神经网络与深度学习讲义20151211

《神经网络与深度学习讲义20151211》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、《神经网络与深度学习》讲义NotesonArtificialNeuralNetworksandDeepLearning邱锡鹏xpqiu@fudan.edu.cn2015年12月11日DRAFT2015-12-1122:00编译时间：目录第一章绪论11.1总结和深入阅读.................................2第二章数学基础32.1向量.......................................32.1.1向量的模................................32.1.2向量的范数.........

2、......................32.2矩阵.......................................42.3矩阵的基本运算.................................42015-12-1122:002.3.1常见的矩阵...............................52.3.2矩阵的范数DRAFT...............................52.4导数.......................................6编译时间：2.4.1常见的向量导数...

3、.........................62.4.2导数法则................................62.5常用函数....................................72.5.1logistic函数...............................72.5.2softmax函数..............................72.6总结和深入阅读.................................9第三章机器学习概述103.1机器学习概述.............

4、.....................10II目录3.1.1损失函数................................123.1.2机器学习算法的类型..........................133.1.3机器学习中的一些概念.........................143.1.4参数学习算法..............................163.2线性回归....................................183.3线性分类................................

5、....203.3.1两类分类................................203.3.2多类线性分类..............................223.4评价方法....................................263.5总结和深入阅读.................................27第四章感知器284.1两类感知器...................................294.1.1感知器学习算法............................294.1.

6、2收敛性证明...............................304.2多类感知器...................................324.2.1多类感知器的收敛性..........................2015-12-1122:00344.3投票感知器...................................354.4总结和深入阅读DRAFT.................................36第五章人工神经网络编译时间：385.1神经元.......................

7、...............395.1.1激活函数................................395.2前馈神经网络..................................415.2.1前馈计算................................415.3反向传播算法..................................425.4梯度消失问题..................................455.5训练方法....................................

8、465.6经验..........................