仿射传播聚类算法的搜索策略优化.docVIP

仿射传播聚类算法的搜索策略优化 　　摘 要：针对多楼层指纹定位中，大规模的指纹样本使得匹配算法复杂度增加，不仅阻碍了系统的实时性，还增加了移动端的能量损耗的问题。依据仿射传播聚类算法理论对指纹库进行分块处理，可以有效减少计算量。复杂环境下的指纹样本搜索通常采用折半查找法，用于在粗定位阶段得出聚类质量最优结果对应的偏向参数，但此方法花费时间较长。在保证计算质量前提下，为了提高聚类速度，研究了其在粗定位阶段的产生与匹配过程，并给出了对折半查找法进行改进的方法。实验结果表明，对于同一样本空间进行聚类，优化后的折半查找法可以减少算法迭代次数，提高系统工作效率，所用的迭代时间74.5%以上都短于传统折半查找法。 　　关键词： 　　仿射传播；聚类分析；折半查找法；偏向参数 　　DOI：10.15938/j.jhust.2018.03.007 　　中图分类号： TN929.5 　　文献标志码： A 　　文章编号： 1007-2683（2018）03-0039-05 　　Search Strategy Optimization of Affine Propagation Clustering Algorithm 　　DONG Jing-wei， ZHANG Tian-qi， LIU Yang， YANG Guang 　　（School of Measurement-control Technology and Communications Engineering，Measurement and Control Technology and Instrument 　　Key Laboratory of Universities in Heilongjiang province， Harbin University of Science and Technology， Harbin 150080， China） 　　Abstract：For multi floor fingerprint location， large scale fingerprint samples increase the complexity of the matching algorithm， which not only hinders the real-time performance of the system， but also increases the energy loss of the mobile terminal.According to the theory of affinity propagation clustering algorithm， we can reduce the amount of computation.Fingerprint sample search in complex environment usually adopts the binary search method， and the optimal parameters for bias results clustering quality of the coarse positioning stages， but this method takes a long time.In the calculation of quality guarantee under the premise， in order to improve the clustering speed， the coarse positioning stage and the matching process， and gives the method to improve the binary search method.The experimental results show that the clustering of the same sample space， the binary search method the optimized algorithm can reduce the number of iterations， improve the work efficiency of the system， and the use of the iteration time is 74.5% shorter than the traditional binary search method. 　　Keywords：affinity propagation； cluster analysis； binary search method； bias p