【Meachine Learning】07 some summary of Meachine Learning

1 cost function

model	cost function	Neural Network Name
linear regression	squared error cost function	MeanSquaredError
logistic regression	loss function	BinaryCrossentropy

2 numpy operation

2.1 dot product and transpose

• 
• matrix multiplication
• more about (*, np.multiply, np.dot, np.matmul, and @)

2.2 boardcast and add

2.2.1 boardcast

boardcast index:

boardcast intuition:

Boardcast Matrix:

2.2.2 add

Boardcast VectorAdd:

2.3 slice

• numpy array slice

3 code demo for different algorithms

3.1 linear regression demo

features	target	code
- population of city	- restaurant profit in target city	predict profit by population
- Size (sqft) - Number of Bedrooms - Number of floors - Age of Home	- Price (1000s dollars)	predict house price by some features

3.2 logistic regression demo

features	target	code
- Exam 1 score - Exam 2 score	- Admitted - Not admitted	predict admission chance by exam score
- test 1 score - test 2 score	- microchip accept - microchip reject	predict microchip inspect result by test score

3.3 neural network demo

• binary_handwritten_digit_recognization
• multiple_handwritten_digit_recognization

3.4 decision tree demo

• diagnose Context Cardiovascular diseases by multiple features

3.5 intuition for model evaluate

• model evaluate for: linear regression
• model evaluate for: Neural Network: linear regression
• model evaluate for: Neural Network: logistic regression

3.6 Clustering demo

• K-Means:
  • ClusteringIntuition
  • KMeans01_ImageCompression
• Anomaly detection:
  • detect computer server anomalies

4 derivative

遇到复杂计算找python绝对不让你失望，sympy是一个Python的科学计算库，用一套强大的符号计算体系完成诸如多项式求值、求极限、解方程、求积分、微分方程、级数展开、矩阵运算等等计算问题。

doc:

• https://blog.csdn.net/cj151525/article/details/95756847

100 other doc summary

100.1 mathmatics

• pupulation mean

100.2 markdown

• mathjax/katex/latex in markdown
• MathJax markdown demo

100.3 concepts in ML

• batch, batch size, iteration/step