Pith. sign in

REVIEW 4 major objections 6 minor 37 references

HIT Model: A Hierarchical Interaction-Enhanced Two-Tower Model for Pre-Ranking Systems

T0 review · 4 major / 6 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 Two-tower pre-ranking model with generators and multi-head matching lifts online GMV by 1.66% and ROI by 1.55% while keeping serving latency near vanilla two-tower.

desk verdict HIT reports a deployed pre-ranking model with real gains, but the printed generation loss contradicts the paper's own cross-tower interaction story, so the code is the make-or-break. read the letter →

arxiv 2505.19849 v2 pith:JKG7QNJS submitted 2025-05-26 cs.IR

Haoqiang Yang , Congde Yuan , Kun Bai , Mengzhuo Guo , Wei Yang , Chao Zhou This is my paper
classification cs.IR
keywords pre-rankingtwo-towermodelonlinedisplayadvertisinguser-adinteractionmulti-headrepresentergeneratornetworkcosinesimilaritylossCTRprediction
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 the Hierarchical Interaction-Enhanced Two-Tower (HIT) model for pre-ranking in online display advertising, and claims it closes most of the accuracy gap to all-to-all interaction models without paying their inference cost. HIT augments a vanilla two-tower backbone with two components: generators that map static user/ad features to 'mimic' vectors meant to carry coarse-grained cross-tower information into each tower, and multi-head representers that project each tower's embedding into several subspaces and score by a max-then-sum over user-interest/ ad-attribute matches. On three public datasets HIT reports the highest AUC and lowest cross-entropy among vanilla, early-interaction, late-interaction, and all-to-all baselines. In a five-day online A/B test on Tencent's advertising platform, HIT reports a 1.66% increase in Gross Merchandise Volume and a 1.55% increase in Return on Investment over the MVKE control, at response latency close to the vanilla two-tower. If correct, HIT offers a drop-in architectural upgrade for pre-ranking systems that need both accuracy and millisecond-level serving.

What carries the argument

The load-bearing machinery is the pair of generators $g_k(\cdot)$ (two-layer MLPs taking only static features) that produce mimic vectors $\mathbf{r}^m_k$, trained by the generation loss $\mathcal{L}_{g_u}=-\frac{1}{N}\sum_i (y_i\,\mathrm{Dist}(\mathbf{r}_u,\mathbf{r}^m_1)+(1-y_i)\,\mathrm{Dist}(\mathbf{r}_u,\mathbf{r}^m_2))$ (and its ad-tower analogue), together with the multi-head representers of Eqs. (5)-(6) and the max-then-sum scoring rule of Eq. (7). The generators' outputs are L2-normalized, concatenated with the user dense vector $\mathbf{e}_u$, and fed into the tower DNN, so the generated information influences the embeddings that the representers see. 'Stop gradient' operations on both the generator input/output and the representer output used in the generation loss are intended to let the generators fit high-level representations without disturbing the backbone or letting the auxiliary loss interfere with the main training. The multi-head representer then partitions each tower's embedding into $J$ subspaces; the matching score is the sum over user heads of the maximum similarity to any ad head, which the paper interprets as finding the most relevant ad attribute for each user interest while keeping ad embeddings pre-cacheable.

What would settle it

Inspect the released source code for the generation loss: if the user-tower generator's target in Eq. (8) is the user tower's own multi-head representation $\mathbf{r}_u$ with the printed negative sign, then the generator never receives ad-side information and the paper's central mechanism is contradicted; alternatively, run an ablation that trains each generator against the opposite tower's representation instead of the same tower's—if HIT's AUC is unchanged, the claimed interaction is not the source of the gains.

Watch

Extended reading notes

Core claim

The paper's central claim is that a two-tower pre-ranking model can be made 'interaction-enhanced' by (1) adding a dual-generator module, in which two small MLPs (one per label type, here positive/negative) produce 'mimic' vectors from static features that are concatenated into the tower inputs, with a cosine-similarity generation loss aligning those mimic vectors with the tower's high-level representer output; and (2) replacing the single dot product with a multi-head representer that projects user and ad embeddings into $J$ latent subspaces and computes $\hat{y} = \sum_{j_u=1}^{J}\max_{j_a}(\mathbf{r}_{u,(j_u)})^\top \mathbf{r}_{a,(j_a)}$. The paper argues that the generators simulate the opposite tower's representations, 'exposing' each tower to cross-domain information before encoding, and that the multi-head representer captures multi-faceted user interests and multi-dimensional ad attributes. It further claims that this design preserves the two-tower serving advantage because ad representations are pre-computed and cached; only the user side is computed at request time. The paper reports consistent offline superiority across MovieLens, Amazon (Electro), and Alibaba datasets, and online gains in GMV and ROI on Tencent's platform.

