参数自适应学习的半监督混合聚类算法

Adaptive parameter learning semi-supervised hybrid clustering

提出目标重构的半监督混合聚类算法SSABC。使用人工蜂群算法结合有标记数据研究半监督聚类的准确率提高问题,利用设定参数衡量有无标记数据的权重,依此重新构造目标函数找出数据聚类中心;使用APL-SSHC算法完成半监督混合聚类的参数自适应学习工作,结合自适应学习理论优化权重参数,将参数的确定与聚类过程结合加快聚类过程。UCI数据集实验结果表明,该算法能够找到合理的聚类中心点,APL-SSHC算法与其它聚类算法相比有更好的聚类效果。

A semi-supervised artificial bee colony algorithm that was capable of reconstructing target was proposed. Combined with the marker data, the artificial bee colony algorithm was used to improve the accuracy of SSABC algorithm. The parameters were used to measure the weight of the labeled data. Based on this, the objective function was reconstructed to identify the data clustering center. The APL-SSHC algorithm was used to perform the parameter adaptive learning and the parameter weight was optimized based on the adaptive learning theory. The clustering process was improved by combining the parameter determination process and the clustering process. Results of UCI data set experiment show that, it is verified that the proposed algorithm can get a reasonable clustering center point, and the APL-SSHC algorithm performs better than other clustering algorithms.