更多精彩内容可以关注公众号:缄默笔记
主要介绍2021年顶会论文中,特征交叉相关的CTR模型,主要介绍7篇论文模型,如下图所示:
下文主要给出各论文标题,公众号排版更美观,各论文导读和模型介绍见:
2021年「顶会论文」特征交叉相关CTR模型汇总https://links.jianshu.com/go?to=https%3A%2F%2Fmp.weixin.qq.com%2Fs%2FbxWdnlAr4jcOVPBNdYd28Q
论文:2: Field-matrixed Factorization Machines for Recommender Systems
Link:https://arxiv.org/pdf/2102.12994.pdf
中文参考:FmFM:FM类浅层CTR模型统一框架
论文:DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems
Link:https://arxiv.org/pdf/2008.13535.pdf
中文参考:DCN-M:Google提出改进版DCN
论文:Looking at CTR Prediction Again: Is Attention All You Need?
Link:https://arxiv.org/pdf/2105.05563.pdf
中文参考:SAM:重新思考CTR模型中Attention的作用
论文:FINT: Field-aware INTeraction Neural Network For CTR Prediction
Link:https://arxiv.org/pdf/2107.01999.pdf
中文参考:FINT:基于特征域交叉的CTR模型
论文:XCrossNet: Feature Structure-Oriented Learning for Click-Through Rate Prediction
Link:https://arxiv.org/pdf/2104.10907.pdf
中文参考:XCrossNet:面向特征结构学习的CTR模型
论文:Disentangled Self-Attentive Neural Networks for Click-Through Rate Prediction
Link:https://arxiv.org/pdf/2101.03654.pdf
中文参考:DESTINE:基于解耦自注意网络的CTR模型
论文:Enhancing Explicit and Implicit Feature Interactions via Information Sharing for Parallel Deep CTR Models
Link:https://dlp-kdd.github.io/assets/pdf/DLP-KDD_2021_paper_12.pdf
中文参考:EDCN:通过信息共享增强并行CTR模型显式和隐式特征交叉