Abstract
The k_means clustering algorithm has very extensive application. The paper gives out_in clustering algorithm based on density. The algorithm combines distance with data density to adapt to data distribution. It can effectively solve the clustering of data. Out_in clustering based on density reduce distorition by move out and move in. Simulation results show that out_in clustering algorithm is more effective than the k_means clustering algorithm.
Highlights
Clustering algorithm have very extensive application in pattern recognition
Out_in clustering algorithm based on density uses local density as probability of data element and uses density of clusters as probability of clusters
In out_in clustering algorithm based on density, Euclidean distance combined with probability distributions, at the same time, mean is still the clustering center, adapting to different data distributions
Summary
Clustering algorithm have very extensive application in pattern recognition. There are several ways to achieve clustering, such as Gas Ref[1,2,3,4], Clustering Algorithm based on Sofm Ref[5,6,7,8]. In out_in clustering algorithm based on density, Euclidean distance combined with probability distributions, at the same time, mean is still the clustering center, adapting to different data distributions. Base on the Euclidean metric, the optimal cluster center of the given data distribution is E = p1 x1 + p2 x2 + ... Out_in clustering algorithm based on density uses a local strategy to find the optimal solution from the current clustering result. Out_in clustering algorithm based on density uses local density as probability of data. This can be achieved by grid division. Let the currently clustering centers is the optimal clustering centers, The distortion J is monotonically decreasing, according to Algorithm 1.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have