3. Basics of Clustering and k-means clustering.pdf
Document Details
Uploaded by FinerLimeTree
University of Southern Denmark - SDU
Tags
Full Transcript
Challenges Assume we are given a hypothetical quality function q to decide for a given partition of a set of n points whether or not this partition constitutes a (good) clustering. â–º na"ive method: test q on all possible clusterings with k partitions (clusters) (2 k ?) â–º problems:...
Challenges Assume we are given a hypothetical quality function q to decide for a given partition of a set of n points whether or not this partition constitutes a (good) clustering. â–º na"ive method: test q on all possible clusterings with k partitions (clusters) (2 k ?) â–º problems: â–º there are O(kn ) many partitions ink clusters â–º and we don't have this function q, actually Therefore, we need heuristic solutions for both problems: â–º efficient search for solutions â–º efficient and effective modelling of q There are many such heuristic solutions around, that is, we have a plethora of clustering algorithms in the literature.
