0

I am in the early stage of studying the Neural Network. Here are the list I made during the online classes.

  1. Shuffle the data
  2. Normalized the data by Sk-learn: StandardScalar
  3. Initialized the weight with He initialization
  4. Use adaptive learning rate
  5. L2, L1 optimization
  6. Train in mini-batch size
  7. Change cost function

They are 7 cases I have to do when I need to find tune the accuracy. Are they any topic left behind? Or some of them is redundant to do?

Do you normally follow all of these? Or not do if the accuracy is high enough?

Sarit
  • 101
  • 2

1 Answers1

0

There is no fixed list of steps to do in order to tune a model. It's a more complex work and depends of multiple factors like your sources of data, your hardware resources, etc. I recommend you take the first three courses of Andrew Ng's Deep Learning specialization on Coursera where you can achieve a a wider view about how to tune a NN and structure a NN project.

Alberto Burgos Plaza
  • 1
  • 1