II
6.2 假说表示 .................................................................................................................. 97
6.3 判定边界 .................................................................................................................. 99
6.4 代价函数 ................................................................................................................ 101
6.5 简化的成本函数和梯度下降 ................................................................................ 104
6.6 高级优化 ................................................................................................................ 107
6.7 多类别分类:一对多 ............................................................................................ 111
七、正则化(Regularization) ................................................................................................. 114
7.1 过拟合的问题 ........................................................................................................ 114
7.2 代价函数 ................................................................................................................ 116
7.3 正则化线性回归 .................................................................................................... 118
7.4 正则化的逻辑回归模型 ........................................................................................ 119
第 4 周 .......................................................................................................................................... 120
第八、神经网络:表述(Neural Networks: Representation)............................................... 120
8.1 非线性假设 ............................................................................................................ 120
8.2 神经元和大脑 ........................................................................................................ 122
8.3 模型表示 1 ............................................................................................................. 126
8.4 模型表示 2 ............................................................................................................. 130
8.5 特征和直观理解 1 ................................................................................................. 132
8.6 样本和直观理解 II ................................................................................................. 134
8.7 多类分类 ................................................................................................................ 136
第 5 周 .......................................................................................................................................... 137
九、神经网络的学习(Neural Networks: Learning) ............................................................. 137
9.1 代价函数 ................................................................................................................ 137
9.2 反向传播算法 ........................................................................................................ 139
9.3 反向传播算法的直观理解 .................................................................................... 142
9.4 实现注意:展开参数 ............................................................................................ 145
9.5 梯度检验 ................................................................................................................ 146
9.6 随机初始化 ............................................................................................................ 148
9.7 综合起来 ................................................................................................................ 149
9.8 自主驾驶 ................................................................................................................ 150
第 6 周 .......................................................................................................................................... 153
十、应用机器学习的建议(Advice for Applying Machine Learning) ................................... 153
10.1 决定下一步做什么 .............................................................................................. 153
10.2 评估一个假设 ...................................................................................................... 156
10.3 模型选择和交叉验证集 ...................................................................................... 158
10.4 诊断偏差和方差 .................................................................................................. 160
10.5 归一化和偏差/方差 ............................................................................................ 162
10.6 学习曲线 .............................................................................................................. 164
10.7 决定下一步做什么 .............................................................................................. 166
十一、机器学习系统的设计(Machine Learning System Design) ....................................... 168
11.1 首先要做什么 ...................................................................................................... 168
11.2 误差分析 .............................................................................................................. 169
11.3 类偏斜的误差度量 .............................................................................................. 172
11.4 查全率和查准率之间的权衡 .............................................................................. 173
- 1
- 2
前往页