2105.08318 https://arxiv.org/abs/2105.08318

    arXiv 2021

  5. [5]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. 2021. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In9th Interna- tional Conference on Learning Representations, IC...

    work page 2021

  6. [6]

    Timothy Dozat. 2016. Incorporating Nesterov Momentum into Adam. InICLR Workshop. 2013–2016. https://openreview.net/forum?id= OM0jvwB8jIp57ZJjtNEZ

    work page 2016

  7. [7]

    Hao Fan, Mengyi Zhu, Yanrong Hu, Hailin Feng, Zhijie He, Hongjiu Liu, and Qingyang Liu. 2025. TiM4Rec: An efficient sequential recommendation model based on time-aware structured state space duality model.Neurocomputing654 (2025), 131270. doi:10.1016/j.neucom.2025.131270

    work page arXiv 2025

  8. [8]

    Hui Fang, Danning Zhang, Yiheng Shu, and Guibing Guo. 2020. Deep Learning for Sequential Recommendation: Algorithms, Influential Factors, and Evaluations. ACM Trans. Inf. Syst.39, 1 (2020), 10:1–10:42. doi:10.1145/3426723

    doi:10.1145/3426723 2020

Show all 61 references
  1. [9]

    Albert Gu and Tri Dao. 2023. Mamba: Linear-Time Sequence Modeling with Selective State Spaces.CoRRabs/2312.00752 (2023). arXiv:2312.00752 doi:10. 48550/ARXIV.2312.00752

    2023 doi arXiv

  2. [10]

    Xu Guo, Tong Zhang, Yufei Xue, Chenxu Wang, Fuyun Wang, and Zhen Cui. 2025. M3Rec: Selective State Space Models with Mixture-of-Modality Experts for Multi- Modal Sequential Recommendation. InICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Proce...

    2025

  3. [11]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yong-Dong Zhang, and Meng Wang. 2020. LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. InProceedings of the 43rd International ACM SIGIR con- ference on research and development in Information Retrieva...

    2020

  4. [12]

    Balázs Hidasi, Alexandros Karatzoglou, Linas Baltrunas, and Domonkos Tikk

  5. [13]

    McAuley, and Wayne Xin Zhao

    Yupeng Hou, Zhankui He, Julian J. McAuley, and Wayne Xin Zhao. 2023. Learn- ing Vector-Quantized Item Representation for Transferable Sequential Recom- menders. InProceedings of the ACM Web Conference 2023, WWW 2023, 30 April 2023 - 4 May 2023, Ying Ding, Jie Tang, Juan F. Seq...

    2023

  6. [14]

    Yupeng Hou, Shanlei Mu, Wayne Xin Zhao, Yaliang Li, Bolin Ding, and Ji-Rong Wen. 2022. Towards Universal Sequence Representation Learning for Recom- mender Systems. InKDD ’22: The 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, August 14 - 18, 2022, Aidong Z...

    2022

  7. [15]

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin de Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-Efficient Transfer Learning for NLP. InProceedings of the 36th Inter- national Conference on Machine Learning, ICML 2...

    2019

  8. [16]

    Wang-Cheng Kang and Julian J. McAuley. 2018. Self-Attentive Sequential Rec- ommendation. InIEEE International Conference on Data Mining, ICDM 2018, November 17-20, 2018. IEEE Computer Society, Singapore, 197–206. doi:10.1109/ ICDM.2018.00035

    2018

  9. [17]

    Angelos Katharopoulos, Apoorv Vyas, Nikolaos Pappas, and François Fleuret

  10. [18]

    Khan, Muzammal Naseer, Munawar Hayat, Syed Waqas Zamir, Fa- had Shahbaz Khan, and Mubarak Shah

    Salman H. Khan, Muzammal Naseer, Munawar Hayat, Syed Waqas Zamir, Fa- had Shahbaz Khan, and Mubarak Shah. 2022. Transformers in Vision: A Survey. ACM Comput. Surv.54, 10s (2022), 200:1–200:41. doi:10.1145/3505244

    2022 doi

  11. [19]

    Chenglin Li, Mingjun Zhao, Huanming Zhang, Chenyun Yu, Lei Cheng, Guo- qiang Shu, Beibei Kong, and Di Niu. 2022. RecGURU: Adversarial Learning of Generalized User Representations for Cross-Domain Recommendation. In WSDM ’22: The Fifteenth ACM International Conference on Web Se...

