Machine Learning 000

Stanford Link

Definition

Arthur Samuel: Machine Learing is the field of study that gives the computer the ablility to learn with out being explicitly  programmed.

Tom Mitchell: a computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at task in T, as measured by P, improves with experience E.

Taxonomy 

• Supervised Learning 
• Unsupervised Learning
• Reinforcement Learning

Supervised Learning

• eg: Housing Price Prediction
  • Given: a dataset that contains n samples(x¹, y1),...,(xⁿ, yⁿ), x is for square feet, y is for price
    
  • Task: if a residence has x square feet, predict its price?

           

• Regression vs Classification
  • regression: if y ∈ R is a continuous variable, e.g., price prediction
  • classification: the label is a discrete variable,e.g., the task of predicting the types of residence
• Supervised Learning in Computer Vision
  • Image Classification, x = raw pixels of the image, y = the main object
  • Object localization and detection, x = raw pixel of the  image, y = the bounding boxes
• Supervised Learning in Natural Language Processing
  • Machine translation

Unsupervised Learning

• Dataset contains no labels: x¹,...,xⁿ
• Goal(vaguely-posed): to find interesting structures in the data

• Clustering

 Reinforcement Learning

•  The algorithm can collect data interactively

 

Other Tools/Toptics In This Course

• Deep learning basics
• Introduction to learning theory
  • Bias偏差 variance方差 tradeoff
  • Feature selection
  • ML advice
• Broader aspects of ML
  • Robustness/fairness