文章目錄
- **Classification**
- 抽象模組化
- **Two Boxes**
- **Two Classes**
- **Gaussian Distribution**
- **Probability from Class**
- **Maximum Likelihood**
- 修改模型
- **Posterior Probability** （後驗機率）
Classification
抽象模組化
- Function（Model）
- Loss Function
- Find the best function:
- eg.perceptron，SVM
Two Boxes
Two Classes
Estimating the Probabilities From training data
Given an , which class does it belong to(來自于Class1的機率）
Generative Model
Gaussian Distribution
Probability from Class
Assume the points are sampled from a Gaussian distribution
Find the Gaussian distribution behind them
Maximum Likelihood
We have the “Water” type Pokémons:
We assume generate from the Gaussian
均值和方差如何求，參考我的博文多元高斯分布的最大似然估計
but 隻有54% accuracy …
修改模型
準确度提升：（2020李宏毅）機器學習-Classification
Posterior Probability （後驗機率）
其中，
數學推導：
==>
令
=>
抽象化：