Load-bearing premise

The model only deserves the name 'interaction-enhanced' if each tower's generator is genuinely trained on information from the other tower; if the training loss only mirrors a tower's own representation, the cross-domain interaction story collapses, even if the accuracy numbers survive.

Editorial extensions

If this is right

  • Pre-ranking systems can adopt HIT as a drop-in upgrade to an existing two-tower pipeline, since it adds only static-feature MLPs and linear head projections to the served model.
  • The reported online A/B gains (GMV +1.66%, ROI +1.55%) suggest that the interaction improvements measured offline by AUC translate into revenue-relevant ranking changes at scale.
  • The max-then-sum matching generalizes late-interaction scoring (e.g., MaxSim-style matching) to a pre-cached two-tower setting, providing a template for other retrieval/pre-ranking models.
  • Because HIT separates generators by label type, the architecture can be extended to multi-task or multi-class labels by increasing the number of generators $K$, as the paper itself notes.
  • The 'stop gradient' recipe shows how to add auxiliary representation-mimicking losses to a two-tower model without destabilizing the main CTR training.

Reading between the lines

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

  • A close reading of Eq. (8) suggests the printed generation loss may train each generator against its own tower's multi-head representation rather than the opposite tower's; if the released code matches the equations, the model's gains would come from a form of self-distillation or regularization rather than the claimed cross-tower interaction, and the mechanism would need to be re-described even i
  • The static-only generator input trades away dynamic features at training time by design; a natural extension is train-time distillation of dynamic signals into the static generator, which could recover some of the lost information without online cost.
  • The paper's hierarchy (coarse-grained generation then fine-grained matching) parallels distillation-plus-MaxSim structures in retrieval; one could test whether a single well-chosen generator shared across label types performs comparably, isolating the value of the dual-generator split.
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 / 6 minor

Summary. The paper proposes HIT, a two-tower pre-ranking model for online display advertising, augmented with two components: generators that are claimed to pre-generate holistic vectors incorporating coarse-grained user-ad interactions, and multi-head representers that project embeddings into multiple latent subspaces for fine-grained matching. The authors report offline improvements over ten baselines on MovieLens, Amazon (Electro), and Alibaba datasets, and online A/B test gains of +1.66% GMV and +1.55% ROI on Tencent's platform, with code released. The central novelty is the generation loss, which is intended to transfer coarse-grained cross-tower information between the user and ad towers.

Significance. If the mechanism worked as described, HIT would be a valuable industrial contribution: it preserves two-tower serving efficiency while incorporating interaction signals, and the paper provides public-dataset experiments, ablation studies, an online A/B test, and open-source code. However, the printed equations do not implement the claimed cross-tower generation, and the sign of the generation loss contradicts its stated objective. The empirical results cannot be interpreted as evidence for the proposed interaction mechanism until this is resolved.

