Clustering

• Also called unsupervised learning, segmentation (in image processing), quantization or discretization (in signal processing).
• In pattern recognition most work assumed attributes are numeric. In general there is a mix of discrete and continuous attributes.

 

Definition: Given N k-dimensional feature vectors, find a “meaningful” partition of the N examples into c subsets. c may be given, or may be “discovered” by the algorithm. much more difficult that classification since in latter partition is given with the data and all that is needed is its description (or its characteristic function).

 

 

---