CS229 Lecture 1 - Welcome

“ML & AI will change the world.”

Goal of this lecture

Be the expert in machine learning.
But machine learning is so pervasive.

Prerequisite

• Big O notation
• Queue, stacks, and binary trees
• Basic with probability
• Basic linear algebra & eigenvector

Best part of CS229

• class project
  • complete meaningful ML project as a team

    Using convex optimization algorithm is the main key

CSxxx class difference

• CS229 : most mathmethical
• CS229a : less math than CS229
• CS231n : mostly about deep learning

Quick overview

“Machine learning is everywhere.
Do meaningful work and take the opportunity of superpower.”

Supervised learning

• used in solving a regression problem & classification problem
• X -> Y
• In actual study, there are multiple features which consist high-dimension
• SVM(support vecotr machine) allows infinite dimesion vector(learn in next lectures)

Regression problem

• Y is continous value
• Example : predicting housing price
  • X : size(squared feet) / Y : price

Classification problem

• Y is discrete value
• Example : classifying breast tumor
  • X : tumor size / Y : malignancy(0 : non-malignant, 1: malignant)

Input X, label Y -> find mapping(+ X is given) -> Return new Y

Unsupervised learning

• Input X, and no label Y
• Find interesting pattern of characteristic by its own
• Example : Cocktail party problem(using ica algorithm)
  • How to seperate voices? (Clustering)

Reinforcement learning

• Don’t know optimal ways
• Keep training

Conclusion

The model looks like this