logo头像
Snippet 博客主题

论文索引

阅读算法的原始论文是学习它的最好途径,为什么呢?一是网上的博客资料质量良莠不齐,有时可能是错的;二是论文中一定会重点阐述算法的核心创新点到底是怎么来的,以及这个算法的完整上下文。


集成学习

  1. CatBoost: gradient boosting with categorical features support:catboost的全面介绍
  2. CatBoost: unbiased boosting with categorical features:详细论述如何处理类别特征以及标签泄露问题

图算法

社区发现

  1. Community structure in social and biological networks: GN算法,社区发现领域的第一篇论文
  2. Finding and evaluating community structure in networks: 详细介绍模块度函数(modurity), IV部分
  3. Fast unfolding of communities in large networks: state-of-art的算法Louvain
  4. Near linear time algorithm to detect community structures in large-scale networks
    Towards Real-Time Community Detection in Large Networks
    : 标签传播算法
  5. Uncovering the overlapping community structure of complex networks in nature and society: 第一篇重叠社区发现算法
  6. Detecting the overlapping and hierarchical community structure in complex networks: 基于局部扩展的算法

图嵌入

  1. DeepWalk: Online Learning of Social Representations
  2. LINE: Large-scale Information Network Embedding
  3. node2vec: Scalable Feature Learning for Networks

图神经网络

  1. Semi-Supervised Classification With Graph Convolution Networks :半监督
  2. Graph Attention Networks
  3. Inductive Representation Learning on Large Graphs :Graphsage
  4. Graph Convolutional Neural Networks for Web-ScaleRecommender Systems :Pinsage

风控算法

  1. Isolation Forest
  2. Isolation-based Anomaly Detection
  3. Outlier Detection with Autoencoder Ensembles
  4. CopyCatch: Stopping Group Attacks by Spotting LockstepBehavior in Social Networks : 基于用户行为和时间的双聚类算法
  5. Uncovering Large Groups of Active Malicious Accounts inOnline Social Networks :Synchrotrap
  6. FRAUDAR: Bounding Graph Fraud in the Face ofCamouflage

计算广告

点击率优化

  1. Practical Lessons from Predicting Clicks on Ads at Facebook: GBDT+LR
  2. Wide & Deep Learning for Recommender Systems: lr memorization & deep generalization
  3. Factorization Machines: fm特征组合
  4. DeepFM: A Factorization-Machine based Neural Network for CTR Prediction: fm替换人工特征组合
  5. Field-aware Factorization Machines for CTR Prediction
  6. Deep Interest Network for Click-Through Rate Prediction: 阿里巴巴din

评论系统未开启,无法评论!