Book Review - Hands on Machine Learning(C1-1)

All of materials(quotes, images, definitions) are from this book.
It’s all just for self-study.

Chapter 1

The Machine Learning Landscape

Why use machine learning?

• training set : The examples that the system uses to learn
• training instance(or sample) : Each training example
• measure : The particular performance measure

  attribute : data type(ex: mileage)

  feature : attribute + value(ex: mileage = 15,000)

• 
• The best solution is to write an algorithm that learn by itself, given many example recordings for each word.
• Data mining : Applying ML techniques to dig into large amounts of data can help discover patterns that were not immediately apparent

Types of Machine Learning Systems

• Whether of not they are trained with human supervision
  • supervised
  • unsupervised
  • semisupervised
  • reinforcement
• Whether or not they can learn incrementally on the fly
  • online
  • batch
• Whether they work by simply comparing new data points to known data points,
  or instead detect patterns in the training data and build a predictive model,
  much like scientists do
  • instance-based
  • model-based

Supervised Learning

• The training data you feed to algorithm includes the desired solutions, called label
• Typical task
  • classification(spam vs ham)
  • regression : to predict target numeric value(price of a car)
    • predictors : given a set of features
    • 
• Some regression algorithms can be used for classification as well, and vice versa.
• The most important supervised learning algorithms(covered in this book)
  

Unsupervised Learning

• The training data is unlabeled
• The system tries to learn without a teacher
• The most important unsupervised learning algorithms(covered in chapter8 and 9)
  
  
• Typical task
  • dimensionality reduction
    • goal is to simplify the data without losing too much information
      => feature extraction : merge several correlated features into one
  • anomaly detection(similar to novelty dection)
    • difference : anomaly detection is more tolerant
  • association rule learning
  • goal is to dig into large amount of data and discover interesting relations between attributes

Semisupervised Learning

• some algorithms can deal with partially labeled training data,
  usually a lot of unlabeled data and a little bit of labeld data
  
• most semisupervised learning algorithms are combinations of unsupervised and supervised algorithms

Reinforcement Learning

• The learning system(agent) in this context, can observe the environment, select and perform actions,
  and get rewards in turn(or penalties in the form of negative rewards)