[ 机器学习 - 吴恩达 ] 单变量线性回归 | 2-2 损失函数 1


房价训练集(俄勒冈州,波特兰市)

Size in \(feet^2\) (x) Price ($) in 1000's (y)
2104 460
1416 232
1534 315
852 178

假设函数: \(h_\theta(x) = \theta_0 + \theta_1x\)

参数:\(\theta_{i's}\)

如何选择 \(\theta_{i's}\)?
想法:对于我们的训练样例 \((x, y)\),选择\(\theta_0, \theta_1\),使\(h_\theta(x)\)靠近\(y\)

Cost Function 损失函数

\[J(\theta_0,\theta_1) = 1/2m\sum_{i=1}^m(h_\theta(x^{(i)}) - y^{(i)})^2 \]

m:训练样本数

总结

假设函数

\[h_\theta(x) = \theta_0 + \theta_1x \]

参数: \(\theta_0, \theta_1\)
损失函数: 均方误差函数(Squared error function)**

\[J(\theta_0,\theta_1) = 1/2m\sum_{i=1}^m(h_\theta(x^{(i)}) - y^{(i)})^2 \]

目标: \(\begin{matrix} minimize J(\theta_0,\theta_1)\\ \theta_0,\theta_1 \end{matrix}\)

简化:

假设函数:

\[h_\theta(x) = \theta_1x \]

参数: \(\theta_1\)
损失函数:

\[J(\theta_1) = 1/2m\sum_{i=1}^m(h_\theta(x^{(i)}) - y^{(i)})^2 \]

目标: \(\begin{matrix} minimize J(\theta_1)\\ \theta_1 \end{matrix}\)

相关