major comments (4)
  1. [Section 4.3.1, Eq. (8)] The generation loss L_gu is written as -1/N Σ [y_i Dist(r_u, r^m_1) + (1-y_i) Dist(r_u, r^m_2)], where r_u is the user tower's own multi-head representation. This compares the user generator outputs with the same tower's representation, not with the ad tower's representation as stated in Section 4.1 ('the generators in the user tower simulate ad representations'). The cross-domain interaction mechanism, which is the paper's central contribution, is therefore absent from the printed mathematics. Furthermore, since Dist is defined as cosine distance, the leading negative sign makes the objective maximize rather than minimize the distance, contradicting the sentence in Section 4.3.1 that the loss 'is designed to minimize the distance between r_u and r^m_k'. The authors should either correct Eq. (8) (and the corresponding ad-tower loss) or provide evidence from the released code that the implemented loss differs from the printed one; as written, the claim of interaction enhancement is unsupported.
  2. [Section 4.1 and Eq. (3)] The description of the generator in Eq. (3) is ambiguous with respect to the generator's target. The text says r^m_{k,δ} denotes 'mimic users/ads' and that the user-tower generator 'simulates ad representations', but the generator takes only user static features x_ub as input and Eq. (8) ties its output to r_u. If the generator is meant to output an ad-like vector from user features, the loss should compare it with r_a (or an ad-side quantity); if it is meant to output a user-like vector, the cross-tower story in the introduction is incorrect. Either way, the manuscript must clarify what the generator produces and against which representation it is trained.
  3. [Section 5.2, Table 2] The claim that 'the improvement is significant at α=0.01' appears without any reported standard deviations, error bars, or description of the statistical test (e.g., paired t-test across repetitions). The table reports only average values over 10 repetitions, which is insufficient to substantiate the significance statement for the paper's main offline result.
  4. [Section 5.5, Table 4] The online A/B test reports +1.66% GMV and +1.55% ROI relative to the MVKE control, but no confidence intervals, significance levels, or information about the number of users/impressions assigned to each arm are provided. Given that these numbers are used to support the deployment claim, a statistical characterization is necessary to assess whether the gains are meaningful and not within normal fluctuation.
minor comments (6)
  1. [Section 4.3.1, Eq. (8)] The sample index i does not appear on r_u, r^m_1, or r^m_2; clarify that these quantities are per-sample.
  2. [Section 4.2, Eq. (7)] The notation j_u and j_a for the head indices is inconsistent with the earlier use of j; use a consistent subscript convention.
  3. [Figure 3] The labels 'Mimic User' and 'Mimic Ad' next to the generators are inconsistent with Eq. (8), which trains the generators against the same tower's representation.
  4. [Section 5.1] Hyperparameters α, J, and z are fixed without any sensitivity analysis or tuning procedure; report how these were chosen.
  5. [Section 5.5, Figure 6] The 'Success Rate' metric is not defined; state what it measures.
  6. [General] The paper uses the ACM template with 'Conference acronym 'XX' and copyright 2018 placeholders; these need to be updated before publication.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: HIT's empirical claims are benchmarked externally; the generation-loss inconsistency is an internal correctness issue, not a circular derivation.

full rationale

The derivation chain of HIT is not circular in the sense targeted by this analysis. The main objective (Eq. 9) is cross-entropy on observed click labels; the generation loss (Eq. 8) is an auxiliary self-distillation term with stop-gradient operations (Sec. 4.3.3), not a fitted quantity that is later reported as a prediction. Offline AUC/CEloss comparisons in Table 2 are against external baselines (DSSM, DAT, MVKE, IntTower, Wide&Deep, DCN, AutoInt, COLD), and the online A/B test in Sec. 5.5 compares against a deployed MVKE control on GMV/ROI. No parameter is fitted to a subset and then 'predicted' on a closely related quantity. The self-citations (Refs. [4], [26]) appear as background examples of Tencent advertising research and are not load-bearing for the HIT architecture. The most salient concern in the paper is not circularity but internal consistency: Eq. (8) contains only r_u, the user tower's own multi-head representation, and with Dist defined as cosine distance the leading negative sign would maximize rather than minimize distance. This would undermine the claimed cross-tower mimicry mechanism if the implementation matches the printed equations. However, an equation that fails to implement the stated mechanism is a correctness/consistency problem, not a case of the paper's conclusions being equivalent to its inputs by construction. Therefore the circularity score is 0.

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

The central claim rests on no physical constants or fitted scaling laws. The free parameters are ordinary ML hyperparameters (loss weight, head count, generator count, head dimension) chosen by hand or by small grid sweeps. The modeling axioms are typical for the two-tower advertising setting: cached ad embeddings bound the architecture, static features are assumed sufficient for the generators, and the random 80/20 split is assumed leak-free. No invented entities are introduced.

free parameters (4)
  • alpha (generation loss weight) = 1e-3
    Balances cross-entropy and generation losses in Eq. (10); chosen by hand; no ablation sweeps it.
  • J (number of representer heads) = 2
    Tuned over 1 to 6 on all datasets (Figure 4); the paper recommends small J, but the choice is data-driven and central to Eq. (7).
  • K (number of generators) = 2
    Set equal to the number of label types (positive/negative) as stated after Eq. (3); a design choice.
  • z (representer output dimension) = 16
    Hand-set along with the DNN widths [300,300,32], embedding dim 32, and generator MLP [64,32] in Section 5.1; p = z*J depends on it.
assumptions (4)
  • standard math Cosine distance and L2 normalization behave as stated in Eqs. (2), (7) and (8).
    The model normalizes embeddings to unit length and uses cosine-based losses without proof; standard background.
  • domain assumption Pre-cached ad embeddings are the binding serving constraint.
    The whole design keeps the two-tower pre-caching property; all-to-all models are excluded from online deployment on this basis (Sections 2 and 5.5).
  • domain assumption Static features alone suffice for generator input.
    Section 4.1 argues static features carry fundamental interests; the ablation in Table 3 Panel C supports this empirically, but it remains an asserted modeling choice.
  • domain assumption Random 80/20 split of interaction samples is a valid evaluation protocol.
    Section 5.1 follows Huang et al. [6]; because user identity can appear in both train and test, shared user features may leak and inflate AUC, and the paper does not address this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HIT Model: A Hierarchical Interaction-Enhanced Two-Tower Model for Pre-Ranking Systems." pith.science (2026). https://pith.science/paper/JKG7QNJS

