自动孔子三.pptVIP

* * Pattern RecognitionChapter 3 Data Clustering (Pattern Recognition Concepts, Methods and Applications J.P. Marques de Sa Springer) Chapter 3 Data Clustering Teaching Purpose: The concept of clustering is introduced. Besides, some common clustering methods and conditions are represented. Request： Grasp the concept of clustering and master some common clustering methods. Focus： The Standardization Issue; Tree clustering; K-Means Clustering Contents Unsupervised Classification The Standardization Issue Tree Clustering Dimensional Reduction K-Means Clustering Cluster Validation 3.1 Unsupervised Classification Unsupervised classification: The classification systems designed designed using a so called unsupervised approach (classifying things without any previous knowledge) to discover the internal similarity structure of the patterns in a useful way. The present chapter is dedicated to the unsupervised classification of feature vectors, also called data clustering. Regarding the type of amalgamation approach two categories of algorithms can be considered: Hierarchical algorithms Centroid adjustment algorithms 3.2 The Standardization Issue There is an arbitrarily large number of data clustering solutions that can be radically different from each other. As an example, consider the +Cross data, which is represented in Figure 3.1. Figure 3.1 Cross data with Euclidian clustering Imagine that we are searching for a 2-cluster solution that minimizes the within-class average error: (3-1) with different pairs of patterns x, y in cluster. (3-2a) (3-2b) (3-2c) (3-2d) 3.3 Tree Clustering Consider the Globular data shown in Figure 3.2 a and assume that the similarity between two clusters is assessed by measuring the similarity of its furthest pair of patterns. This is illustrated in the binary tree shown as a vertical icicle plot in Figure 3.3 a, for the complete