pith. machine review for the scientific record. sign in

arxiv: 2604.12965 · v1 · submitted 2026-04-14 · 💻 cs.IR

Recognition: unknown

Efficient Retrieval Scaling with Hierarchical Indexing for Large Scale Recommendation

Dongqi Fu , Kaushik Rangadurai , Haiyu Lu , Yunchen Pu , Siyang Yuan , Minhui Huang , Yiqun Liu , Golnaz Ghasemiesfeh
show 8 more authors
Xingfeng He Fangzhou Xu Andrew Cui Vidhoon Viswanathan Lin Yang Liang Wang Jiyan Yang Chonglin Sun
Authors on Pith no claims yet

Pith reviewed 2026-05-10 14:22 UTC · model grok-4.3

classification 💻 cs.IR
keywords hierarchical indexinglarge-scale retrievalrecommendation systemsresidual quantizationcross-attentiontest-time trainingadvertisement recommendationfoundational models
0
0 comments X

The pith

Jointly learning a hierarchical index with cross-attention and residual quantization enables efficient exact retrieval from large foundational recommendation models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests whether a hierarchical structure can be learned over the memory of large retrieval models used in recommendations to cut search costs without sacrificing exact matches. The method jointly trains the index using cross-attention and residual quantization so that the model organizes its own embeddings into levels that support fast traversal. This setup was deployed to handle daily ad recommendations for billions of Facebook and Instagram users. Intermediate nodes in the learned index turn out to mark high-quality data subsets, and fine-tuning the model on those subsets improves results further, extending the idea of test-time training to recommendation systems. A reader would care because it offers one path to run ever-larger foundational models in production without falling back on distillation or heavy caching.

Core claim

We propose jointly learning a hierarchical index using cross-attention and residual quantization for large-scale retrieval models. Such a hierarchical organization over the memory of foundational retrieval models would reduce retrieval costs while preserving exactness. Real-world deployment supports daily advertisement recommendations for billions of users. The intermediate nodes correspond to a small set of high-quality data whose fine-tuning further improves inference performance and concretizes test-time training in the recommendation domain.

What carries the argument

The jointly learned hierarchical index built with cross-attention and residual quantization, whose intermediate nodes identify high-quality data subsets.

If this is right

  • Retrieval latency drops while top-k exactness stays at 100 percent relative to the base model.
  • The same model can serve larger candidate pools without proportional compute growth.
  • Fine-tuning only on the high-quality subsets identified by intermediate nodes raises inference quality.
  • Production systems for billions of users can keep using the full foundational model instead of distilling it.
  • Both internal Meta data and public benchmarks show gains against strong baselines.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same joint-learning pattern could organize memories in other large embedding models outside recommendations.
  • High-quality subset discovery might become a general technique for identifying useful training examples at inference time.
  • Hybrid systems could combine this index with existing vector search libraries to handle both learned and hand-crafted hierarchies.
  • Test-time training on these subsets might reduce the need for full periodic retraining cycles.

Load-bearing premise

A hierarchical organization can be jointly learned over the memory of foundational retrieval models such that search costs are reduced while exactness is preserved without degrading the underlying representations.

What would settle it

Running the hierarchical index on a held-out query set and checking whether it returns exactly the same top-k items and scores as exhaustive search over the full model, or whether end-to-end latency exceeds the baseline in production traces.

Figures

Figures reproduced from arXiv: 2604.12965 by Andrew Cui, Chonglin Sun, Dongqi Fu, Fangzhou Xu, Golnaz Ghasemiesfeh, Haiyu Lu, Jiyan Yang, Kaushik Rangadurai, Liang Wang, Lin Yang, Minhui Huang, Siyang Yuan, Vidhoon Viswanathan, Xingfeng He, Yiqun Liu, Yunchen Pu.

Figure 1
Figure 1. Figure 1: Overall Pipeline of MoNN Foundation Retrieval Model with HILL Index. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A MoNN Block In general, Modular Neural Network (MoNN) enhances the learning of sophisticated user and item interactions beyond a single dot product while maintaining high efficiency. As shown in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A Hierarchical Index Example Learnt by HILL. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: 3-Layer MoNN Illustration Example. Loss Function. Each layer of multi-layer MoNN has its own loss function, i.e., with 1 <user, item> prediction loss function and <user, index node> prediction loss function. 5 Related Work To support retrieval models, prior works suggest that organizing candidate items into index structures can reduce the search space and expedite the identification of relevant item-user p… view at source ↗
read the original abstract