@misc{pith2026250519849,
  author       = {Pith},
  title        = {Pith review of: HIT Model: A Hierarchical Interaction-Enhanced Two-Tower Model for Pre-Ranking Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JKG7QNJS}},
  note         = {Machine review of arXiv:2505.19849}
}
abstract

Online display advertising platforms rely on pre-ranking systems to efficiently filter and prioritize candidate ads from large corpora, balancing relevance to users with strict computational constraints. The prevailing two-tower architecture, though highly efficient due to its decoupled design and pre-caching, suffers from cross-domain interaction and coarse similarity metrics, undermining its capacity to model complex user-ad relationships. In this study, we propose the Hierarchical Interaction-Enhanced Two-Tower (HIT) model, a new architecture that augments the two-tower paradigm with two key components: $\textit{generators}$ that pre-generate holistic vectors incorporating coarse-grained user-ad interactions through a dual-generator framework with a cosine-similarity-based generation loss as the training objective, and $\textit{multi-head representers}$ that project embeddings into multiple latent subspaces to capture fine-grained, multi-faceted user interests and multi-dimensional ad attributes. This design enhances modeling effectiveness without compromising inference efficiency. Extensive experiments on public datasets and large-scale online A/B testing on Tencent's advertising platform demonstrate that HIT significantly outperforms several baselines in relevance metrics, yielding a $1.66\%$ increase in Gross Merchandise Volume and a $1.55\%$ improvement in Return on Investment, alongside similar serving latency to the vanilla two-tower models. The HIT model has been successfully deployed in Tencent's online display advertising system, serving billions of impressions daily. The code is available at https://github.com/HarveyYang123/HIT_model.

Figures

Figures reproduced from arXiv: 2505.19849 by the authors.

Figure 1
Figure 1. Online display advertising system. The pre-ranking system, encompassing the targeting and scoring tasks highlighted in the shaded area of [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. The proposed HIT model is comprised of two new com [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figure 2
Figure 2. Illustrations of four types of two-tower model structures. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figures from the paper (4 more)
Figure 3
Figure 3. Figure 3: Overview of the HIT model architecture. 4.1 Coarse-grained Generation The key component of coarse-grained generation is the generator, which is a simple multilayer perceptron network 𝑔𝑘 (·) with two hidden layers that accepts only static features (x𝑢𝑏 or x𝑎𝑏) as input …
Figure 4
Figure 4. Figure 4: Evaluation metrics concerning 𝐽. Q2.5. What is the best number of heads in our multi-head representer? A critical setting for capturing fine-grained semantic relationships between user and ad embeddings is the number of heads 𝐽. We have tuned different values ranging f…
Figure 6
Figure 6. Figure 6: Results of online efficiency. Millisecond and Success Rate represent the inference time and the success rate of responses under QPS=35,000. illustrated in [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 5
Figure 5. Figure 5: Visualization of mimic/original representations. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

37 extracted references · 12 canonical work pages

  1. [1]

    Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, et al

  2. [2]

    Jiaping Gui, Stuart Mcilroy, Meiyappan Nagappan, and William GJ Halfond. 2015. Truth in advertising: The hidden cost of mobile ads for software developers. In 2015 IEEE/ACM 37th IEEE International Conference on Software Engineering , Vol. 1. IEEE, 100–110

    work page 2015

  3. [3]

    Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. 2017. DeepFM: a factorization-machine based neural network for CTR prediction.arXiv preprint arXiv:1703.04247 (2017)

    arXiv 2017

  4. [4]

