Article Details

Title · Authors · Abstract · Keywords · DOI · References

Research on Advertisement CTR Prediction Based on Multi-branch Feature Fusion Models

WANG Zhige1, PAN Shengke2

(1.School of Computing, Guangzhou College of Applied Science and Technology, Zhaoqing 526072, China; 2.Guangdong Zhengda Heshun Project Management Co., Ltd., Heyuan 517000, China)

Section：Artificial Intelligence · Vol. 2026, Issue 20 · Pages：42-48,55 · 2096-4706（2026）20-0042-08 · CLC number：TP391.3；TP183 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2026.20.009

Abstract

Click-Through Rate (CTR) prediction is a core task in online advertising and recommender systems, and its accuracy directly influences both user experience and platform revenue. To overcome the limitations of existing models in multimodal feature interaction modeling, dynamic context awareness, and branch-level feature fusion, this paper presents MFNet, a multi-branch feature fusion CTR prediction model that integrates disentangled attention, contextual feature extraction, and field-matrixed factorization machines. The multi-branch fusion layer of the model couples dynamic contextual awareness, disentangled attention-based interactions, and field-matrixed factorization-based association modeling to deeply fuse feature representations. Experiments on the public Criteo and MovieLens-1M datasets show that, compared with the DESTINE model, MFNet improves AUC by 0.07 and 0.44 percentage points and reduces the cross-entropy loss (Logloss) by 0.0071 and 0.0013, respectively.

Keywords

CTR prediction; multi-branch feature fusion; disentangled attention; contextual feature extraction; field-matrixed factorization machine

