first week of machine learning on Coursera
来源:互联网 发布:centos修改ip地址 编辑:程序博客网 时间:2024/06/18 16:59
first week of machine learning on Coursera
@(Coursera)
惯例是,先在Matlab/octava上实现算法原型,确定可用再迁移到其他编译环境。因为Matlab/octava集成了很多机器学习算法和常用的计算,对于算法实现速度很快,而且代码比较简单。
平方误差函数是解决回归问题最常用的代价函数(cost function)。
我们的目的是使我们作出的假设函数hypothesis function最接近于实际的训练集样本点集
所以我们的目的就是找到一组
我们使用梯度下降法来寻找
梯度下降法的直观描述就是,当人在山顶,每次迈出一步长
这里的
temp0:
temp1:
通过梯度下降不断的更新
步长太大可能会导致无法收敛:
![Alt text](./屏幕快照 2017-09-23 下午7.14.33.png)
线性模型时:
假设函数
成本函数cost function:
这里为什么乘以
此时,
Batch:表示步长,也称为学习速率,就是上式中的
Vector:an N*1 matrix
矩阵计算:
单位矩阵(Identity matrix):对角线元素为1,其余元素为0的方阵。
矩阵乘法:
除非:
矩阵的逆,当矩阵是个方阵时,
- first week of machine learning on Coursera
- Second week of machine learning on Coursera
- Third week of machine learning on Coursera
- Sixth week of machine learning on Coursera
- Seventh week of machine learning on Coursera
- Fourth and fifth week of machine learning on Coursera
- Machine Learning - Andrew Ng on Coursera (Week 1)
- Machine Learning - Andrew Ng on Coursera (Week 2)
- Machine Learning - Andrew Ng on Coursera (Week 3)
- Machine Learning - Andrew Ng on Coursera (Week 4)
- Machine Learning - Andrew Ng on Coursera (Week 5)
- Machine Learning - Andrew Ng on Coursera (Week 6)
- Outline of Machine Learning created by Andrew Ng on Coursera
- Coursera Machine Learning Week 1.1: Introduction
- Coursera Machine Learning Week 3.1: Logistic Regression
- Coursera Machine Learning Week 3.2: Regularization
- Coursera Machine Learning Week 8.1: Clustering
- coursera Machine Learning Week 1学习笔记
- 资源链接
- heap和stack有什么区别
- 随笔
- final, finally, finalize的区别。
- 哈希表
- first week of machine learning on Coursera
- 理论篇~第三章 数据模型设计
- Redis在Java中的Jedis常用API
- [读书笔记]第八章 导游拿商店的回扣愈多,团员看风景的机会愈少
- Centos 6.4系统如何从动态获取ip改为静态获取ip
- 三星港版 A8000 6.0.1 ROOT
- MYSQL EXPLAIN 查看查询优化器-后台开发应知应会
- javaWeb开发入门
- spring boot 整合spring data jpa