谱聚类算法研究及其在文本聚类中的应用.pdf

摘要煳聊 谱聚类作为一类新兴的、高性能的聚类方法，目前已成为国际上机器学习领域爵弃 关键词：谱聚类，文本聚类，离散化，相似性度量，干细胞优化 垒!!!!!!! 侦士论文 ——‘—————————————————————————————————————————————————————————————————一一 ：=：：=：： Abstract asaclassofnewand Spectralclustering effective methodhas clustering beenthe major researchinthefieldofmachine with traditional learning．Comparedclustering algorithms． candeal spectralclustering with algorithm withan clusteringproblem arbitrary shaped and samplespace tothe converge solution．Basedon globaloptimal researchonspectral clustering，thispaper the and improves the of algorithmimplements Chinesetext system clustering． The mainresearch resultsareas following： discretizationa (1)Since as makesa pretreatment contributiontodata step great mining， itiscombinedwith density-sensitive normalize spectral anddiscretizethe clustering．Firstly measure onginaldata，then thedistanc