    2022

  12. [20]

    Youhua Li, Hanwen Du, Yongxin Ni, Pengpeng Zhao, Qi Guo, Fajie Yuan, and Xiaofang Zhou. 2024. Multi-Modality is All You Need for Transferable Rec- ommender Systems. In40th IEEE International Conference on Data Engineer- ing, ICDE 2024, May 13-16, 2024. IEEE, Utrecht, The Nethe...

    2024

  13. [21]

    Jiahao Liang, Xiangyu Zhao, Muyang Li, Zijian Zhang, Wanyu Wang, Haochen Liu, and Zitao Liu. 2023. MMMLP: Multi-modal Multilayer Perceptron for Sequen- tial Recommendations. InProceedings of the ACM Web Conference 2023, WWW 2023, 30 April 2023 - 4 May 2023, Ying Ding, Jie Tang...

    2023

  14. [22]

    Chengkai Liu, Jianghao Lin, Jianling Wang, Hanzhou Liu, and James Caverlee

  15. [23]

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. 2023. Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing.ACM Comput. Surv.55, 9 (2023), 195:1–195:35. doi:10.1145/3560815

    2023 doi

  16. [24]

    Shiyu Liu, Jinzhao Su, Changdong Wang, Shengli Sun, Chen Lin, and Zhenhua Huang. 2026. Adaptive Temporal Expert Routing with Hierarchical Wavelet Enhancement for Multi-Modal Sequential Recommendation.ACM Trans. Inf. Syst.(March 2026). doi:10.1145/3801160 Just Accepted. Confere...

    2026 doi

  17. [25]

    Jie Lu, Dianshuang Wu, Mingsong Mao, Wei Wang, and Guangquan Zhang. 2015. Recommender system application developments: A survey.Decis. Support Syst. 74 (2015), 12–32. doi:10.1016/J.DSS.2015.03.008

    2015 doi

  18. [26]

    Jianmo Ni, Jiacheng Li, and Julian J. McAuley. 2019. Justifying Recommendations using Distantly-Labeled Reviews and Fine-Grained Aspects. InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natu...

    2019 doi

  19. [27]

    Smith, Albert Gu, Anushan Fernando, Çaglar Gülçehre, Razvan Pascanu, and Soham De

    Antonio Orvieto, Samuel L. Smith, Albert Gu, Anushan Fernando, Çaglar Gülçehre, Razvan Pascanu, and Soham De. 2023. Resurrecting Recurrent Neural Networks for Long Sequences. InInternational Conference on Machine Learning, ICML 2023, 23-29 July 2023 (Proceedings of Machine Lea...

    2023

  20. [28]

    Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Köpf, Edward Z. Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, ...

    2019

  21. [29]

    Ahmed Rashed, Shereen Elsayed, and Lars Schmidt-Thieme. 2022. Context and Attribute-Aware Sequential Recommendation via Cross-Attention. InProceedings of the 16th ACM Conference on Recommender Systems(Seattle, WA, USA)(RecSys ’22). Association for Computing Machinery, New York...

    2022

  22. [30]

    Steffen Rendle, Christoph Freudenthaler, and Lars Schmidt-Thieme. 2010. Factor- izing personalized Markov chains for next-basket recommendation. InProceedings of the 19th International Conference on World Wide Web, WWW 2010, Raleigh, North Carolina, USA, April 26-30, 2010. ACM...

    2010

  23. [31]

    Schein, Alexandrin Popescul, Lyle H

    Andrew I. Schein, Alexandrin Popescul, Lyle H. Ungar, and David M. Pennock

  24. [32]

    Yehjin Shin, Jeongwhan Choi, Hyowon Wi, and Noseong Park. 2024. An Atten- tive Inductive Bias for Sequential Recommendation beyond the Self-Attention. In Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applications of ...

    2024

  25. [33]

    Kunzhe Song, Qingfeng Sun, Can Xu, Kai Zheng, and Yaming Yang. 2023. Self- Supervised Multi-Modal Sequential Recommendation.CoRRabs/2304.13277 (2023). arXiv:2304.13277 doi:10.48550/ARXIV.2304.13277

    2023 doi arXiv

  26. [34]

    Weijie Su, Xizhou Zhu, Yue Cao, Bin Li, Lewei Lu, Furu Wei, and Jifeng Dai

  27. [35]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang

  28. [36]

    Hongyan Tang, Junning Liu, Ming Zhao, and Xudong Gong. 2020. Progressive Lay- ered Extraction (PLE): A Novel Multi-Task Learning (MTL) Model for Personalized Recommendations. InRecSys 2020: Fourteenth ACM Conference on Recommender Systems, September 22-26, 2020, Rodrygo L. T. ...

