Pith. sign in

REVIEW 4 major objections 3 minor 1 cited by

Spectrum-based Modality Representation Fusion Graph Convolutional Network for Multimodal Recommendation

T0 review · 4 major / 3 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 SMORE fuses item text and image features in the frequency domain, where a trainable filter suppresses modality noise, and reports consistent gains over nine baselines on three Amazon datasets.

desk verdict Novel frequency-domain fusion for multimodal rec, but the main denoising claim is undercut by an equation inconsistency that must be fixed. read the letter →

arxiv 2412.14978 v1 pith:VWVFFKFB submitted 2024-12-19 cs.IR cs.MM

Rongqing Kenneth Ong , Andy W. H. Khong This is my paper · ORCID
classification cs.IRcs.MM
keywords multimodalrecommendationspectrum-basedfusionFouriertransformmodalitynoisesuppressiongraphconvolutionalnetworkcollaborativefilteringmodality-awarepreferencecontrastivelearning
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 proposes SMORE, a multimodal recommender that fuses item text and image features in the frequency domain rather than concatenating or summing them. It claims that a trainable complex filter applied to Fourier-transformed features suppresses modality-specific noise, preventing the cross-modality noise amplification that direct fusion causes. SMORE also learns item-item graphs from both uni-modal and fused features, and balances uni-modal and fusion preferences with behavioral gating. On three Amazon datasets (Baby, Sports, Clothing), the paper reports that SMORE outperforms both general and multimodal baselines on Recall@K and NDCG@K. If the claim holds, spectrum-based fusion is a viable denoising mechanism for multimodal recommendation.

What carries the argument

The load-bearing device is the adaptive spectral filter: after an MLP projects each modality into a shared space, the fast Fourier transform converts the features into complex spectra, and a modality-specific trainable complex filter $\delta_m$ attenuates noise via pointwise multiplication in Eq. (4). The filtered uni-modal spectra are then fused pointwise in Eq. (5), an operation equivalent to circular convolution in the spatial domain, capturing cross-modality correlations at logarithmic cost. The inverse FFT returns the denoised uni-modal and fused features to the original space for downstream graph learning and preference modeling.

What would settle it

Compare SMORE's Recall@20 against a variant in which the learned complex filter is replaced by an identity or an equally sized spatial-domain filter: if accuracy barely changes, the frequency-domain denoising mechanism is not the explanation for the gains. A more direct test is to inject known noise into modality features, pass them through the trained filter, and measure whether noise energy decays more than signal energy in the output spectra.

Watch

Extended reading notes

Core claim

SMORE's central claim is that fusing modalities after projecting them into the frequency domain, filtering each spectrum with a trainable complex weight, and combining the filtered spectra by pointwise product recovers cross-modal correlations while attenuating modality-specific noise. The pointwise product in the frequency domain is equivalent to circular convolution in the spatial domain, so the fusion captures interactions between text and image features. On top of this, SMORE constructs separate item-item graphs for each modality and a fusion graph by max-pooling edge weights across modalities, then propagates denoised features through light graph convolution. A modality-aware preference module gates the uni-modal and fused features with behavioral signals and applies an InfoNCE contrastive loss to align behavioral and modality views. The paper reports consistent improvements over nine baselines on three real-world datasets, with ablations showing each component contributes.

Load-bearing premise

The load-bearing premise is that the trainable complex filter genuinely separates useful semantic content from modality-specific noise in the frequency domain, so that filtered spectra fused pointwise yield cleaner features.

Editorial extensions

If this is right

  • Direct fusion methods (concatenation, element-wise sum, attention) risk amplifying modality-specific noise; SMORE presents a form of fusion designed to suppress it.
  • The same spectral fusion module can extend beyond two modalities, since the pointwise product and per-modality filters apply to any number of feature streams.
  • Because FFT and pointwise aggregation cost logarithmic time, the fusion step avoids the quadratic cost of co-attention-based fusion.
  • Explicitly modeling uni-modal and fusion preferences, then balancing them with behavioral gating, is presented as a more realistic model of user choice than single fused representations.

