Course:CPSC312-2023-Kmeans++

Authors:Ali,

Contents

• 1What is the problem?
• 2What is the something extra?
• 3What did we learn from doing this?
• 4Links to code etc

What is the problem?[edit | wikitext]

K-means can be used as a vector quantization to segment images into meaningful components. However it doesn't guarantee convergence due to its dependence on initial conditions of the segmentations.

What is the something extra?[edit | wikitext]

Use Haskell to implement K-means++ to select the initial conditions for K-means and determine the feasible scalability limit of the algorithm as it requires k passes over the dataset (the image in this case)