Potential energy cluster algorithm for automatically determining cluster center

An automatic determination and clustering algorithm technology, applied in the field of cluster analysis, can solve the problems of artificially setting the number of clusters and incomplete consideration of the allocation mechanism, and achieve good clustering effect, good clustering effect, and clustering accuracy high effect

CN106557786AInactive Publication Date: 2017-04-05JIANGNAN UNIV
0 Cites 2 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2017-04-05
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The present invention discloses a potential energy cluster algorithm for automatically determining a cluster center. The objective of the invention is mainly to solve the two defects that the potential-based hierarchical agglomerative clustering (PHA) needs to artificially arrange the number and the distribution mechanism only depends on the distance so as to weaken the potential energy influence. The method comprises: obtaining the potential energy of each data point, arranging the potential energy of each data point from big to small, finding out the distance from father nodes of each data point to the father nodes, calculating the [Gamma]i value of each node, employing the K-means algorithm to perform clustering (K is 2) in the one-dimensional space according to the [Gamma]i value of each node, and selecting one class with few data points as a clustering center set. After the clustering center is determined, each cluster center represents one class, and the residual data points are included into the cluster where the samples with having the smaller potential energy and having the nearest distance are located. The potential energy cluster algorithm for automatically determining the cluster center automatically determines the number of the clusters and has high accuracy and higher practicality.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the technical field of cluster analysis, and relates to the improvement and optimization of a potential-based fast hierarchical clustering algorithm (Potential-based hierarchical agglomerative clustering, PHA). Specifically, it is a potential energy clustering algorithm for automatically determining cluster centers, which can be used in pattern recognition, data mining, image processing and other fields. Background technique

[0002] Clustering algorithm is an important branch of data mining, which gathers data through some similarity measure and clustering criterion without prior knowledge about potential data distribution. Cluster analysis plays an important role in many fields, including biology, artificial intelligence, customer relationship management, information retrieval, machine learning, etc.

[0003] However, there are many widely used clustering algorithms that cannot automatically determine the number of clusters....

Examples

Embodiment Construction

[0042] In order to clarify the purpose, technical solutions and advantages of the present invention, the present invention will be further described in detail below in conjunction with specific embodiments and accompanying drawings.

[0043] refer to figure 1 , the specific implementation process of the present invention comprises the following steps:

[0044] Step 1: first find the potential energy Φ of each data point i , where the potential energy Φ between every two points ij It is defined as follows:

[0045]

[0046] where r ij is the Euclidean distance between point i and point j, and δ is used to avoid the case where the denominator is zero. About δ:

[0047]

[0048] δ=mean(MinD i ) / S

[0049] where MinD i is the shortest distance from point i to other points, S is a scale factor, generally set to 10, and N is the number of data points. The potential energy Φ between every two points is calculated ij After that, the potential energy Φ of each point i ...