The increase in data volume, computational resources, and model parameters during training has led to the development of numerous large-scale industrial retrieval models for recommendation tasks. However, effectively and efficiently deploying these large-scale foundational retrieval models remains a critical challenge that has not been fully addressed. Common quick-win solutions for deploying these massive models include relying on offline computations (such as cached user dictionaries) or distilling large models into smaller ones. Yet, both approaches fall short of fully leveraging the representational and inference capabilities of foundational models. In this paper, we explore whether it is possible to learn a hierarchical organization over the memory of foundational retrieval models. Such a hierarchical structure would enable more efficient search by reducing retrieval costs while preserving exactness. To achieve this, we propose jointly learning a hierarchical index using cross-attention and residual quantization for large-scale retrieval models. We also present its real-world deployment at Meta, supporting daily advertisement recommendations for billions of Facebook and Instagram users. Interestingly, we discovered that the intermediate nodes in the learned index correspond to a small set of high-quality data. Fine-tuning the model on this set further improves inference performance, and concretize the concept of "test-time training" within the recommendation system domain. We demonstrate these findings using both internal and public datasets with strong baseline comparisons and hope they contribute to the community's efforts in developing the next generation of foundational retrieval models.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 0 minor

Summary. The paper claims that jointly learning a hierarchical index over foundational retrieval models via cross-attention and residual quantization enables efficient search with reduced costs while preserving exactness. It reports a production deployment at Meta supporting daily ad recommendations for billions of Facebook and Instagram users, observes that intermediate index nodes correspond to high-quality data, and shows that fine-tuning the model on this subset further improves inference performance (framed as test-time training). These findings are said to be demonstrated on internal and public datasets with strong baseline comparisons.

Significance. If the central claims hold—particularly that exact top-k retrieval accuracy is preserved while search cost drops—this would be a significant contribution to scaling large retrieval models in industrial recommendation systems. The reported Meta deployment provides real-world evidence of practicality at extreme scale, and the high-quality node observation plus test-time fine-tuning insight could open new directions for efficient adaptation of foundational models.

major comments (3)
  1. Abstract: the central claim that the hierarchical index 'preserves exactness' while reducing retrieval costs is load-bearing, yet the abstract (and the provided manuscript description) supplies no quantitative metrics, recall@K tables, ablation results, or error analysis comparing the indexed model to the flat baseline; without this, the exactness guarantee cannot be evaluated.
  2. Method description: no equations, algorithm box, or derivation is presented for the joint learning procedure that combines cross-attention with residual quantization, leaving unclear how the construction avoids the approximation errors (pruning, reconstruction, or candidate filtering) that typically arise in hierarchical or quantized indexes.
  3. Experiments section: the manuscript states that results are shown on internal and public datasets with 'strong baseline comparisons,' but the text contains no tables, figures, or specific numbers verifying that top-k accuracy remains identical to the non-hierarchical baseline or that any quantization error is bounded.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We appreciate the focus on ensuring the central claims are clearly supported by quantitative evidence and formal method descriptions. We address each major comment below and outline the revisions we will make.

read point-by-point responses

  1. Referee: [—] Abstract: the central claim that the hierarchical index 'preserves exactness' while reducing retrieval costs is load-bearing, yet the abstract (and the provided manuscript description) supplies no quantitative metrics, recall@K tables, ablation results, or error analysis comparing the indexed model to the flat baseline; without this, the exactness guarantee cannot be evaluated.

    Authors: We agree that the abstract would benefit from explicit quantitative support for the exactness claim. In the revised manuscript we will update the abstract to include key metrics (e.g., recall@K values on internal and public datasets) demonstrating that top-k accuracy matches the flat baseline while retrieval cost is reduced. We will also add a brief reference to the supporting tables and ablations that appear in the experiments section. revision: yes

  2. Referee: [—] Method description: no equations, algorithm box, or derivation is presented for the joint learning procedure that combines cross-attention with residual quantization, leaving unclear how the construction avoids the approximation errors (pruning, reconstruction, or candidate filtering) that typically arise in hierarchical or quantized indexes.

    Authors: We acknowledge that a more formal presentation is needed. We will insert the missing equations for the cross-attention-based node assignment and residual quantization objective, include an algorithm box that outlines the joint training procedure, and add a short derivation explaining why the resulting index permits exact top-k retrieval (i.e., no pruning or reconstruction error is introduced because every item remains reachable via its assigned path and the residual codes are used only for ordering within leaves). revision: yes

  3. Referee: [—] Experiments section: the manuscript states that results are shown on internal and public datasets with 'strong baseline comparisons,' but the text contains no tables, figures, or specific numbers verifying that top-k accuracy remains identical to the non-hierarchical baseline or that any quantization error is bounded.

    Authors: We agree that the experimental claims require explicit numerical backing. In the revision we will add tables reporting recall@K (and other ranking metrics) for the hierarchical index versus the flat baseline on both the Meta production dataset and the public benchmarks, together with figures that quantify the reduction in retrieval cost and ablations on the number of hierarchy levels and quantization bits. An error-analysis subsection will be included to show that any observed discrepancy is within the expected floating-point tolerance and that quantization error is bounded by construction. revision: yes

Circularity Check

0 steps flagged

No significant circularity; method is a novel construction with no self-referential derivations

full rationale

The paper proposes a new joint learning procedure for hierarchical indexes via cross-attention and residual quantization, supported by deployment at Meta and empirical comparisons on internal/public datasets. No equations, algorithms, or derivation steps are presented that reduce by construction to fitted inputs, self-citations, or renamed known results. The central claim is an engineering construction whose exactness preservation is asserted via the method itself rather than derived from prior self-referential premises. The absence of any load-bearing mathematical chain or uniqueness theorem imported from the same authors keeps the work self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only view provides no explicit free parameters, axioms, or invented entities; the approach assumes standard cross-attention and residual quantization can be jointly optimized for hierarchy without further specification of loss terms or constraints.

