wziji / deep_ctr

212 stars 77 forks source link

点击预估模型

1. Recall

算法 论文 公众号或知乎文章介绍
Word2vec Efficient Estimation of Word Representations in Vector Space
YouTubeNet Deep Neural Networks for YouTube Recommendations 推荐系统召回模型之YouTubeNet
DSSM Sampling-Bias-Corrected Neural Modeling for Large Corpus Item Recommendations 实践DSSM召回模型
MIND Multi-Interest Network with Dynamic Routing for Recommendation at Tmall 推荐系统召回模型之MIND用户多兴趣网络实践

2. Rank

算法 论文 公众号文章介绍
FFM Field-aware Factorization Machines for CTR Prediction FFM算法原理及Bi-FFM算法实现
Wide & Deep Wide & Deep Learning for Recommender Systems
NFM Neural Factorization Machines for Sparse Predictive Analytics NFM模型理论与实践
AFM Attentional Factorization Machines: Learning the Weight of Feature Interactions via Attention Networks 注意力机制在深度推荐算法中的应用之AFM模型
DeepFM DeepFM: A Factorization-Machine based Neural Network for CTR Prediction DeepFM实践
BST Behavior sequence transformer for e-commerce recommendation in Alibaba Transformer 在美团搜索排序中的实践

3. Multi-Task

算法 论文 公众号文章介绍
ESMM Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate ESMM多任务学习算法在推荐系统中的应用
MMoE Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts 多任务学习之MMOE模型

4. Recall_ANN

算法 开源地址 公众号文章介绍
Annoy https://github.com/spotify/annoy Annoy最近邻检索技术之 “图片检索”
Faiss https://github.com/facebookresearch/faiss

代码参考

https://github.com/shenweichen/DeepCTR

https://github.com/shenweichen/DeepMatch

待学习及分享

Recall

Poly-encoders: Transformer Architectures and Pre-training Strategies for Fast and Accurate Multi-sentence Scoring

Controllable Multi-Interest Framework for Recommendation, 代码:https://github.com/THUDM/ComiRec

Pre-Rank

COLD: Towards the Next Generation of Pre-Ranking System

Rank

DIN:Deep Interest Network for Click-Through Rate Prediction

DIEN:Deep Interest Evolution Network for Click-Through Rate Prediction, 代码: https://github.com/mouna99/dien

MIMN:Practice on Long Sequential User Behavior Modeling for Click-Through Rate Prediction

Search-based Interest Model:Search-based User Interest Modeling with Lifelong Sequential Behavior Data for Click-Through Rate Prediction

Multi-Task

YouTube,2019: Recommending What Video to Watch Next-A Multitask Ranking System