Machine Learning by Andrew Ng -

• Supervised Learning

• Unsupervised Learning

  
  Supervised Learning and Unsupervised Learning.png

Supervised Learning

"Right Answers“ given; Predict continuous output.

Already know that is the data set of 'correct answers' that we would like that algorithms have predict on that data set.

Regression Problem

Predict continuous valued output (e.g., price)

Regression Problem.png

Classification Problem

Discrete valued output (0 or 1) (or more than two values)

Classification Problem.png

A represents a linear classification problem and B represents a non-linear classification problem.

About infinite number of features?

A learning algorithm that might deal with an infinite number of features.
So how to deal with an infinite number of features? And how to store an infinite number of attributes?

The answer is Support Vector Machine

Support Vector Machine

There will be a neat mathematical trick that will allow a computer to deal with an infinite number of features.

Support Vector Machine.png

Unsupervised Learning

Unsupervised learning is a type of machine learning algorithm used to draw inferences from datasets consisting of input data without labeled responses.

The most common unsupervised learning method is cluster analysis, which is used for exploratory data analysis to find hidden patterns or grouping in data.

Given a data set, we're not told what to do with it and we're not told each data point is. Instead, we just find some structure in the data.

Unsupervised Learning.png

Given this data set, an Unsupervised Learning algorithm might decide that the data lives in two different clusters.

Unsupervised Learning algorithm may break these data into these two separate clusters. So this is called a clustering algorithm.

Examples of Unsupervised Learning Algorithm

• Organize computing clusters
• Social network analysis
• Market segmentation
• Astronomical data analysis
• Cocktail party problem (/effect)
  Cocktail Party Problem.png
  Cocktail party problem algorithm with Octave programming environment
  [W,s,v] = svd ( ( repmat ( sum ( x. * x, 1 ), size ( x, 1), 1 ). * x ) * x' );