Reading between the lines

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

  • The paper's evidence for denoising is indirect (accuracy and t-SNE); a direct measurement of spectra before and after filtering would test whether the filter is actually removing noise energy.
  • The spectral filtering idea could transfer to other multimodal tasks beyond recommendation, such as retrieval or classification, wherever pretrained encoders inject per-modality noise.
  • Since the fusion graph is built by max-pooling modality edge weights, a learned or attention-weighted fusion graph might change the balance between uni-modal and fusion signals; the paper does not compare these variants.
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 / 3 minor

Summary. The manuscript proposes SMORE, a multimodal recommendation model that projects item modality features into the frequency domain, applies learned complex filters to suppress modality-specific noise, fuses the filtered spectra, and combines the result with graph-based collaborative and modality-aware preference signals. The model is evaluated on three Amazon subsets (Baby, Sports, Clothing) against general and multimodal baselines, reporting consistent improvements in Recall@K and NDCG@K. The central claim is that spectrum-based fusion simultaneously captures uni-modal and fusion preferences while suppressing cross-modality noise.

Significance. If the mechanism works as claimed, the paper makes a useful contribution: frequency-domain fusion is relatively underexplored in multimodal recommendation, the model is largely implementable, the code is publicly released, and the experiments cover three standard datasets with multiple baselines and five seeds. The ablations in Section 5.3 and the hyperparameter analysis in Section 5.4 are informative. However, the manuscript's own equations contain a load-bearing inconsistency in the fusion path, and the denoising claim is supported only by qualitative t-SNE visualizations and indirect downstream accuracy. The significance of the contribution depends on whether the implementation actually follows the intended filtered-fusion variant.

major comments (4)
  1. [§4.1, Eq. (5)] As written, the fusion spectrum is computed as bH_i,f = δ_f(∏_{m∈M} eH_i,m), i.e., the pointwise product is taken over the unfiltered spectra eH_i,m from Eq. (2), not over the filtered spectra bH_i,m from Eq. (4). This means the per-modality denoising filter δ_m is bypassed in the fusion path, and the sentence following Eq. (5) claiming that the pointwise product captures correlations 'while minimizing noise contamination during fusion' is not supported by the equations. If the intended operation is ∏ bH_i,m, the equation and the implementation must be corrected and verified; if Eq. (5) is deliberate, the central denoising-while-fusion claim needs to be reformulated, because the fusion input is exactly the noisy spectrum whose amplification the paper motivates in Fig. 1.
  2. [§5.5, Figs. 6 and 7] RQ4's evidence for denoising is qualitative: the manuscript shows t-SNE plots and kernel density estimates, but provides no quantitative measurement of noise suppression. In addition, the trainable filter W^c in Eq. (4) has no explicit denoising objective; it is trained end-to-end with BPR and contrastive losses, so improved downstream accuracy alone cannot distinguish noise suppression from added model capacity. Please report a quantitative comparison, e.g., an ablation that runs the same model with and without the per-modality filter in the fusion path, and, if feasible, a spectral or representation-level noise metric.
  3. [Table 2] The manuscript states that results are stable across five seeds and statistically significant with p < 0.01, but Table 2 reports only mean values with no standard deviations or confidence intervals. This is particularly important because some improvements are small, e.g., Recall@10 on Clothing is 0.0659 for SMORE versus 0.0649 for MGCN and 0.0635 for FREEDOM. Without per-seed variance, the consistency claim cannot be assessed by the reader; please include standard deviations or per-seed results, or justify why the paired t-test is sufficient.
  4. [§4.2.1, Eq. (15)] The user fusion feature h_u,f is defined in Eq. (15) by aggregating h_i,m rather than h_i,f. As written, the user-side fusion representation is not actually formed, even though the later modality-aware preference module in Eqs. (18)-(21) consumes H_f. This appears to be a typo, but it needs correction and the implementation should be checked to confirm that fusion features are propagated to users.
minor comments (3)
  1. [§4.2.1, Eq. (11)] The notation ¥S^m_a and ¥S^{m'}_b in Eq. (11) should presumably be ¥S^m_{a,b} and ¥S^{m'}_{a,b}; as written, the max appears to be taken over row or column vectors rather than scalar edge weights.
  2. [§4.1, Eq. (3)] The statement that the FFT reduces complexity to 'a logarithmic scale' is imprecise; the FFT complexity is O(n log n), not O(log n). Please rephrase.
  3. [§5.1.2] The abbreviation 'STOA' should be 'SOTA' for state-of-the-art.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: SMORE is an empirical model evaluated on held-out interactions, and its few self-citations are background references rather than load-bearing evidence.

