Skip to content

Commit

Permalink
xdeepfm
Browse files Browse the repository at this point in the history
  • Loading branch information
shataowei committed Dec 1, 2019
1 parent c51dc08 commit 9429fe8
Show file tree
Hide file tree
Showing 3 changed files with 27 additions and 17 deletions.
30 changes: 15 additions & 15 deletions .idea/workspace.xml

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

3 changes: 2 additions & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -365,7 +365,8 @@
- MLR
- Neural Network全家桶
- XDeepFM
- [整个过程中有什么亮点?有哪些决定性的提升](推荐/XDeepFM.md#L164)
- [选用的原因](推荐/XDeepFM.md#L164)
- [什么叫显示隐式?什么叫元素级/向量级?什么叫做高阶/低阶特征交互](推荐/XDeepFM.md#L164)
# 风控
- 孤立森林
- 评分卡
Expand Down
11 changes: 10 additions & 1 deletion 推荐/XDeepFM.md
Original file line number Diff line number Diff line change
@@ -1 +1,10 @@
#
# 选用的原因?
- 类似deepfm和FNN等模型的高阶的特征交互来自于dnn部分,但是这样的特征交互是不可控且隐式的,难以描述的
- 向量级别的特征交互而不是元素级交互
- 经验上,vector-wise的方式构建的特征交叉关系比bit-wise的方式更容易学习
- 我也不知道具体好在哪,如果有大佬会可以指导一下,感恩

# 什么叫显示隐式?什么叫元素级/向量级?什么叫做高阶/低阶特征交互?
- 显示是可以写出feature交互的公式,隐式相反
- 元素级是以feature值交互,向量级是feature向量级点乘处理
- 高阶特征是类似DNN这种多层特征交互,低阶特征交互是FM这种特征单层处理方式

0 comments on commit 9429fe8

Please sign in to comment.