Clustering and Categorisation

• Two standard applications are clustering and categorising.
• Clustering takes a bunch of data points and groups them together.
• Nearest neighbor and SOMs work well for this.
• Categoristaion takes data typically of vectors with the correct category.
• The system then learns to categorise based on these training data.
• It is then tested on other data.
• University of California at Irvine's Categorisation Benchmark
• I've supervised a lot of categoristation theses.