full rationale

This paper presents an empirical recommender-system architecture rather than a formal derivation, and no reported result reduces to a fitted input by construction. The model parameters (complex filter weights W^c, MLP projections, gates, graph affinities) are trained with the BPR and contrastive objectives in Eqs. (24)-(25), and the effectiveness claims in Section 5.2 are measured on held-out 10% test splits of the Amazon Baby, Sports, and Clothing datasets against independently implemented baselines. No predicted metric is a renamed fitted constant, and no uniqueness theorem or prior result by the same authors is invoked to force the architecture. The two self-citations ([5] and [23]) appear only in a general list of references for bipartite-graph modeling and are not load-bearing. The reviewer-noted issue that Eq. (5) multiplies the unfiltered spectra eH_{i,m} rather than the filtered spectra bH_{i,m} is a potential typo or implementation ambiguity that could weaken the stated denoising mechanism, but it is not circular: Eq. (5) is not defined in terms of the paper's conclusion, and the reported accuracy gains are not derived from that equation by definition. The t-SNE evidence in Section 5.5 is qualitative, but it does not equate an input with an output. Overall, no circular step can be exhibited under the required standard, so the appropriate finding is a low score.

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

No new physical or conceptual entities are introduced; the spectral filter is a learned weight tensor, not a new entity. The central claims rest on standard signal-processing assumptions about FFT and on learned model parameters rather than on hand-fitted constants.

free parameters (4)
  • lambda1 (contrastive loss weight) = 0.01 (Baby), 0.03 (Sports)
    Tuned per dataset; performance varies strongly with it (Fig. 4).
  • K_m (top-K neighbors per modality) = K_t=10, K_v=40 for Baby; K_t=K_v=10 for Sports
    Graph sparsification parameter in Eq. (9); dataset-dependent optimal values found in Section 5.4.
  • lambda2 (L2 regularization weight) = not reported
    Appears in total loss Eq. (25) but value is not given.
  • tau (InfoNCE temperature) = not reported
    Temperature in Eq. (22), not disclosed.
assumptions (4)
  • standard math DFT/FFT and the convolution theorem apply to the feature matrices as used in Section 4.1.
    Justifies pointwise product of spectra as circular convolution.
  • domain assumption A learned pointwise complex filter can separate informative components from modality-specific noise in the frequency domain.
    Central mechanistic premise; supported only by downstream performance and t-SNE.
  • domain assumption Raw modality feature similarity (cosine similarity in Eq. 8) reflects semantic item-item relatedness.
    Used to build item graphs in Section 4.2.
  • domain assumption InfoNCE contrastive loss improves user and item representation alignment in this setting.
    Standard self-supervised assumption, incorporated in Eq. (22).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spectrum-based Modality Representation Fusion Graph Convolutional Network for Multimodal Recommendation." pith.science (2026). https://pith.science/paper/VWVFFKFB

@misc{pith2026241214978,
  author       = {Pith},
  title        = {Pith review of: Spectrum-based Modality Representation Fusion Graph Convolutional Network for Multimodal Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VWVFFKFB}},
  note         = {Machine review of arXiv:2412.14978}
}
read the original abstract

