Classification

This exercise will introduce you to the foundations of linear classification, logistic regression and decision boundaries. Additionally, you will create Histogram of Oriented Gradients (HOG) feature descriptors. HOG features are useful for describing global properties using local gradient information in images. The HOG features will be used in a later exercise about classification.

Learning goals

After this week's exercises, you should be able to:

  • Identify and and implement a decision boundary for classification.
  • Visualize the results and reason about the importance of choice of boundary charateristics.
  • Implement and reason about logistic regression for classification.
  • Reason about and experiment with HoG features and how they can be used for classification.
  • Reason about the importance of data preprocessing in terms of classification.