Optimal Classification/clustering

From Wikibooks, open books for an open world
Jump to navigation Jump to search

Many people consider clustering the most important unsupervised learning problem. Clustering deals with finding structure in a collection of unlabeled data.

Clustering has many practical applications.

Many clustering algorithms ( k-clustering algorithms ) require a human to specify ahead of time how many buckets (categories, labels, classes, etc.) to divide up the input into.

  • Median cut clustering -- perhaps the fastest clustering algorithm
  • K-means clustering -- a very fast clustering algorithm, with the disadvantage that it does not provide the same result with each run.
  • QT clustering algorithm
  • expectation-maximization algorithm
  • canopy clustering algorithm
  • constrained clustering
  • Fuzzy clustering by Local Approximation of MEmberships (FLAME clustering)
  • self-organizing map
  • vector quantization
  • Learning Vector Quantization


further reading[edit | edit source]