pith-pipeline@v0.9.0 · 5601 in / 1165 out tokens · 70278 ms · 2026-05-10T14:22:39.488810+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

75 extracted references · 50 canonical work pages · 4 internal anchors

  1. [1]

    Newsha Ardalani, Carole-Jean Wu, Zeliang Chen, Bhargav Bhushanam, and Adnan Aziz. 2022. Understanding Scaling Laws for Recommendation Models. CoRRabs/2208.08489 (2022). doi:10.48550/ARXIV.2208.08489 arXiv:2208.08489 Efficient Retrieval Scaling with Hierarchical Indexing for Large Scale Recommendation EDBT ’26, 24-27 March 2026, Tampere (Finland)

    work page doi:10.48550/arxiv.2208.08489 2022

  2. [2]

    Oren Barkan and Noam Koenigstein. 2016. ITEM2VEC: Neural item em- bedding for collaborative filtering. In26th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2016, Vietri sul Mare, Salerno, Italy, September 13-16, 2016, Francesco A. N. Palmieri, Aurelio Uncini, Kostas I. Diamantaras, and Jan Larsen (Eds.). IEEE, 1–6. doi:10.110...

    work page doi:10.1109/mlsp.2016.7738886 2016

  3. [3]

    Jane Bromley, Isabelle Guyon, Yann LeCun, Eduard Säckinger, and Roopak Shah. 1993. Signature Verification Using a Siamese Time Delay Neural Net- work. InAdvances in Neural Information Processing Systems 6, [7th NIPS Confer- ence, Denver, Colorado, USA, 1993], Jack D. Cowan, Gerald Tesauro, and Joshua Alspector (Eds.). Morgan Kaufmann, 737–744. http://pape...

    1993

  4. [4]

    Chong Chen, Min Zhang, Yongfeng Zhang, Yiqun Liu, and Shaoping Ma. 2020. Efficient Neural Matrix Factorization without Sampling for Recommendation. ACM Trans. Inf. Syst.38, 2 (2020), 14:1–14:28. doi:10.1145/3373807

    work page doi:10.1145/3373807 2020

  5. [5]

    Jeongwhan Choi, Seoyoung Hong, Noseong Park, and Sung-Bae Cho. 2023. Blurring-Sharpening Process Models for Collaborative Filtering. InProceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2023, Taipei, Taiwan, July 23-27, 2023, Hsin-Hsi Chen, Wei-Jou (Edward) Duh, Hen-Hsen Huang, Makoto P. ...

    work page doi:10.1145/3539618 2023

  6. [6]

    Jeongwhan Choi, Jinsung Jeon, and Noseong Park. 2021. LT-OCF: Learnable- Time ODE-based Collaborative Filtering. InCIKM ’21: The 30th ACM Interna- tional Conference on Information and Knowledge Management, Virtual Event, Queensland, Australia, November 1 - 5, 2021, Gianluca Demartini, Guido Zuc- con, J. Shane Culpepper, Zi Huang, and Hanghang Tong (Eds.)....

    work page doi:10.1145/3459637.3482449 2021

  7. [7]

    Paul Covington, Jay Adams, and Emre Sargin. 2016. Deep Neural Networks for YouTube Recommendations. InProceedings of the 10th ACM Conference on Recommender Systems, Boston, MA, USA, September 15-19, 2016, Shilad Sen, Werner Geyer, Jill Freyne, and Pablo Castells (Eds.). ACM, 191–198. doi:10. 1145/2959100.2959190

    work page arXiv 2016

  8. [8]

    Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szil- vasy, Pierre-Emmanuel Mazaré, Maria Lomeli, Lucas Hosseini, and Hervé Jégou. 2024. The Faiss library.CoRRabs/2401.08281 (2024). doi:10.48550/ ARXIV.2401.08281 arXiv:2401.08281

    work page internal anchor Pith review arXiv 2024

  9. [9]

    Travis Ebesu, Bin Shen, and Yi Fang. 2018. Collaborative Memory Network for Recommendation Systems. InThe 41st International ACM SIGIR Conference on Research & Development in Information Retrieval, SIGIR 2018, Ann Arbor, MI, USA, July 08-12, 2018, Kevyn Collins-Thompson, Qiaozhu Mei, Brian D. Davison, Yiqun Liu, and Emine Yilmaz (Eds.). ACM, 515–524. doi:...

    work page arXiv 2018

  10. [10]

    Wenqi Fan, Xiaorui Liu, Wei Jin, Xiangyu Zhao, Jiliang Tang, and Qing Li

  11. [11]

    Graph Trend Filtering Networks for Recommendation. InSIGIR ’22: The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, Madrid, Spain, July 11 - 15, 2022, Enrique Amigó, Pablo Castells, Julio Gonzalo, Ben Carterette, J. Shane Culpepper, and Gabriella Kazai (Eds.). ACM, 112–121. doi:10.1145/3477495.3531985

    work page doi:10.1145/3477495.3531985 2022

  12. [12]