    2020

  29. [37]

    Jiaxi Tang and Ke Wang. 2018. Personalized Top-N Sequential Recommendation via Convolutional Sequence Embedding. InProceedings of the Eleventh ACM International Conference on Web Search and Data Mining, WSDM 2018, Marina Del Rey, CA, USA, February 5-9, 2018. ACM, 565–573. doi:...

    2018

  30. [38]

    Ilya O. Tolstikhin, Neil Houlsby, Alexander Kolesnikov, Lucas Beyer, Xiao- hua Zhai, Thomas Unterthiner, Jessica Yung, Andreas Steiner, Daniel Key- sers, Jakob Uszkoreit, Mario Lucic, and Alexey Dosovitskiy. 2021. MLP-Mixer: An all-MLP Architecture for Vision. InAdvances in Ne...

    2021

  31. [39]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. InAdvances in Neural Information Processing Systems 30: Annual Con- ference on Neural Information Processing Systems 2...

    2017

  32. [40]

    In 8th International Conference on Learning Representations, ICLR 2020, April 26-30,

    VL-BERT: Pre-training of Generic Visual-Linguistic Representations. In 8th International Conference on Learning Representations, ICLR 2020, April 26-30,

    2020

  33. [41]

    https://openreview.net/forum? id=SygXPaEYvH

    OpenReview.net, Addis Ababa, Ethiopia. https://openreview.net/forum? id=SygXPaEYvH

  34. [42]

    Sheng, and Mehmet A

    Shoujin Wang, Liang Hu, Yan Wang, Longbing Cao, Quan Z. Sheng, and Mehmet A. Orgun. 2019. Sequential Recommender Systems: Challenges, Progress and Prospects. InProceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI 2019, August 10-16,...

    2019 doi

  35. [43]

    Chuhan Wu, Fangzhao Wu, Tao Qi, Chao Zhang, Yongfeng Huang, and Tong Xu

  36. [44]

    Jingjing Xu, Xu Sun, Zhiyuan Zhang, Guangxiang Zhao, and Junyang Lin

  37. [45]

    McAuley, and Dong Wang

    Zhenrui Yue, Yueqi Wang, Zhankui He, Huimin Zeng, Julian J. McAuley, and Dong Wang. 2024. Linear Recurrent Units for Sequential Recommendation. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining, WSDM 2024, Merida, Mexico, March 4-8, 2024. AC...

    2024

  38. [46]

    Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. 2023. Sig- moid Loss for Language Image Pre-Training. InIEEE/CVF International Conference on Computer Vision, ICCV 2023, October 1-6, 2023. IEEE, Paris, France, 11941–11952. doi:10.1109/ICCV51070.2023.01100

    2023

  39. [47]

    Shengzhe Zhang, Liyi Chen, Dazhong Shen, Chao Wang, and Hui Xiong. 2025. Hierarchical Time-Aware Mixture of Experts for Multi-Modal Sequential Recom- mendation. InProceedings of the ACM on Web Conference 2025, WWW 2025, Sydney, NSW, Australia, 28 April 2025- 2 May 2025, Guodon...

    2025

  40. [48]

    Jose, Chenyun Yu, Beibei Kong, Zhijin Wang, Bo Hu, and Zang Li

    Jie Wang, Fajie Yuan, Mingyue Cheng, Joemon M. Jose, Chenyun Yu, Beibei Kong, Zhijin Wang, Bo Hu, and Zang Li. 2024. TransRec: Learning Transferable Recommendation from Mixture-of-Modality Feedback. InWeb and Big Data - 8th International Joint Conference, APWeb-W AIM 2024, Aug...

    2024

  41. [49]

    Jinpeng Wang, Ziyun Zeng, Yunxiao Wang, Yuting Wang, Xingyu Lu, Tianxiang Li, Jun Yuan, Rui Zhang, Hai-Tao Zheng, and Shu-Tao Xia. 2023. MISSRec: Pre- training and Transferring Multi-modal Interest-aware Sequence Representation for Recommendation. InProceedings of the 31st ACM...

