Software Courses/Improving Deep Neural Networks

[Improving: Hyper-parameter tuning, Regularization and Optimization] Optimization - Adam

김 정 환 2020. 4. 17. 11:42
반응형

This note is based on Coursera course by Andrew ng.

(It is just study note for me. It could be copied or awkward sometimes for sentence anything, because i am not native. But, i want to learn Deep Learning on English. So, everything will be bettter and better :))

 

 

 

INTRO

Adam stands for Adaptive Moment Estimation. 

The Adam optimization algorithm is basically taking Momentum and RMSprop and putting them together. 

 

MAIN

 

The algorithm has a number of hyper parameters. 

  • α : needs to be tune
  • β1 : default is 0.9      (this is computing the moving weighted average of dw and db)
  • β2 : default is 0.999  (this is computing the moving weighted average of dw^2 and db^2)
  • ε : default 10^-8

 

CONCLUSION

 

 

 

 

 

 

 

 

반응형