这一讲主要介绍当数据线性不可分时，通过非线性映射将数据映射到高维特征空间，使得数据在高维特征空间是线性可分的，进而实现机器学习。

1. Quadratic Hypotheses

2. Nonlinear Transform

The Nonlinear Transform Steps：

Nonlinear Model via Nonlinear $\Phi$  + Linear Models

3. Price of Nonlinear Transform

Q-th order polynomial transform:

4. Structured Hypothesis Sets

Structured Hypothesis Sets

linear model first: simple, efficient, safe, and workable!