HEU机器学习基础0.03-CSDN博客

本文链接：https://blog.csdn.net/Scales_0704/article/details/131726958

HEU机器学习基础0.03

线性回归Extra

文章目录

HEU机器学习基础0.03
- 前言
多元线性回归multiple linear regression

前言

Preview
based on Andrew Ng

多元线性回归multiple linear regression

有多个输入特征的线性回归模型
$x_j$ = $j^{th}$ feature
n = number of features
$\vec{x}^{i}$ = features of $i^{th}$ training example
$x_j$ = value of feature j in $i^{th}$ training example

初始模型single linear regression model

$f_{w,b}(x) = wx+b$

multiple linear regression

$f_{w,b}(x) = w_1x_1+w_2x_2+...+w_nx_n+b$

向量化vectorization

基本步骤

$\vec{w}$ = [ $w_1$ $w_2$ $w_3$ … $w_n$ ]
b = constant
$\vec{x}$ = [ $x_1$ $x_2$ $x_3$ … $x_n$ ]

减少变量数量，加快运行速度

Final result

$f_{w,b}(\vec{x}) = \vec{w}\cdot\vec{x}+b$

	Previous notation	Vector notation
Parameter	$w_1$ … $w_n$	$\vec{w}$ = [ $w_1$ … $w_n$ ]
Model	$f_{w,b}(x) = w_1x_1+w_2x_2+...+w_nx_n+b$	$f_{w,b}(\vec{x}) = \vec{w}\cdot\vec{x}+b$
Cost function	$J(w_1, ... ,w_n, b)$	$J(\vec{w},b)$
Gradient descent	repeat{ $w_j = w_j - \alpha\frac{\partial}{\partial w_j}\partial J(w_1,w_2,...w_n,b)$ $\alpha\frac{\partial}{\partial b}\partial J(w_1,w_2,...w_n,b)$ }	repeat{ $w_j = w_j - \alpha\frac{\partial}{\partial w_j}\partial J(\vec w,b)$ $\alpha\frac{\partial}{\partial b}\partial J(\vec w, b)$ }