DeepCTR’s Documentation

DeepCTR is a project that introduces classic CTR (Click Through Rate) prediction model and implements popular network designed for CTR prediction task. What’s more, we do a great number of experiments on open data and provide as benchmark.

Installation

• Dependency

Models

• Introduction
• Traditional Models
  • FM (Factorization Machines)
  • FFM (Field-aware Factorization Machines)
  • GBDT+LR
• Deep Models
  • Overview
  • FNN (Factorization-supported Neural Network)
  • PNN (Product-based Neural Network)
  • Wide & Deep
  • DeepFM
  • NFM (Neural FM)
  • AFM (Attentional FM)
  • DCN (Deep & Cross Network)
  • DIN (Deep Interest Network)
  • xDeepFM (eXtreme DeepFM)

Benchmarks

• Benchmarks

Contributors

• Contributors