Incorporating multi-modal features as side information has recently become a trend in recommender systems. To elucidate user-item preferences, recent studies focus on fusing modalities via concatenation, element-wise sum, or attention mechanisms. Despite having notable success, existing approaches do not account for the modality-specific noise encapsulated within each modality. As a result, direct fusion of modalities will lead to the amplification of cross-modality noise. Moreover, the variation of noise that is unique within each modality results in noise alleviation and fusion being more challenging. In this work, we propose a new Spectrum-based Modality Representation (SMORE) fusion graph recommender that aims to capture both uni-modal and fusion preferences while simultaneously suppressing modality noise. Specifically, SMORE projects the multi-modal features into the frequency domain and leverages the spectral space for fusion. To reduce dynamic contamination that is unique to each modality, we introduce a filter to attenuate and suppress the modality noise adaptively while capturing the universal modality patterns effectively. Furthermore, we explore the item latent structures by designing a new multi-modal graph learning module to capture associative semantic correlations and universal fusion patterns among similar items. Finally, we formulate a new modality-aware preference module, which infuses behavioral features and balances the uni- and multi-modal features for precise preference modeling. This empowers SMORE with the ability to infer both user modality-specific and fusion preferences more accurately. Experiments on three real-world datasets show the efficacy of our proposed model. The source code for this work has been made publicly available at https://github.com/kennethorq/SMORE.

Figures

Figures reproduced from arXiv: 2412.14978 by the authors.

Figure 1
Figure 1. An illustrative example of user multi-modality pref [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. An illustrative overview of the proposed architecture, comprising three key components: (i) spectrum modality fusion, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Variation of SMORE with 𝜆1 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: Variation of SMORE with 𝐾𝑚 and mitigating cross-modality noise through attenuation and sup￾pression, while multi-modal graph learning and modality-aware preference modules aim to encode semantically associative uni￾versal signals and decipher accurate user preference, …
Figure 3
Figure 3. Figure 3: Ablation studies on the proposed SMORE directly injects modality features into ID representations, GRCN ex￾hibits moderate performance improvement by relying on modality features implicitly to enhance the interaction graph. From an alter￾nate perspective, the two MGCN …
Figure 6
Figure 6. Figure 6: Distribution of fusion features for Baby Dataset [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Survey on Multimodal Recommender Systems: Recent Advances and Future Directions

    cs.IR 2025-01 unverdicted novelty 3.0 of 10

    A survey that categorizes recent multimodal recommender system research into feature extraction, encoder, fusion, and loss-function technologies, with taxonomies and future directions.

Reference graph

Works this paper leans on

56 extracted references · 36 canonical work pages · cited by 1 Pith paper

  1. [1]

    Saeed Anwar and Nick Barnes. 2019. Real image denoising with feature attention. In Proceedings of the IEEE/CVF Int’l Conference on Computer Vision . 3155–3164

    work page 2019

  2. [2]

    Jie Chen, Haw-ren Fang, and Yousef Saad. 2009. Fast Approximate kNN Graph Construction for High Dimensional Data via Recursive Lanczos Bisection.Journal of Machine Learning Research 10, 9 (2009)

    work page 2009

  3. [3]

    Xu Chen, Hanxiong Chen, Hongteng Xu, Yongfeng Zhang, Yixin Cao, Zheng Qin, and Hongyuan Zha. 2019. Personalized fashion recommendation with visual explanations based on multimodal attention network: Towards visually explain- able recommendation. In Proceedings of the 42nd Int’l ACM SIGIR Conference on Spectrum-based Modality Representation Fusion Graph Co...

    work page 2019

  4. [4]

    Yang Chen, Yueqi Duan, Runzhong Zhang, and Yap-Peng Tan. 2024. Adaptive Margin Contrastive Learning for Ambiguity-aware 3D Semantic Segmentation. In 2024 IEEE Int’l Conference on Multimedia and Expo (ICME) . IEEE, 1–6

    work page 2024

  5. [5]

    Hong Wei Chun, Rongqing Kenneth Ong, and Andy W. H. Khong. 2024. Rea- sonable Sense of Direction: Making Course Recommendations Understandable with LLMs. In 2024 IEEE 67th International Midwest Symposium on Circuits and Systems (MWSCAS). IEEE, 1408–1412

    work page 2024

  6. [6]