    Yan Fang, Jingtao Zhan, Qingyao Ai, Jiaxin Mao, Weihang Su, Jia Chen, and Yiqun Liu. 2024. Scaling Laws For Dense Retrieval. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Informa- tion Retrieval, SIGIR 2024, Washington DC, USA, July 14-18, 2024, Grace Hui Yang, Hongning Wang, Sam Han, Claudia Hauff, Guido Zucc...

    work page doi:10.1145/3626772.3657743 2024

  13. [13]

    Chao Feng, Wuchao Li, Defu Lian, Zheng Liu, and Enhong Chen. 2022. Rec- ommender Forest for Efficient Retrieval. InAdvances in Neural Informa- tion Processing Systems 35: Annual Conference on Neural Information Pro- cessing Systems 2022, NeurIPS 2022, New Orleans, LA, USA, November 28 - December 9, 2022, Sanmi Koyejo, S. Mohamed, A. Agarwal, Danielle Belg...

    2022

  14. [14]

    Weihao Gao, Xiangjun Fan, Chong Wang, Jiankai Sun, Kai Jia, Wenzhi Xiao, Ruofan Ding, Xingyan Bin, Hui Yang, and Xiaobing Liu. 2020. Deep re- trieval: learning a retrievable structure for large-scale recommendations.arXiv preprint arXiv:2007.07203(2020)

    work page arXiv 2020

  15. [15]

    Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable Feature Learning for Networks. InProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, August 13- 17, 2016, Balaji Krishnapuram, Mohak Shah, Alexander J. Smola, Charu C. Aggarwal, Dou Shen, and Rajeev Rastogi (Eds.). ACM, 855–864...

    work page arXiv 2016

  16. [16]

    Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. 2017. DeepFM: A Factorization-Machine based Neural Network for CTR Prediction. InProceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI 2017, Melbourne, Australia, August 19-25, 2017, Carles Sierra (Ed.). ijcai.org, 1725–1731. doi:10.24963/IJCAI.2017/239

    work page doi:10.24963/ijcai.2017/239 2017

  17. [18]

    Moritz Hardt and Yu Sun. 2024. Test-Time Training on Nearest Neighbors for Large Language Models. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net. https://openreview.net/forum?id=CNL2bku4ra

    2024

  18. [19]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yong-Dong Zhang, and Meng Wang. 2020. LightGCN: Simplifying and Powering Graph Convolution Net- work for Recommendation. InProceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, SIGIR 2020, Virtual Event, China, July 25-30, 2020, Jimmy X. Huang, Yi Chang,...

    work page doi:10.1145/3397271.3401063 2020

  19. [20]

    Xinran He, Junfeng Pan, Ou Jin, Tianbing Xu, Bo Liu, Tao Xu, Yanxin Shi, Antoine Atallah, Ralf Herbrich, Stuart Bowers, and Joaquin Quiñonero Can- dela. 2014. Practical Lessons from Predicting Clicks on Ads at Facebook. InProceedings of the Eighth International Workshop on Data Mining for On- line Advertising, ADKDD 2014, August 24, 2014, New York City, N...

    work page doi:10.1145/2648584.2648589 2014

  20. [21]

    Training Compute-Optimal Large Language Models

    Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch, Elena Buchatskaya, Trevor Cai, Eliza Rutherford, Diego de Las Casas, Lisa Anne Hendricks, Jo- hannes Welbl, Aidan Clark, Tom Hennigan, Eric Noland, Katie Millican, George van den Driessche, Bogdan Damoc, Aurelia Guy, Simon Osindero, Karen Si- monyan, Erich Elsen, Jack W. Rae, Oriol Vinyals, and Laurent S...

    work page internal anchor Pith review doi:10.48550/arxiv.2203.15556 2022

  21. [22]

    Belongie, and Deborah Estrin

    Cheng-Kang Hsieh, Longqi Yang, Yin Cui, Tsung-Yi Lin, Serge J. Belongie, and Deborah Estrin. 2017. Collaborative Metric Learning. InProceedings of the 26th International Conference on World Wide Web, WWW 2017, Perth, Australia, April 3-7, 2017, Rick Barrett, Rick Cummings, Eugene Agichtein, and Evgeniy Gabrilovich (Eds.). ACM, 193–201. doi:10.1145/3038912.3052639

    work page doi:10.1145/3038912.3052639 2017

  22. [23]

    Jun Hu, Bryan Hooi, Shengsheng Qian, Quan Fang, and Changsheng Xu

  23. [24]

    MGDCF: Distance Learning via Markov Graph Diffusion for Neural Collaborative Filtering.IEEE Trans. Knowl. Data Eng.36, 7 (2024), 3281–3296. doi:10.1109/TKDE.2023.3348537

    work page doi:10.1109/tkde.2023.3348537 2024

  24. [25]

