Cyan's Blog

Search

Search IconIcon to open search

D2L-44-Batch_Normalization-批量归一化

Last updated Mar 5, 2022 Edit Source

# Batch Normalization

2022-03-05

Tags: #BatchNormalization #Normalization #DeepLearning #Regularization

# Motivation

# Batch Normalization Layer

BN层可以作用在全连接层或者卷积层的输出之前和输入之后, 并且在激活函数之前.

# 全连接层

# 卷积层

Batchnorm

# 规范化定义

# 其他相似的Normalization

Stanford CS231n BN

# 预测与训练的不同


  1. 只是为了解释的清楚, 实际计算的时候并不是我说的这样的 ↩︎

  2. 这个图居然是Drawio画的, 可惜原作者的链接没了 calculation of mean and variance in batch normalization in convolutional neural network - Stack Overflow  ↩︎

  3. 7.5. Batch Normalization — Dive into Deep Learning 0.17.2 documentation ↩︎