    Enyan Dai, Wei Jin, Hui Liu, and Suhang Wang. 2022. Towards robust graph neural networks for noisy graphs with sparse labels. InProceedings of the Fifteenth ACM Int’l Conference on Web Search and Data Mining . 181–191

    work page 2022

  7. [7]

    Jun Gao, Di He, Xu Tan, Tao Qin, Liwei Wang, and Tie-Yan Liu. 2019. Repre- sentation degeneration problem in training natural language generation models. arXiv preprint arXiv:1907.12009 (2019)

    arXiv 2019

  8. [8]

    Xavier Glorot and Yoshua Bengio. 2010. Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the Thirteenth Int’l Con- ference on Artificial Intelligence and Statistics . JMLR Workshop and Conference Proceedings, 249–256

    work page 2010

Show all 56 references
  1. [9]

    Li He, Xianzhi Wang, Dingxian Wang, Haoyuan Zou, Hongzhi Yin, and Guandong Xu. 2023. Simplifying graph-based collaborative filtering for recommendation. In Proceedings of the sixteenth ACM Int’l Conference on Web Search and Data Mining . 60–68

    2023

  2. [10]

    Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. In Proceedings of the 25th Int’l Conference on World Wide Web . 507–517

    2016

  3. [11]

    Ruining He and Julian McAuley. 2016. VBPR: visual bayesian personalized ranking from implicit feedback. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 30

    2016

  4. [12]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. In Proceedings of the 43rd Int’l ACM SIGIR Conference on Research and Development in Information Retrieval . 639–648

    2020

  5. [13]

    Michael Heideman, Don Johnson, and Charles Burrus. 1984. Gauss and the history of the fast Fourier transform. IEEE ASSP Magazine 1, 4 (1984), 14–21

    1984

  6. [14]

    Saeed Izadi, Darren Sutton, and Ghassan Hamarneh. 2023. Image denoising in the deep learning era. Artificial Intelligence Review 56, 7 (2023), 5929–5974

    2023

  7. [15]

    Yungi Kim, Taeri Kim, Won-Yong Shin, and Sang-Wook Kim. 2024. MONET: Modality-Embracing Graph Convolutional Network and Target-Aware Attention for Multimedia Recommendation. In Proceedings of the 17th ACM Int’l Conference on Web Search and Data Mining . 332–340

    2024

  8. [16]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980 (2014)

    2014 arXiv

  9. [17]

    Thomas N Kipf and Max Welling. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016)

    2016 arXiv

  10. [18]

    An Lao, Qi Zhang, Chongyang Shi, Longbing Cao, Kun Yi, Liang Hu, and Duoqian Miao. 2024. Frequency spectrum is more effective for multimodal representation and fusion: A multimodal spectrum rumor detector. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. ...

    2024

  11. [19]

    Qiang Liu, Shu Wu, and Liang Wang. 2017. Deepstyle: Learning user preferences for visual recommendation. In Proceedings of the 40th Int’l ACM SIGIR Conference on Research and Development in Information Retrieval . 841–844

    2017

  12. [20]

    Yunshan Ma, Xiaohao Liu, Yinwei Wei, Zhulin Tao, Xiang Wang, and Tat-Seng Chua. 2024. Leveraging multimodal features and item-level user feedback for bundle construction. In Proceedings of the 17th ACM Int’l Conference on Web Search and Data Mining . 510–519

    2024

  13. [21]

    Kelong Mao, Jieming Zhu, Xi Xiao, Biao Lu, Zhaowei Wang, and Xiuqiang He

  14. [22]

    Michael Michelashvili and Lior Wolf. 2019. Speech denoising by accumulating per-frequency modeling fluctuations. arXiv preprint arXiv:1904.07612 (2019)

    2019 arXiv

  15. [23]

    Rongqing Kenneth Ong, Wei Qiu, and Andy W. H. Khong. 2023. Quad-Tier Entity Fusion Contrastive Representation Learning for Knowledge Aware Recommen- dation System. In Proceedings of the 32nd ACM Int’l Conference on Information and Knowledge Management. 1949–1959