    Jui-Ting Huang, Ashish Sharma, Shuying Sun, Li Xia, David Zhang, Philip Pronin, Janani Padmanabhan, Giuseppe Ottaviano, and Linjun Yang. 2020. Embedding-based Retrieval in Facebook Search. InKDD ’20: The 26th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Virtual Event, CA, USA, August 23-27, 2020, Rajesh Gupta, Yan Liu, Jiliang Tang, and B...

    work page doi:10.1145/3394486.3403305 2020

  25. [26]

    Jeff Johnson, Matthijs Douze, and Hervé Jégou. 2021. Billion-Scale Similarity Search with GPUs.IEEE Trans. Big Data7, 3 (2021), 535–547. doi:10.1109/ TBDATA.2019.2921572

    work page arXiv 2021

  26. [27]

    Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B. Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei

  27. [28]

    Scaling Laws for Neural Language Models

    Scaling Laws for Neural Language Models.CoRRabs/2001.08361 (2020). arXiv:2001.08361 https://arxiv.org/abs/2001.08361

    work page internal anchor Pith review Pith/arXiv arXiv 2001

  28. [29]

    Johannes Klicpera, Aleksandar Bojchevski, and Stephan Günnemann. 2019. Predict then Propagate: Graph Neural Networks meet Personalized PageRank. In7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net. https://openreview.net/ forum?id=H1gL-2A9Ym

    2019

  29. [30]

    Taeyong Kong, Taeri Kim, Jinsung Jeon, Jeongwhan Choi, Yeon-Chang Lee, Noseong Park, and Sang-Wook Kim. 2022. Linear, or Non-Linear, That is the Question!. InWSDM ’22: The Fifteenth ACM International Conference on Web Search and Data Mining, Virtual Event / Tempe, AZ, USA, February 21 - 25, 2022, K. Selcuk Candan, Huan Liu, Leman Akoglu, Xin Luna Dong, an...

    work page doi:10.1145/3488560.3498501 2022

  30. [31]

    Dongha Lee, SeongKu Kang, Hyunjun Ju, Chanyoung Park, and Hwanjo Yu

  31. [32]

    Bootstrapping User and Item Representations for One-Class Collabora- tive Filtering. InSIGIR ’21: The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, Virtual Event, Canada, July 11-15, 2021, Fernando Diaz, Chirag Shah, Torsten Suel, Pablo Castells, Rosie Jones, and Tetsuya Sakai (Eds.). ACM, 1513–1522. doi:10....

    work page doi:10.1145/3404835.3462935 2021

  32. [33]

    Doyup Lee, Chiheon Kim, Saehoon Kim, Minsu Cho, and Wook-Shin Han. 2022. Autoregressive Image Generation using Residual Quantization. arXiv:2203.01941 [cs.CV]

    work page arXiv 2022

  33. [34]

    Haitao Li, Qingyao Ai, Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Zheng Liu, and Zhao Cao. 2023. Constructing Tree-based Index for Efficient and Effective Dense Retrieval. InProceedings of the 46th International ACM SIGIR Confer- ence on Research and Development in Information Retrieval, SIGIR 2023, Taipei, Taiwan, July 23-27, 2023, Hsin-Hsi Chen, Wei-Jou (Edwa...

    work page doi:10.1145/3539618.3591651 2023

  34. [35]

    Wuchao Li, Kai Zheng, Defu Lian, Qi Liu, Wentian Bao, Yunen Yu, Yang Song, Han Li, and Kun Gai. 2025. Making Transformer Decoders Better Differentiable Indexers. InThe Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025. OpenReview.net. https: //openreview.net/forum?id=bePaRx0otZ

    2025

  35. [36]

    Krishnan, Matthew D

    Dawen Liang, Rahul G. Krishnan, Matthew D. Hoffman, and Tony Jebara. 2018. Variational Autoencoders for Collaborative Filtering. InProceedings of the 2018 World Wide Web Conference on World Wide Web, WWW 2018, Lyon, France, April 23-27, 2018, Pierre-Antoine Champin, Fabien Gandon, Mounia Lalmas, and Panagiotis G. Ipeirotis (Eds.). ACM, 689–698. doi:10.114...

    work page doi:10.1145/3178876.3186150 2018

  36. [37]

    Mingfu Liang, Xi Liu, Rong Jin, Boyang Liu, Qiuling Suo, Qinghai Zhou, Song Zhou, Laming Chen, Hua Zheng, Zhiyuan Li, Shali Jiang, Jiyan Yang, Xiaozhen EDBT ’26, 24-27 March 2026, Tampere (Finland) Dongqi Fu et al. Xia, Fan Yang, Yasmine Badr, Ellie Wen, Shuyu Xu, Hansey Chen, Zhengyu Zhang, Jade Nie, Chunzhi Yang, Zhichen Zeng, Weilin Zhang, Xingliang Hu...

    work page doi:10.1145/3701716.3715223 2026

  37. [38]

    Fan Liu, Zhiyong Cheng, Lei Zhu, Zan Gao, and Liqiang Nie. 2021. Interest- aware Message-Passing GCN for Recommendation. InWWW ’21: The Web Conference 2021, Virtual Event / Ljubljana, Slovenia, April 19-23, 2021, Jure Leskovec, Marko Grobelnik, Marc Najork, Jie Tang, and Leila Zia (Eds.). ACM / IW3C2, 1296–1305. doi:10.1145/3442381.3449986

    work page doi:10.1145/3442381.3449986 2021

  38. [39]

    Zhiwei Liu, Lin Meng, Fei Jiang, Jiawei Zhang, and Philip S. Yu. 2022. De- oscillated Adaptive Graph Collaborative Filtering. InTopological, Algebraic and Geometric Learning Workshops 2022, 25-22 July 2022, Virtual (Proceed- ings of Machine Learning Research, Vol. 196), Alexander Cloninger, Timothy Doster, Tegan Emerson, Manohar Kaul, Ira Ktena, Henry Kvi...

