w task 2

Posted 2023-06-02 ylxn

Method:

• 2 sentences
• introduce the topic
• give a general answer
• 30-40 words.

 

Four types of question:

• Discussion
• Opinion
• Problem and solution
• two-part question

 

Discussion:

Plan: Topic- A or B

Answer： sometimes A but usually better B

People have different views about A opposed to B

While there are some benefits to ... , I believe that B is better

 

Opinon:

my own opinion

Plan: Topic - community service for all teenagers.

Answer - 3 choices: agree, disagree, or balanced opinion

It is sometimes argued that .... 

(agree) I completely agree that ... is a good idea

(disagree) In my opinion, it would be wrong to force .. 

(balanced) while i disagree with the idea A , I do believe that A could benefit ...

 

Problem and solution

Plan:

Topic

Answer-serveral reasons, a variety of measures

It is true that ... deter sb from doint sth

There are several reasons why ..., but ... could certainly take stpes to adress this issue. 

 

Two part question

Topic: - 

Answer- several factors, unrealistic/impossible

 

 

break the law

unsalaried work

job satisfaction

feel fufilled professionally

 

谢谢！

Datewhale一起吃瓜 Task2啃瓜第三章

文章目录

• 线性模型
• 关键：找到合适的w和b
• 如何找到合适的 w和b？
• 偏导为什么可以？
• 推广

线性模型

任务：找出一条线能够对数据进行划分或预测趋势

关键：找到合适的w和b

更适合于连续性的数值，如果数据是离散的如色泽是青绿色，我们需要对特征值进行编码
首先我们考虑是否是有序的
若有序，如身高的高中低，我们可以编为1，0.5，0
若无序，如一共有三种色泽，某种色泽是青绿色，我们可以标为[0 1 0]

如何找到合适的 w和b？

均方误差最小化，利用最小二乘法找到一组合适的w和b使得均方误差最小

偏导为什么可以？

在我们求最佳参数的过程中，参数在不断变化，我们对其进行偏导，并让其等于0，这时候参数不再变化，他可能是极大值，也可能是极小值。而均方误差最大值是无穷大，因此极大值是不存在的，求得的结果只可能是在极小值上，因此我们求出的结果，即是均方误差最小的时候，就是我们想要的最佳参数

推广

根据具体情况可能有以下两种结果：

若此时根据归纳偏好，或者引入正则化，选出最佳的参数