当前位置:网站首页>ML笔记-matrix fundamental, Gradient Descent
ML笔记-matrix fundamental, Gradient Descent
2022-06-22 13:53:00 【阿达斯加】
matrix fundamental
basic concepts:



错题:


matrix calculation:





Gradient Descent:





We can speed up gradient descent by having each of our input values in roughly the same range. This is because θ will descend quickly on small ranges and slowly on large ranges, and so will oscillate inefficiently down to the optimum when the variables are very uneven.
The way to prevent this is to modify the ranges of our input variables so that they are all roughly the same. Ideally:
−1 ≤ x(i) ≤ 1
or
−0.5 ≤ x(i) ≤ 0.5
Two techniques to help with this are feature scaling and mean normalization. Feature scaling involves dividing the input values by the range (i.e. the maximum value minus the minimum value) of the input variable, resulting in a new range of just 1. Mean normalization involves subtracting the average value for an input variable from the values for that input variable resulting in a new average value for the input variable of just zero. To implement both of these techniques, adjust your input values as shown in this formula:
Where μ_iμi is the average of all the values for feature (i) and s_isi is the range of values (max - min), or s_isi is the standard deviation.
边栏推荐
- RealNetworks vs. Microsoft: the battle in the early streaming media industry
- [Software Engineering] design module
- Mysql学习笔记2022
- 轻松上手Fluentd,结合 Rainbond 插件市场,日志收集更快捷
- 谷歌竞价账户可以探测到全球市场吗?
- 2022orace database installation and use
- 拜登签署两项旨在加强政府网络安全的新法案
- At 19:00 this Thursday evening, the 7th live broadcast of battle code Pioneer - how third-party application developers contribute to open source
- 【浙江大学】考研初试复试资料分享
- flutter video_ Player monitors and automatically plays the next song
猜你喜欢

网站存在的价值是什么?为什么要搭建独立站

Fluentd is easy to get started. Combined with the rainbow plug-in market, log collection is faster
![[PR] basic process](/img/e6/13a5703cfb8e5c51ed06fe9b0d9485.png)
[PR] basic process

visual studio开发过程中常见操作

Common real machine debugging plug-ins for unity commercial games

Database connection pool: implementation of connection pool function point

NF-ResNet:去掉BN归一化,值得细读的网络信号分析 | ICLR 2021

Zhongshanshan: engineers after being blasted will take off | ONEFLOW u

Is the encryption market a "natural disaster" or a "man-made disaster" in the cold winter?

Show me my personal work list for the past two years. I earn 6K a month in my spare time. It's so delicious to have a sideline
随机推荐
Groovy之列表操作
C # paging calculation total pages, current page data set
PHP内置协议(支持的协议和封装协议)
拜登签署两项旨在加强政府网络安全的新法案
Network address translation nat
数据库连接池:压力测试
After 17 years, Liu Yifei became popular again: ordinary people don't want to be eliminated, but they also need to understand this
Unity's rich text color sets the color to be fully transparent
Follow up to the "June 21" gas explosion accident in Tai'an, Shandong Province: comprehensive investigation and remediation of potential gas safety hazards in restaurants
[live broadcast review] battle code pioneer phase VI: build a test subsystem and empower developers to provide
数据库连接池:连接池功能点的实现
Front and back management system of dessert beverage store based on SSM framework dessert mall cake store [source code + database]
Redistemplate serialization
I rely on the sideline to buy a house in full one year: the industry you despise will make a lot of money in the next ten years!
UE4 obtains local files through blueprints
What are strong and weak symbols in embedded systems?
轻松上手Fluentd,结合 Rainbond 插件市场,日志收集更快捷
网站存在的价值是什么?为什么要搭建独立站
Software architecture
那些没考上大学的人,后来过的怎样