Let X = { x 1, x 2,…, x n } be a set of samples to be clustered; P = { p 1, p 2,…, p c } is a set of c clustering prototypes, where c is the number of clusters.
K-means performs the clustering by minimizing the distances of the vectors to their cluster prototype.
If any distance is smaller than the distance of the corresponding vector to its own cluster prototype, it is evidence that k-means has potential to operate between the clusters.
The proposed hybrid system is designed and tested in a real environment, with an event filter cluster prototype based on the architecture of the Compact Muon Solenoid experiment at CERN.
Due to this re-weighting, all points in the vicinity of the chosen cluster prototype receive a larger γ-index.
The lowest γ-index indicates that the corresponding filter is inside a region with many other filter examples and should therefore be chosen as cluster prototype.
To ensure that we also sample prototypes from other clusters, an incremental procedure of choosing and re-weighting is applied to determine a predefined number of cluster prototype filters.
Of the clinical chemistry measurements listed in Table 3 for each cluster prototype, ALT and AST levels clearly distinguish Cluster 1 samples labelled with the prototype feature as moderate necrosis of the centrilobular region of the liver from the two other clusters.
Ion/molecule reactions with C 2H 6: [V2P2O10].+ has also been chosen as cluster prototype to demonstrate the reactions of the various cluster ions [V x P4− x O10].+ (x=0, 2 4) with ethane, and the potential-energy profile is shown in Figure 4 (Cartesian coordinates of all structures can be found in the Supporting Information and the relative electronic and free energies are summarized in Table 5).
First one is to compute cluster prototypes, v i for each cluster i. Cluster prototypes refer to the average values of features of member objects (real data, z k ), weighted by membership degree of the object.
