search

RaleighZ / statnlp_fundamental_reading

Group for Fundamental NLP Reading and Learning
5 stars 0 forks source link

Details of pooling in CNN for NLP #12

Open RaleighZ opened 5 years ago

RaleighZ commented 5 years ago
  1. Dynamic pooling
  2. K-max pooling, retain the relative position information?
nanguoshun commented 5 years ago

The answer for 2 is YES. K-max pooling RETAINs the order of features but it insensitive to their specific positions. WHY k-max pooling: K-max pooling makes it possible to pool the k most active features, and it is is the generalization of max pooling.

RaleighZ commented 5 years ago

The answer for 2 is YES. K-max pooling RETAINs the order of features but it insensitive to their specific positions. WHY k-max pooling: K-max pooling makes it possible to pool the k most active features, and it is is the generalization of max pooling.

It would be better with the reference for K-max pooling RETAINs the order of features