Name	Name	Last commit message	Last commit date
parent directory ..
README.md	README.md
main.py	main.py

Name

Last commit message

Last commit date

Practice 15 - Batch Normalization in TensorFlow

This practice is refer to the following resources credited to Morvan.

莫烦PYTHON - TensorFlow: 什么是批标准化 (Batch Normalization)
莫烦PYTHON - TensorFlow: Batch Normalization 批标准化

Execution

Run main.py with a argument

# Make sure your current directory is in this folder
# Train with "ReLU" as the activation function
$ python3 main.py relu
# Train with "tanh" as the activation function
$ python3 main.py tanh

If succeed, you will get the following result

# Train with "ReLU" as the activation function
[INFO] Using ReLU as activation function
# Train with "tanh" as the activation function
[INFO] Using tanh as activation function

Besides, you will see the following figures
- Training with tf.nn.relu as the activation function
  - Input data
  - Histogram
  - Loss rate
- Training with tf.nn.tanh as the activation function
  - Input data
  - Histogram
  - Loss rate