    Mengzhuo Guo, Wuqi Zhang, Congde Yuan, Binfeng Jia, Guoqing Song, Hua Hua, Shuangyang Wang, and Qingpeng Zhang. 2024. A Bayesian Multi-Armed Bandit Algorithm for Bid Shading in Online Display Advertising. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management . 4506–4513

    work page 2024

  5. [5]

    Po-Sen Huang, Xiaodong He, Jianfeng Gao, Li Deng, Alex Acero, and Larry Heck. 2013. Learning deep structured semantic models for web search using clickthrough data. In Proceedings of the 22nd ACM international conference on Information & Knowledge Management . 2333–2338

    2013

  6. [6]

    Tongwen Huang, Zhiqi Zhang, and Junlin Zhang. 2019. FiBiNET: combining fea- ture importance and bilinear feature interaction for click-through rate prediction. In Proceedings of the 13th ACM conference on recommender systems . 169–177

    2019

  7. [7]

    Samuel Humeau, Kurt Shuster, Marie-Anne Lachaux, and Jason Weston. 2019. Poly-encoders: Transformer architectures and pre-training strategies for fast and accurate multi-sentence scoring. arXiv preprint arXiv:1905.01969 (2019)

    arXiv 2019

  8. [8]

    Cong Jiang, Zhongde Chen, Bo Zhang, Yankun Ren, Xin Dong, Lei Cheng, Xinxing Yang, Longfei Li, Jun Zhou, and Linjian Mo. 2024. GATS: Generative Audience Targeting System for Online Advertising. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval . 2920–2924

    work page 2024

Show all 37 references
  1. [9]

    Dan Li, Yang Yang, Hongyin Tang, Jiahao Liu, Qifan Wang, Jingang Wang, Tong Xu, Wei Wu, and Enhong Chen. 2022. VIRT: Improving representation-based text matching via virtual interaction. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing ...

    2022

  2. [10]

    Xiangyang Li, Bo Chen, HuiFeng Guo, Jingjie Li, Chenxu Zhu, Xiang Long, Sujian Li, Yichao Wang, Wei Guo, Longxia Mao, et al. 2022. Inttower: the next generation of two-tower model for pre-ranking system. In Proceedings of the 31st ACM International Conference on Information & ...

    2022

  3. [11]

    Yuxiang Lu, Yiding Liu, Jiaxiang Liu, Yunsheng Shi, Zhengjie Huang, Shikun Feng Yu Sun, Hao Tian, Hua Wu, Shuaiqiang Wang, Dawei Yin, et al. 2022. Ernie- search: Bridging cross-encoder with dual-encoder via self on-the-fly distillation for dense passage retrieval. arXiv prepri...

    2022 arXiv

  4. [12]

    H Brendan McMahan, Gary Holt, David Sculley, Michael Young, Dietmar Ebner, Julian Grady, Lan Nie, Todd Phillips, Eugene Davydov, Daniel Golovin, et al

  5. [13]

    Wentao Ouyang, Xiuwu Zhang, Shukui Ren, Li Li, Kun Zhang, Jinmei Luo, Zhaojie Liu, and Yanlong Du. 2021. Learning graph meta embeddings for cold-start ads in click-through rate prediction. In Proceedings of the 44th International ACM SIGIR Conference on Research and Developmen...

    2021

  6. [14]

    Yanru Qu, Han Cai, Kan Ren, Weinan Zhang, Yong Yu, Ying Wen, and Jun Wang

  7. [15]

    Weiping Song, Chence Shi, Zhiping Xiao, Zhijian Duan, Yewen Xu, Ming Zhang, and Jian Tang. 2019. Autoint: Automatic feature interaction learning via self- attentive neural networks. InProceedings of the 28th ACM international conference on information and knowledge management ...

    2019

  8. [16]

    Laurens Van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-SNE. Journal of machine learning research 9, 11 (2008)

    2008

  9. [17]

    In 2016 IEEE 16th international conference on data mining (ICDM)

    Product-based neural networks for user response prediction. In 2016 IEEE 16th international conference on data mining (ICDM) . IEEE, 1149–1154

    2016

  10. [18]

    Yaozheng Wang, Dawei Feng, Dongsheng Li, Xinyuan Chen, Yunxiang Zhao, and Xin Niu. 2016. A mobile recommendation system based on logistic regression and gradient boosting decision trees. In 2016 international joint conference on neural networks (IJCNN). IEEE, 1896–1902

    2016

  11. [19]

    Zekun Wang, Wenhui Wang, Haichao Zhu, Ming Liu, Bing Qin, and Furu Wei

  12. [20]