    2023

  16. [24]

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. 2018. Representation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748 (2018)

    2018 arXiv

  17. [25]

    Ruihong Qiu, Zi Huang, Hongzhi Yin, and Zijian Wang. 2022. Contrastive learning for representation degeneration problem in sequential recommendation. In Proceedings of the fifteenth ACM Int’l Conference on Web Search and Data Mining. 813–823

    2022

  18. [26]

    Yongming Rao, Wenliang Zhao, Zheng Zhu, Jiwen Lu, and Jie Zhou. 2021. Global filter networks for image classification.Advances in Neural Information Processing Systems 34 (2021), 980–993

    2021

  19. [27]

    Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084 (2019)

    2019 arXiv

  20. [28]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme

  21. [29]

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. 2015. U-net: Convolu- tional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th Int’l conference, Munich, Germany, October 5-9, 2015, proceedings, part ...

    2015

  22. [30]

    Karen Simonyan and Andrew Zisserman. 2014. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)

    2014 arXiv

  23. [31]

    Zhulin Tao, Xiaohao Liu, Yewei Xia, Xiang Wang, Lifang Yang, Xianglin Huang, and Tat-Seng Chua. 2022. Self-supervised learning for multimedia recommenda- tion. IEEE Transactions on Multimedia (2022)

    2022

  24. [32]

    Laurens Van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-SNE. Journal of Machine Learning Research 9, 11 (2008)

    2008

  25. [33]

    Charles Van Loan. 1992. Computational frameworks for the fast Fourier transform . SIAM

    1992

  26. [34]

    Qifan Wang, Yinwei Wei, Jianhua Yin, Jianlong Wu, Xuemeng Song, and Liqiang Nie. 2021. Dualgnn: Dual graph neural network for multimedia recommendation. IEEE Transactions on Multimedia 25 (2021), 1074–1084

    2021

  27. [35]

    Wei Wei, Chao Huang, Lianghao Xia, and Chuxu Zhang. 2023. Multi-modal self-supervised learning for recommendation. In Proceedings of the ACM Web Conference 2023. 790–800

    2023

  28. [36]

    Yinwei Wei, Xiang Wang, Liqiang Nie, Xiangnan He, and Tat-Seng Chua. 2020. Graph-refined convolutional network for multimedia recommendation with implicit feedback. In Proceedings of the 28th ACM Int’l Conference on Multimedia . 3541–3549

    2020

  29. [37]

    Yinwei Wei, Xiang Wang, Liqiang Nie, Xiangnan He, Richang Hong, and Tat-Seng Chua. 2019. MMGCN: Multi-modal graph convolution network for personalized recommendation of micro-video. In Proceedings of the 27th ACM Int’l Conference on Multimedia. 1437–1445

    2019

  30. [38]

    Lanling Xu, Zhen Tian, Bingqian Li, Junjie Zhang, Daoyuan Wang, Hongyu Wang, Jinpeng Wang, Sheng Chen, and Wayne Xin Zhao. 2024. Sequence-level Semantic Representation Fusion for Recommender Systems. InProceedings of the 33rd ACM Int’l Conference on Information and Knowledge M...

    2024

  31. [39]

    Shuo Yang, Zhaopan Xu, Kai Wang, Yang You, Hongxun Yao, Tongliang Liu, and Min Xu. 2023. Bicro: Noisy correspondence rectification for multi-modality data via bi-directional cross-modal similarity consistency. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pa...

