上一讲我们得到 learning can happen with target distribution $P(y|x)$ and low $E_{in}$ w.r.t. error measure.

1. Linear Regression Problem 

  linear regression hypothesis: $h(x) = w^T x$, $h(x)$: like perceptron, but without the $sign$.

  linear regression: find lines/hyperplanes with small residuals.

  The Error Measure:

  popular/historical error measure: squared error $err(\hat{y}; y) = (\hat{y} - y)^2$.

2. Linear Regression Algorithm

practical suggestion: use well-implemented $y$ routine instead of $(X^T X)^{-1} X^T$ for numerical stability when almost-singular.

Linear Regression Algorithm:

3. Generalization Issue

4. Linear Regression for Binary Classification