    2022

  39. [40]

    Ze Liu, Jin Zhang, Chao Feng, Defu Lian, Jie Wang, and Enhong Chen. 2024. Learning Deep Tree-based Retriever for Efficient Recommendation: Theory and Method.arXiv preprint arXiv:2408.11345(2024)

    work page arXiv 2024

  40. [41]

    Jianxin Ma, Peng Cui, Kun Kuang, Xin Wang, and Wenwu Zhu. 2019. Dis- entangled Graph Convolutional Networks. InProceedings of the 36th Inter- national Conference on Machine Learning, ICML 2019, 9-15 June 2019, Long Beach, California, USA (Proceedings of Machine Learning Research, Vol. 97), Kamalika Chaudhuri and Ruslan Salakhutdinov (Eds.). PMLR, 4212–422...

    2019

  41. [42]

    Jianxin Ma, Chang Zhou, Peng Cui, Hongxia Yang, and Wenwu Zhu

  42. [43]

    Learning Disentangled Representations for Recommendation. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancouver, BC, Canada, Hanna M. Wallach, Hugo Larochelle, Alina Beygelzimer, Florence d’Alché-Buc, Emily B. Fox, and Roman Gar- nett (Eds.). 5...

    2019

  43. [44]

    Kelong Mao, Jieming Zhu, Jinpeng Wang, Quanyu Dai, Zhenhua Dong, Xi Xiao, and Xiuqiang He. 2021. SimpleX: A Simple and Strong Baseline for Collaborative Filtering. InCIKM ’21: The 30th ACM International Conference on Information and Knowledge Management, Virtual Event, Queensland, Australia, November 1 - 5, 2021, Gianluca Demartini, Guido Zuccon, J. Shane...

    work page doi:10.1145/3459637 2021

  44. [45]

    Biswajit Paria, Chih-Kuan Yeh, Ian E. H. Yen, Ning Xu, Pradeep Ravikumar, and Barnabás Póczos. 2020. Minimizing FLOPs to Learn Efficient Sparse Representations. arXiv:2004.05665 [cs.LG]

    work page arXiv 2020

  45. [46]

    Bryan Perozzi, Rami Al-Rfou, and Steven Skiena. 2014. DeepWalk: online learning of social representations. InThe 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’14, New York, NY, USA - August 24 - 27, 2014, Sofus A. Macskassy, Claudia Perlich, Jure Leskovec, Wei Wang, and Rayid Ghani (Eds.). ACM, 701–710. doi:10.1145/...

    work page arXiv 2014

  46. [47]

    Nikhil Rao, Hsiang-Fu Yu, Pradeep Ravikumar, and Inderjit S. Dhillon

  47. [48]

    Collaborative Filtering with Graph Information: Consistency and Scalable Methods. InAdvances in Neural Information Processing Sys- tems 28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015, Montreal, Quebec, Canada, Corinna Cortes, Neil D. Lawrence, Daniel D. Lee, Masashi Sugiyama, and Roman Gar- nett (Eds.). 2107–2115....

    2015

  48. [49]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt- Thieme. 2009. BPR: Bayesian Personalized Ranking from Implicit Feedback. InUAI 2009, Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence, Montreal, QC, Canada, June 18-21, 2009, Jeff A. Bilmes and Andrew Y. Ng (Eds.). AUAI Press, 452–461. https://www....

    2009

  49. [50]

    Yu Rong, Wenbing Huang, Tingyang Xu, and Junzhou Huang. 2020. DropEdge: Towards Deep Graph Convolutional Networks on Node Classification. In8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020. OpenReview.net. https://openreview.net/forum? id=Hkx1qkrKPr

    2020

  50. [51]

    Kyuyong Shin, Hanock Kwak, Su Young Kim, Max Nihlén Ramström, Jisu Jeong, Jung-Woo Ha, and Kyung-Min Kim. 2023. Scaling Law for Recommen- dation Models: Towards General-Purpose User Representations. InThirty- Seventh AAAI Conference on Artificial Intelligence, AAAI 2023, Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence, IAAI 2...

    work page doi:10.1609/aaai.v37i4.25582 2023

  51. [52]

    Jianing Sun, Yingxue Zhang, Wei Guo, Huifeng Guo, Ruiming Tang, Xiuqiang He, Chen Ma, and Mark Coates. 2020. Neighbor Interaction Aware Graph Convolution Networks for Recommendation. InProceedings of the 43rd Inter- national ACM SIGIR conference on research and development in Information Retrieval, SIGIR 2020, Virtual Event, China, July 25-30, 2020, Jimmy...