    Ruoxi Wang, Bin Fu, Gang Fu, and Mingliang Wang. 2017. Deep & cross network for ad click predictions. In Proceedings of the ADKDD’17 . 1–7

    2017

  13. [21]

    Yanheng Wei, Lianghua Huang, Yanhao Zhang, Yun Zheng, and Pan Pan. 2022. An intelligent advertisement short video production system via multi-modal retrieval. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval . 3368–3372

    2022

  14. [22]

    Melanie Wiese, Carla Martínez-Climent, and Dolores Botella-Carrubi. 2020. A framework for Facebook advertising effectiveness: A behavioral perspective. Journal of Business Research 109 (2020), 76–87

    2020

  15. [23]

    Zhenhui Xu, Meng Zhao, Liqun Liu, Lei Xiao, Xiaopeng Zhang, and Bifeng Zhang

  16. [24]

    Zhe Wang, Liqin Zhao, Biye Jiang, Guorui Zhou, Xiaoqiang Zhu, and Kun Gai

  17. [25]

    Yantao Yu, Weipeng Wang, Zhoutian Feng, and Daiyue Xue. 2021. A dual aug- mented two-tower model for online large-scale recommendation. DLP-KDD (2021)

    2021

  18. [26]

    Congde Yuan, Mengzhuo Guo, Chaoneng Xiang, Shuangyang Wang, Guoqing Song, and Qingpeng Zhang. 2022. An actor-critic reinforcement learning model for optimal bidding in online display advertising. In Proceedings of the 31st ACM International Conference on Information & Knowledg...

    2022

  19. [27]

    Hengyu Zhang, Junwei Pan, Dapeng Liu, Jie Jiang, and Xiu Li. 2024. Deep Pattern Network for Click-Through Rate Prediction. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1189–1199

    2024

  20. [28]

    Wei Zhang, Dai Li, Chen Liang, Fang Zhou, Zhongke Zhang, Xuewei Wang, Ru Li, Yi Zhou, Yaning Huang, Dong Liang, et al. 2024. Scaling User Modeling: Large-scale Online User Representations for Ads Personalization in Meta. In Companion Proceedings of the ACM on Web Conference 20...

    2024

  21. [29]

    Xiangyu Zhao, Changsheng Gu, Haoshenglun Zhang, Xiwang Yang, Xiaobing Liu, Jiliang Tang, and Hui Liu. 2021. Dear: Deep reinforcement learning for online advertising impression in recommender systems. In Proceedings of the AAAI conference on artificial intelligence , Vol. 35. 750–758

    2021

  22. [30]

    Xinyang Yi, Ji Yang, Lichan Hong, Derek Zhiyuan Cheng, Lukasz Heldt, Aditee Kumthekar, Zhe Zhao, Li Wei, and Ed Chi. 2019. Sampling-bias-corrected neural modeling for large corpus item recommendations. In Proceedings of the 13th ACM conference on recommender systems . 269–277

    2019

  23. [31]

    Guorui Zhou, Xiaoqiang Zhu, Chenru Song, Ying Fan, Han Zhu, Xiao Ma, Yanghui Yan, Junqi Jin, Han Li, and Kun Gai. 2018. Deep interest network for click-through rate prediction. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining ...

    2018

  24. [36]

    Zuowu Zheng, Changwang Zhang, Xiaofeng Gao, and Guihai Chen. 2022. HIEN: hierarchical intention embedding network for click-through rate prediction. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval . 322–331

    2022

  25. [2013]

    In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining

    Ad click prediction: a view from the trenches. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining . 1222–1230

  26. [2016]

    In Proceedings of the 1st workshop on deep learning for recommender systems

    Wide & deep learning for recommender systems. In Proceedings of the 1st workshop on deep learning for recommender systems . 7–10

  27. [2020]

    CoRR abs/2007.16122 (2020)

    COLD: Towards the Next Generation of Pre-Ranking System. CoRR abs/2007.16122 (2020). arXiv preprint arXiv:2007.16122 (2020)

    2020 arXiv

  28. [2021]

    arXiv preprint arXiv:2112.08723 (2021)

    Distilled dual-encoder model for vision-language understanding. arXiv preprint arXiv:2112.08723 (2021)

    2021 arXiv

  29. [2022]

    In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

    Mixture of virtual-kernel experts for multi-objective user profile modeling. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 4257–4267

Pith tools

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