References

  1. [1] Zhang W N，Qin J R，Guo W，et al. Deep Learning for Click-Through Rate Estimation [C]//Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence Survey Track，2021：4695-4703.
  2. [2] Rendle S. Factorization machines [C]//2010 IEEE International Conference on Data Mining，2010：995-1000.
  3. [3] Pan J W，Xu J，Ruiz A L，et al. Field-weighted Factorization Machines for Click-Through Rate Prediction in Display Advertising [PP/OL].arXiv:1806.03514 [cs.LG].https://arxiv.org/abs/1806.03514.
  4. [4] Sun Y，Pan J W，Zhang A，et al. FM2: Field-matrixed Factorization Machines for Recommender Systems [PP/OL].arXiv:2102.12994 [cs.IR].https://arxiv.org/abs/2102.12994.
  5. [5] He X N，Chua T S. Neural Factorization Machines for Sparse Predictive Analytics [C]//Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval，2017：355-364.
  6. [6] Covington P，Adams J，Sargin E. Deep Neural Networks for YouTube Recommendations [C]//ACM Conference on Recommender Systems，2016：191-198.
  7. [7] Liu B，Tang R M，Chen Y Z，et al. Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction [C]//WWW '19：The World Wide Web Conference，2019：1119-1129.
  8. [8] Guo H F，Chen B，Tang R M，et al. An Embedding Learning Framework for Numerical Features in CTR Prediction [C]//KDD'21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining，2021：2910-2918.
  9. [9] Xiao J，Ye H，He X N，et al. Attentional Factorization Machines: Learning the Weight of Feature Interactions via Attention Networks [C]//Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence Main track，2017：3119-3125.
  10. [10] Vaswani A，Shazeer N，Parmar N，et al. Attention is All You Need [C]//NIPS'17: Proceedings of the 31st International Conference on Neural Information Processing Systems，2017：6000-6010.
  11. [11] Song W，Shi C，Xiao Z，et al. Autoint: Automatic Feature Interaction Learning via Self-attentive Neural Networks [C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management，2019：1161-1170.
  12. [12] Xu Y，Zhu Y，Yu F，et al. Disentangled Self-Attentive Neural Networks for Click-Through Rate Prediction [C]//Proceedings of the 30th ACM International Conference on Information & Knowledge Management，2021：3553-3557.
  13. [13] Leng J，Liu Y，Wang Z，et al. CrossNet: Detecting Objects as Crosses [J].IEEE Transactions on Multimedia，2022，24：861-875.
  14. [14] Cheng H-T，Koc L，Harmsen J，et al. Wide & Deep Learning for Recommender Systems [C]//DLRS 2016: Proceedings of the 1st Workshop on Deep Learning for Recommender Systems，2016：7-10.
  15. [15] Guo H-F，Tang R-M，Ye Y-M，et al. DeepFM: A Factorization-machine Based Neural Network for CTR Prediction [C]//IJCAI'17: Proceedings of the 26th International Joint Conference on Artificial Intelligence，2017：1725-1731.
  16. [16] Lian J X，Zhou X H，Zhang F Z，et al. xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems [PP/OL].arXiv:1803.05170 [cs.LG].https://arxiv.org/abs/1803.05170.
  17. [17] Yu Y T，Wang Z，Yuan B. An Input-aware Factorization Machine for Sparse Prediction [C]//Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence，2019：1466-1472.
  18. [18] Lu W T，Yu Y T，Chang Y Z，et al. A Dual Input-aware Factorization Machine for CTR Prediction [C]//Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI-20)，2020：3139-3145.
  19. [19] Huang T，Zhang Z，Zhang J. FiBiNET: Combining Feature Importance and Bilinear Feature Interaction for Click-through Rate Prediction [C]//Proceedings of the 13th ACM Conference on Recommender Systems，2019：169-177.
  20. [20] Hu J，Shen L，Albanie S，et al. Squeeze-and-Excitation Networks [C]//IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，42（8）：2011-2023.
  21. [21] Wang F Y，Wang Y X，Li D S，et al. Enhancing CTR Prediction with Context-Aware Feature Re-presentation Learning [C]//SIGIR'22: Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval，2022：343-352.
  22. [22] Li Z Y，Cheng W，Chen Y. Interpretable Click-Through Rate Prediction through Hierarchical Attention [C]//WSDM'20: Proceedings of the 13th International Conference on Web Search and Data Mining，2020：313-321.
  23. [23] Li Y J，Tarlow D，Brockschmidt M ，et al. Gated Graph Sequence Neural Networks [PP/OL].arXiv:1511.05493 [cs.LG].https://arxiv.org/abs/1511.05493?context = cs.AI.
  24. [24] Kipf T N，Welling M. Semi-supervised Classification with Graph Convolutional Networks [PP/OL].arXiv:1609.02907 [cs.LG].https://arxiv.org/abs/1609.02907.
  25. [25] Li Z，Cui Z，Wu S，et al. Fi-GNN: Modeling Feature Interactions via Graph Neural Networks for CTR Prediction [C]//The 28th ACM International Conference on Information and Knowledge Management，2019：539-548.
  26. [26] Qu Y R，Cai H，Ren K，et al. Product-Based Neural Networks for User Response Prediction[PP/OL].arXiv:1611.00144 [cs.LG].https://arxiv.org/abs/1611.00144?context = cs.
  27. [27] Zhao Z S，Yang S，Liu G H，et al. FINT: Field-aware INTeraction Neural Network For CTR Prediction [PP/OL].arXiv:2107.01999 [cs.IR].https://arxiv.org/abs/2107.01999.
  28. [28] Wang R，Shivanna R，Cheng D，et al. DCNV2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems [C]//Proceedings of the Web Conference 2021，2021：1785-1797.
  29. [29] Yan B C，Wang P J，Zhang K，et al. APG: Adaptive Parameter Generation Network for Click-Through Rate Prediction [J].Advances in Neural Information Processing Systems，2022，35：24740-24752.

Funding

Course Group Teaching Team of Intelligent Algorithm Research Institute, Guangzhou University of Applied Science and Technology (2025TD005)

About the authors

Wang Zhige (1997-), male, Han ethnicity, from Chizhou, Anhui, teaching assistant, master's student, research interests include click-through rate prediction and recommendation and object detection; Pan Shengke (1992-), male, Han ethnicity, from Heyuan, Guangdong, AI engineer, bachelor's degree, research interests include recommender systems.