ImageNet Classification with Deep Convolutional Neural Networks 阅读理解及问题
来源:互联网 发布:斑马梦龙网络计划 编辑:程序博客网 时间:2024/04/30 07:12
看了这篇论文和网上的阅读笔记,还没看代码。有一些问题列在这里,看看自己以后能不能回答。
1. 用GPU加速训练,论文提到是用GPU做2D卷积,而实际的网络都是3D的卷积计算,这个怎么对应?怎样计算3D卷积?
2. 论文中的深度CNNs使用激活函数f(x) = max(0, x), 称为ReLU Nonlinearity。这比sigmoid、双曲正切作为激活函数在训练速度上要快好几倍。那么ReLU在哪些模型上比较适用?是否能说在深度CNNs的训练上都可以用ReLU代替sigmoid? 毕竟在训练速度上有很大的优势。
3. CNN结构图如下
第4、5个卷积层和第3个卷积层的特征图大小相同(13X13),论文说它们之间没有正则化和池化运算,这么说只做了卷积运算?卷积运算不改变特征图大小?如果是这样那么第一层的大小55X55怎么算的,怎么不是56X56(224/4=56)?
4. 在防止过拟合所采用的方法中,altering the intensities of the RGB channels这一步没看明白,这里对哪个对象做PCA?是每个像素、每张图还是整个训练集?
0 0
- ImageNet Classification with Deep Convolutional Neural Networks 阅读理解及问题
- ImageNet Classification with Deep Convolutional Neural Networks 阅读笔记
- ImageNet Classification with Deep Convolutional Neural Networks(阅读)
- ImageNet Classification with Deep Convolutional Neural Networks阅读笔记
- ImageNet Classification with Deep Convolutional Neural Networks 阅读笔记
- ImageNet Classification with Deep Convolutional Neural Networks AlexNet阅读笔记
- 《ImageNet Classification with Deep Convolutional Neural Networks》论文阅读
- ImageNet Classification with deep convolutional neural networks
- ImageNet Classification with Deep Convolutional neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with deep convolutional neural networks
- Imagenet classification with deep convolutional neural networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- ImageNet Classification with Deep Convolutional Neural Networks
- Linux命令-系统相关
- MFC学习笔记之VC++ Debug Assertion Failed! 错误(File: strex.cpp Line: 690)
- Search in Rotated Sorted Array -- leetcode
- struts2+jquery+json集成
- Linux命令-磁盘文件相关
- ImageNet Classification with Deep Convolutional Neural Networks 阅读理解及问题
- 构造器与静态工厂
- linux命令-文件权限相关
- 指针数组与数组指针的区别
- Navicat for Oracle 安装后连接配置问题(ORA-12541或者ORA-28547等)
- inux命令-用户相关
- Java Frame Panel JFrame JPanel
- intel万兆网卡驱动简要分析
- Linux命令-包相关