    work page doi:10.1145/3397271.3401123 2020

  52. [53]

    Peijie Sun, Le Wu, Kun Zhang, Xiangzhi Chen, and Meng Wang. 2024. Neighborhood-Enhanced Supervised Contrastive Learning for Collaborative Filtering.IEEE Trans. Knowl. Data Eng.36, 5 (2024), 2069–2081. doi:10.1109/ TKDE.2023.3317068

    work page arXiv 2024

  53. [54]

    Efros, and Moritz Hardt

    Yu Sun, Xiaolong Wang, Zhuang Liu, John Miller, Alexei A. Efros, and Moritz Hardt. 2020. Test-Time Training with Self-Supervision for Generalization under Distribution Shifts. InProceedings of the 37th International Conference on Machine Learning, ICML 2020, 13-18 July 2020, Virtual Event (Proceedings of Machine Learning Research, Vol. 119). PMLR, 9229–92...

    2020

  54. [55]

    Jian Tang, Meng Qu, Mingzhe Wang, Ming Zhang, Jun Yan, and Qiaozhu Mei

  55. [56]

    Line: Large-scale information network embedding,

    LINE: Large-scale Information Network Embedding. InProceedings of the 24th International Conference on World Wide Web, WWW 2015, Florence, Italy, May 18-22, 2015, Aldo Gangemi, Stefano Leonardi, and Alessandro Panconesi (Eds.). ACM, 1067–1077. doi:10.1145/2736277.2741093

    work page doi:10.1145/2736277.2741093 2015

  56. [57]

    Kipf, and Max Welling

    Rianne van den Berg, Thomas N. Kipf, and Max Welling. 2017. Graph Convo- lutional Matrix Completion.CoRRabs/1706.02263 (2017). arXiv:1706.02263 http://arxiv.org/abs/1706.02263

    work page arXiv 2017

  57. [58]

    Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph Attention Networks. In6th International Conference on Learning Representations, ICLR 2018, Vancouver, BC, Canada, April 30 - May 3, 2018, Conference Track Proceedings. OpenReview.net. https: //openreview.net/forum?id=rJXMpikCZ

    2018

  58. [59]

    Xiang Wang, Xiangnan He, Meng Wang, Fuli Feng, and Tat-Seng Chua. 2019. Neural Graph Collaborative Filtering. InProceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2019, Paris, France, July 21-25, 2019, Benjamin Piwowarski, Max Chevalier, Éric Gaussier, Yoelle Maarek, Jian-Yun Nie, and Fal...

    2019

  59. [60]

    doi:10.1145/3331184.3331267

    work page doi:10.1145/3331184.3331267

  60. [61]

    Xiang Wang, Hongye Jin, An Zhang, Xiangnan He, Tong Xu, and Tat-Seng Chua. 2020. Disentangled Graph Collaborative Filtering. InProceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, SIGIR 2020, Virtual Event, China, July 25-30, 2020, Jimmy X. Huang, Yi Chang, Xueqi Cheng, Jaap Kamps, Vanessa Murdoc...

    work page doi:10.1145/3397271 2020

  61. [62]

    Jiancan Wu, Xiang Wang, Fuli Feng, Xiangnan He, Liang Chen, Jianxun Lian, and Xing Xie. 2021. Self-supervised Graph Learning for Recommendation. InSIGIR ’21: The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, Virtual Event, Canada, July 11-15, 2021, Fernando Diaz, Chirag Shah, Torsten Suel, Pablo Castells, Ro...

    work page doi:10.1145/3404835.3462862 2021

  62. [63]

    Keyulu Xu, Chengtao Li, Yonglong Tian, Tomohiro Sonobe, Ken-ichi Kawarabayashi, and Stefanie Jegelka. 2018. Representation Learning on Graphs with Jumping Knowledge Networks. InProceedings of the 35th In- ternational Conference on Machine Learning, ICML 2018, Stockholmsmässan, Stockholm, Sweden, July 10-15, 2018 (Proceedings of Machine Learning Re- search...

    2018

  63. [64]

    Jheng-Hong Yang, Chih-Ming Chen, Chuan-Ju Wang, and Ming-Feng Tsai

  64. [65]

    InPro- ceedings of the 12th ACM Conference on Recommender Systems, RecSys 2018, Vancouver, BC, Canada, October 2-7, 2018, Sole Pera, Michael D

    HOP-rec: high-order proximity for implicit recommendation. InPro- ceedings of the 12th ACM Conference on Recommender Systems, RecSys 2018, Vancouver, BC, Canada, October 2-7, 2018, Sole Pera, Michael D. Ekstrand, Xavier Amatriain, and John O’Donovan (Eds.). ACM, 140–144. doi:10.1145/ 3240323.3240381

    work page arXiv 2018

  65. [66]

