【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第1张图片
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第2张图片
RSKT-2014
International conference on rough sets and knowledge technology


文章目录

  • 1 Background and Motivation
  • 2 Review of Convolutional Neural Networks
  • 3 Advantages / Contributions
  • 4 Method
  • 5 Experiments
    • 5.1 Datasets
    • 5.2 Experimental Results
  • 6 Conclusion(own) / Future work


1 Background and Motivation

池化层的作用(一文看尽深度学习中的9种池化方法!)

  • 增大网络感受野
  • 抑制噪声,降低信息冗余
  • 降低模型计算量,降低网络优化难度,防止网络过拟合
  • 使模型对输入图像中的特征位置变化更加鲁棒

作者针对 max 和 ave pooling 的缺点,

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第3张图片

提出了 mix pooling——randomly employs the local max pooling and average pooling methods when training CNNs

2 Review of Convolutional Neural Networks

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第4张图片

  • Convolutional Layer,包括卷积操作和 activation function
  • Non-linear Transformation Layer,也即 normalization 层,现在比较流行的是 BN 等,以前的是 LCN(local contrast normalization) 和 AlexNet 的 LRN(the local response normalization) 等,PS:论文中 LCN 的公式感觉有问题,LRN 细节也原版论文也有差距,形式基本一致
  • Feature Pooling Layer

3 Advantages / Contributions

借鉴 dropout, 混合max 和 ave 池化,提出 mixed pooling

4 Method

1)mixed pooling 公式
在这里插入图片描述

λ \lambda λ is a random value being either 0 or 1

2)mixed pooling 反向传播

先看看 max 和 ave pooling 的反向传播

max pooling
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第5张图片
(图片来源于网络,侵删!!!)

ave pooling
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第6张图片
(图片来源于网络,侵删!!!)

mixed pooling

得记录下 λ \lambda λ 的取值,才能正确反向传播

the pooling history about the random value λ \lambda λ in Eq. must be recorded during forward propagation.

3)Pooling at Test Time

统计训练时某次 pooling 采用 max 和 ave 的频次 F m a x k F_{max}^{k} Fmaxk F a v e k F_{ave}^{k} Favek,谁的频次高测试的时候该处的 pooling 就用谁,开始玄学了是吧,哈哈哈哈

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第7张图片

5 Experiments

5.1 Datasets

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第8张图片

  • CIFAR-10
  • CIFAR-100
  • SVHN

5.2 Experimental Results

1)CIFAR-10
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第9张图片

train error 高,acc 高

作者解释 This indicates that the proposed mixed pooling outperforms max pooling and average pooling to address the over-fitting problem

可视化结果
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第10张图片
可以看出 mixed pooling 包含更多的信息

2)CIFAR-100
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第11张图片
3)SVHN
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第12张图片
4)Time Performance
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第13张图片

6 Conclusion(own) / Future work

LRN

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第14张图片
k , n , α , β k, n, \alpha, \beta k,n,α,β 都是超参数, a , b a,b a,b 输入输出特征图, x , y x,y x,y 空间位置, i i i 通道位置

以下内容来自 深度学习的局部响应归一化LRN(Local Response Normalization)理解

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第15张图片

import tensorflow as tf
import numpy as np
x = np.array([i for i in range(1,33)]).reshape([2,2,2,4])
y = tf.nn.lrn(input=x,depth_radius=2,bias=0,alpha=1,beta=1)
 
with tf.Session() as sess:
    print(x)
    print('#############')
    print(y.eval())

【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第16张图片
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第17张图片


LCN
【Mixed Pooling】《Mixed Pooling for Convolutional Neural Networks》_第18张图片
《What is the best multi-stage architecture for object recognition?》

你可能感兴趣的:(CNN,深度学习,cnn,机器学习,mixed,pooling)