EdmundYan-CSDN博客

原创 SIGIR2020|基于自注意力机制和多鉴别器的序列推荐

Sequential Recommendation with Self-Attentive Multi-Adversarial Networkhttps://arxiv.org/pdf/2005.10602.pdf背景序列推荐中，对于CTR的预估都是使用极大似然估计，然而极大似然估计存在的问题没法很好的处理数据的稀疏性以及曝光的偏差。文中提出了一个多因素的生成对抗网络(MFGAN)，包括一个基于Transformer的生成器，来预测下一次点击；以及多个鉴别器，从不同的因素来评估生成器的推荐。

2021-07-04 21:10:14 650

原创 KDD2020|基于邻域的异构图交互推荐模型

An Efficient Neighborhood-based Interaction Model for Recommendation on Heterogeneous Graph

2021-06-23 21:20:32 627 2

原创 SIGIR2020|基于物品频次信息的个性化建模对下个“篮子”预测

Modeling Personalized Item Frequency Information for Next-basket Recommendation

2021-06-17 16:33:11 501

原创 SIGIR2020|个性化和可解释的替代商品推荐

Try This Instead: Personalized and Interpretable Substitute Recommendationhttps://arxiv.org/pdf/2005.09344.pdf

2021-06-15 22:34:47 512 1

原创 KDD2020|基于CTR预估的自动神经网络结构搜索

Towards Automated Neural Interaction Discovery for Click-Through Rate Prediction

2021-06-14 21:14:03 323

原创 KDD2020|序列推荐中的解纠缠自监督学习

Disentangled Self-Supervision in Sequential Recommenders背景

2021-06-13 16:59:27 1101

原创 KDD2020|对大规模推荐模型输入的神经搜索

Neural Input Search for Large Scale Recommendation Modelshttps://arxiv.org/pdf/1907.04471.pdf这是一篇谷歌推荐的短文，

2021-06-10 20:00:08 306

原创 KDD2020|PinnerSage:Pinterest推荐中的多模式用户嵌入框架

PinnerSage: Multi-Modal User Embedding Framework for Recommendations at Pinterest

2021-06-09 21:38:00 851

原创 KDD2020|淘宝推荐中的“特权”特征蒸馏

Privileged Features Distillation at Taobao Recommendationshttps://arxiv.org/pdf/1907.05171.pdfRef:Xu C, Li Q, Ge J, et al. Privileged Features Distillation at Taobao Recommendations[C]//Proceedings of the 26th ACM SIGKDD International Conference on .

2021-06-09 14:04:30 1111

原创 KDD2020|推荐系统中基于属性的无偏学习：算法和案例研究

Attribute-based Propensity for Unbiased Learning in Recommender Systems: Algorithm and Case Studies

2021-06-08 10:44:21 818

原创 KDD2020|可控的多兴趣推荐框架

Controllable Multi-Interest Framework for Recommendationhttps://arxiv.org/pdf/2005.09347.pdfRef:Cen Y, Zhang J, Zou X, et al. Controllable Multi-Interest Framework for Recommendation[C]//Proceedings of the 26th ACM SIGKDD International Conference on.

2021-06-07 14:46:12 652