    Hamilton, and Jure Leskovec

    Rex Ying, Ruining He, Kaifeng Chen, Pong Eksombatchai, William L. Hamilton, and Jure Leskovec. 2018. Graph Convolutional Neural Networks for Web-Scale Recommender Systems. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2018, London, UK, August 19-23, 2018, Yike Guo and Faisal Farooq (Eds.). ACM, 974...

    work page arXiv 2018

  66. [67]

    Mert Yuksekgonul, Daniel Koceja, Xinhao Li, Federico Bianchi, Jed McCaleb, Xiaolong Wang, Jan Kautz, Yejin Choi, James Zou, Carlos Guestrin, et al. 2026. Learning to discover at test time.arXiv preprint arXiv:2601.16175(2026)

    work page arXiv 2026

  67. [68]

    Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. 2021. SoundStream: An End-to-End Neural Audio Codec. arXiv:2107.03312 [cs.SD]

    work page arXiv 2021

  68. [69]

    Interformer: Towards effective heterogeneous interaction learning for click-through rate prediction,

    Zhichen Zeng, Xiaolong Liu, Mengyue Hang, Xiaoyi Liu, Qinghai Zhou, Chaofei Yang, Yiqun Liu, Yichen Ruan, Laming Chen, Yuxin Chen, Yujia Hao, Jiaqi Xu, Jade Nie, Xi Liu, Buyun Zhang, Wei Wen, Siyang Yuan, Kai Wang, Wen-Yen Chen, Yiping Han, Huayu Li, Chunzhi Yang, Bo Long, Philip S. Yu, Hanghang Tong, and Jiyan Yang. 2024. InterFormer: Towards Effective H...

    work page doi:10.48550/arxiv.2411.09852 2024

  69. [70]

    Jiaqi Zhai, Lucy Liao, Xing Liu, Yueming Wang, Rui Li, Xuan Cao, Leon Gao, Zhaojie Gong, Fangda Gu, Jiayuan He, Yinghai Lu, and Yu Shi. 2024. Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations. InForty-first International Conference on Ma- chine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024. ...

    2024

  70. [71]

    Buyun Zhang, Liang Luo, Yuxin Chen, Jade Nie, Xi Liu, Shen Li, Yanli Zhao, Yuchen Hao, Yantao Yao, Ellie Dingqiao Wen, Jongsoo Park, Maxim Naumov, and Wenlin Chen. 2024. Wukong: Towards a Scaling Law for Large-Scale Recommendation. InForty-first International Conference on Ma- chine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024. OpenReview.net. h...

    2024

  71. [72]

    Buyun Zhang, Liang Luo, Xi Liu, Jay Li, Zeliang Chen, Weilin Zhang, Xiaohan Wei, Yuchen Hao, Michael Tsang, Wenjun Wang, Yang Liu, Huayu Li, Yasmine Badr, Jongsoo Park, Jiyan Yang, Dheevatsa Mudigere, and Ellie Wen. 2022. DHEN: A Deep and Hierarchical Ensemble Network for Large-Scale Click- Through Rate Prediction. arXiv:2203.11014 [cs.IR] https://arxiv.o...

    work page arXiv 2022

  72. [73]

    Buyun Zhang, Liang Luo, Xi Liu, Jay Li, Zeliang Chen, Weilin Zhang, Xiaohan Wei, Yuchen Hao, Michael Tsang, Wenjun Wang, Yang Liu, Huayu Li, Yasmine Badr, Jongsoo Park, Jiyan Yang, Dheevatsa Mudigere, and Ellie Wen. 2022. DHEN: A Deep and Hierarchical Ensemble Network for Large-Scale Click- Through Rate Prediction.CoRRabs/2203.11014 (2022). doi:10.48550/A...

    work page internal anchor Pith review doi:10.48550/arxiv 2022

  73. [74]

    Gaowei Zhang, Yupeng Hou, Hongyu Lu, Yu Chen, Wayne Xin Zhao, and Ji- Rong Wen. 2024. Scaling Law of Large Sequential Recommendation Models. In Proceedings of the 18th ACM Conference on Recommender Systems, RecSys 2024, Bari, Italy, October 14-18, 2024, Tommaso Di Noia, Pasquale Lops, Thorsten Joachims, Katrien Verbert, Pablo Castells, Zhenhua Dong, and B...

    work page doi:10.1145/3640457.3688129 2024

  74. [75]

    Yinan Zhang, Pei Wang, Xiwei Zhao, Hao Qi, Jie He, Junsheng Jin, Changping Peng, Zhangang Lin, and Jingping Shao. 2022. IA-GCN: Interactive Graph Convolutional Network for Recommendation.CoRRabs/2204.03827 (2022). doi:10.48550/ARXIV.2204.03827 arXiv:2204.03827

    work page doi:10.48550/arxiv.2204.03827 2022

  75. [76]

    Han Zhu, Xiang Li, Pengye Zhang, Guozheng Li, Jie He, Han Li, and Kun Gai. 2018. Learning Tree-based Deep Model for Recommender Systems. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2018, London, UK, August 19-23, 2018, Yike Guo and Faisal Farooq (Eds.). ACM, 1079–1088. doi:10.1145/3219819.3219826

    work page doi:10.1145/3219819.3219826 2018