Skip to content

Deep neural network model based on the tree

Notifications You must be signed in to change notification settings

zhuhan1236/Tree_Deep_Model

 
 

Repository files navigation

深度树模型实验


目录

  • 实验环境
  • 代码结构
  • 算法模型
  • 进度
  • 参考文献

实验环境

系统环境: ubuntu 18.04LTS
IDE:PyCharm 2018.3

实验数据集

代码结构

code-structure
文件说明
tdm.py: 代码入口,负责完整深度树模型的训练和测试
sample_init.py: 数据处理及生成程序,负责数据预处理及树样本的生成
construct_tree.py: 样本二叉树生成程序,负责树模型的生成
deep_network.py: DNN的实现程序,负责网络的搭建
prediction.py: 树节点预测及模型评测程序,负责模型预测及性能验证

算法模型

algorithm-structure
深度树算法流程(文献[1]):

  1. 构造随机二叉树
  2. 基于树模型生成样本
  3. 训练DNN模型直到收敛
  4. 基于DNN模型得到样本的Embedding,重新构造聚类二叉树
  5. 循环上述2~4过程

进度

完成功能测试,跑通模型
TODO: 改进性能,验证模型

参考文献

[1] Learning Tree-based Deep Model for Recommender Systems, Han Zhu, Xiang Li, Pengye Zhang, etc.
[2] Deep Interest Network for Click-Through Rate Prediction, Guorui Zhou, Chengru Song, Xiaoqiang Zhu, etc.
[3] Empirical Evaluation of Rectified Activations in Convolution Network, Bing Xu, Naiyan Wang, Tianqi Chen, etc.
[4] Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification, Kaiming He, Xiangyu Zhang, Shaoqing Ren, etc.
[5] Distributed Representations of Words and Phrases and their Compositionality, Tomas Mikolov, Ilya Sutskever, Kai Chen, etc.

About

Deep neural network model based on the tree

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%