    2023

  32. [40]

    Zixuan Yi, Xi Wang, Iadh Ounis, and Craig Macdonald. 2022. Multi-modal graph contrastive learning for micro-video recommendation. In Proceedings of the 45th Int’l ACM SIGIR Conference on Research and Development in Information Retrieval . 1807–1811

    2022

  33. [41]

    Penghang Yu, Zhiyi Tan, Guanming Lu, and Bing-Kun Bao. 2023. Multi-view graph convolutional network for multimedia recommendation. In Proceedings of the 31st ACM Int’l Conference on Multimedia . 6576–6585

    2023

  34. [42]

    Jinghao Zhang, Yanqiao Zhu, Qiang Liu, Shu Wu, Shuhui Wang, and Liang Wang

  35. [43]

    Jinghao Zhang, Yanqiao Zhu, Qiang Liu, Mengqi Zhang, Shu Wu, and Liang Wang

  36. [44]

    Qingyang Zhang, Yake Wei, Zongbo Han, Huazhu Fu, Xi Peng, Cheng Deng, Qinghua Hu, Cai Xu, Jie Wen, Di Hu, et al. 2024. Multimodal fusion on low-quality data: A comprehensive survey. arXiv preprint arXiv:2404.18947 (2024)

    2024 arXiv

  37. [45]

    In Proceedings of the 29th ACM Int’l Conference on Multimedia

    Mining latent structures for multimedia recommendation. In Proceedings of the 29th ACM Int’l Conference on Multimedia . 3872–3880

  38. [46]

    Yilun Zheng, Sitao Luan, and Lihui Chen. 2024. What is missing in homophily? disentangling graph homophily for graph neural networks. arXiv preprint arXiv:2406.18854 (2024)

    2024 arXiv

  39. [47]

    Hongyu Zhou, Xin Zhou, Zhiwei Zeng, Lingzi Zhang, and Zhiqi Shen. 2023. A comprehensive survey on multimodal recommender systems: Taxonomy, evalua- tion, and future directions. arXiv preprint arXiv:2302.04473 (2023)

    2023 arXiv

  40. [48]

    Hongyu Zhou, Xin Zhou, Lingzi Zhang, and Zhiqi Shen. 2023. Enhancing dyadic relations with homogeneous graphs for multimodal recommendation. arXiv preprint arXiv:2301.12097 (2023)

    2023 arXiv

  41. [49]

    Kai Zhao, Yukun Zheng, Tao Zhuang, Xiang Li, and Xiaoyi Zeng. 2022. Joint learning of e-commerce search and recommendation with a unified graph neural network. In Proceedings of the Fifteenth ACM Int’l Conference on Web Search and Data Mining. 1461–1469

    2022

  42. [50]

    Xin Zhou and Zhiqi Shen. 2023. A tale of two graphs: Freezing and denoising graph structures for multimodal recommendation. In Proceedings of the 31st ACM Int’l Conference on Multimedia . 935–943

    2023

  43. [51]

    Xin Zhou, Aixin Sun, Yong Liu, Jie Zhang, and Chunyan Miao. 2023. Selfcf: A simple framework for self-supervised collaborative filtering. ACM Transactions on Recommender Systems 1, 2 (2023), 1–25

    2023

  44. [52]

    Xin Zhou, Hongyu Zhou, Yong Liu, Zhiwei Zeng, Chunyan Miao, Pengwei Wang, Yuan You, and Feijun Jiang. 2023. Bootstrap latent representations for multi-modal recommendation. In Proceedings of the ACM Web Conference 2023 . 845–854

    2023

  45. [53]

    Xin Zhou. 2023. Mmrec: Simplifying multimodal recommendation. InProceedings of the 5th ACM Int’l Conference on Multimedia in Asia Workshops . 1–2

    2023

  46. [2012]

    arXiv preprint arXiv:1205.2618 (2012)

    BPR: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618 (2012)

    2012 arXiv

  47. [2021]

    In Proceedings of the 30th ACM Int’l Conference on Information & Knowledge Management

    UltraGCN: ultra simplification of graph convolutional networks for rec- ommendation. In Proceedings of the 30th ACM Int’l Conference on Information & Knowledge Management. 1253–1262

  48. [2022]

    IEEE Transactions on Knowledge and Data Engineering (2022)

    Latent structure mining with contrastive modality fusion for multimedia recommendation. IEEE Transactions on Knowledge and Data Engineering (2022)

    2022

Pith tools

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