    2023

  42. [50]

    Kun Zhou, Hui Yu, Wayne Xin Zhao, and Ji-Rong Wen. 2022. Filter-enhanced MLP is All You Need for Sequential Recommendation. InWWW ’22: The ACM Web Conference 2022, April 25 - 29, 2022, Frédérique Laforest, Raphaël Troncy, Elena Simperl, Deepak Agarwal, Aristides Gionis, Ivan H...

    2022

  43. [52]

    MM-Rec: Visiolinguistic Model Empowered Multimodal News Recommen- dation. InSIGIR ’22: The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, July 11 - 15, 2022, Enrique Amigó, Pablo Castells, Julio Gonzalo, Ben Carterette, J. Shane C...

    2022

  44. [54]

    Understanding and Improving Layer Normalization. InAdvances in Neural Information Processing Systems 32: Annual Conference on Neural Infor- mation Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancou- ver, BC, Canada. 4383–4393. https://proceedings.neurips.cc/pap...

    2019

  45. [58]

    Sheng, Jiajie Xu, De- qing Wang, Guanfeng Liu, and Xiaofang Zhou

    Tingting Zhang, Pengpeng Zhao, Yanchi Liu, Victor S. Sheng, Jiajie Xu, De- qing Wang, Guanfeng Liu, and Xiaofang Zhou. 2019. Feature-level Deeper Self-Attention Network for Sequential Recommendation. InProceedings of the Twenty-Eighth International Joint Conference on Artifici...

    2019 doi

  46. [59]

    Kun Zhou, Hui Wang, Wayne Xin Zhao, Yutao Zhu, Sirui Wang, Fuzheng Zhang, Zhongyuan Wang, and Ji-Rong Wen. 2020. S3-Rec: Self-Supervised Learning for Sequential Recommendation with Mutual Information Maximization. InCIKM ’20: The 29th ACM International Conference on Informatio...

    2020

  47. [61]

    pre-train and transfer

    successfully adapts the MLP-Mixer [ 38] architecture to SR. M3Rec [10] integrates MoE architecture, pioneering the application of Mamba to multi-modal SR. While these works partially address the shortcomings of pure ID-based modeling, they remain subopti- mal in achieving univ...

    2018

  48. [2002]

    Methods and metrics for cold-start recommendations. InSIGIR 2002: Pro- ceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, August 11-15, 2002, Kalervo Järvelin, Miche- line Beaulieu, Ricardo A. Baeza-Yates, and S...

    2002

  49. [2016]

    In4th International Conference on Learning Representations, ICLR 2016, San Juan, Puerto Rico, May 2-4, 2016, Conference Track Proceedings

    Session-based Recommendations with Recurrent Neural Networks. In4th International Conference on Learning Representations, ICLR 2016, San Juan, Puerto Rico, May 2-4, 2016, Conference Track Proceedings. 1–10. http://arxiv.org/abs/ 1511.06939

    2016 arXiv

  50. [2019]

    InProceedings of the 28th ACM International Con- ference on Information and Knowledge Management, CIKM 2019, Beijing, China, November 3-7, 2019

    BERT4Rec: Sequential Recommendation with Bidirectional Encoder Repre- sentations from Transformer. InProceedings of the 28th ACM International Con- ference on Information and Knowledge Management, CIKM 2019, Beijing, China, November 3-7, 2019. ACM, 1441–1450. doi:10.1145/33573...

    2019

  51. [2020]

    InProceedings of the 37th International Conference on Machine Learn- ing, ICML 2020, 13-18 July 2020, Virtual Event (Proceedings of Machine Learn- ing Research, Vol

    Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention. InProceedings of the 37th International Conference on Machine Learn- ing, ICML 2020, 13-18 July 2020, Virtual Event (Proceedings of Machine Learn- ing Research, Vol. 119). PMLR, 5156–5165. http://pr...

    2020

  52. [2022]

    http://papers.nips.cc/paper_files/paper/2022/hash/ d46662aa53e78a62afd980a29e0c37ed-Abstract-Conference.html

    New Orleans, LA, USA. http://papers.nips.cc/paper_files/paper/2022/hash/ d46662aa53e78a62afd980a29e0c37ed-Abstract-Conference.html

    2022

  53. [2024]

    arXiv:2403.03900 doi:10.48550/ ARXIV.2403.03900

    Mamba4Rec: Towards Efficient Sequential Recommendation with Selective State Space Models.CoRRabs/2403.03900 (2024). arXiv:2403.03900 doi:10.48550/ ARXIV.2403.03900

    2024 doi arXiv

Pith tools

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