A Star Spectrum Outliers Mining System Based on PSO
Jianghui Cai, Jifu Zhang and Xujun Zhao
Outlier mining is a kind of effective way of finding the spectrum data of unknown celestial body. Many clustering algorithms are used to detect outliers as by-products of the clustering processes. The rational of using clustering algorithms to detect outliers is based on the understanding that outliers and cluster objects are mutually complementally. That is, an outlier shall not be in any cluster, and a cluster object shall not be an outlier. However, in high-dimensional space, they fail to retain their effectiveness. Consequently, for high-dimensional data, the notion of finding meaningful outliers becomes substantially more complex and nonobvious. In this paper, we discuss new techniques for outlier detection. A kind of outlier mining algorithm based on particle swarm optimization is put forward. Using VC++, Oracle9i as development tools, the outlier mining system on star spectrum data is designed and realized. The running results of the system show that it is feasible and valuable to apply this method to mining the outlier in spectrum data.
Keywords: PSO, outliers, star optical spectrum data.