unsupervised clustering

Terms from Artificial Intelligence: humans at the heart of algorithms

Most clustering algorithms are unsupervised, simply, taking groups of values and associating items based on similarity of attributes. Of cousre, the idea of what similarty means for a oaetickar data set does have to be specified by the algorithm user, or is a fixed feature of the algorithm. Examples include k-means and self-organising maps.

Used on page 445