Clustering
  • is an unsupervised learning finding patterns in unlabeled data

Clustering - Types

Type

Description

Example Algorithms

Problem

Hard Clustering

  • assigns an instance to 1 cluster
  • clusters do not overlap
  • element either belongs to a cluster or it does not
  • each point/instance is given a “hard” assignment to exactly one cluster-center
  • does not allow uncertainty in cluster-center/class-membership
  • does not allow point/instance to belong to more than one cluster-center

Soft Clustering

  • assigns probabilities that an instance belongs to each cluster
  • clusters may overlap
  • strength of association between clusters and instances

Clustering - Other