基于仿生模式识别思想的时间序列匹配  

Time Series Matching Based on Bionic Pattern Recognition

作　　者：闾海荣[1] 韩慧[1] 王文渊[1] 

机构地区：[1]清华大学自动化系智能技术与系统国家重点实验室,北京100084

出　　处：《电子学报》2007年第7期1323-1326,共4页Acta Electronica Sinica

摘　　要：仿生模式识别是模式识别理论的一种新模型,它的出发点是要"认识"事物而非"区分"事物,理论的基点在于它确认了样本点在特征空间的连续性.本文根据仿生模式识别的基本思想,提出了一种时间序列匹配的新方法.该方法利用同类样本间的连续性规律,将时间序列排序,并在相邻的时间序列之间添加了若干新的时间序列,以增加样本点.对力感键盘按键压力序列进行分类实验的结果表明,新方法优于基于动态时间弯折的传统方法.Bionic Pattern Recognition is a new model of pattern recognition principle. It＇ s based on ＂matter cognition＂ instead of ＂matter classification＂ and the basic idea of this model is that the sample points in the feature space are continuous. We proposed a new time series matching algorithm in this paper according to the basic idea of bionic pattern recognition. It makes use of the continuity rule between the samples that belong to the same class to sort the time series and to add new samples between the neighbor time series, so that the number of samples increases. We test the algorithm by classifying pressure series of pressure keyboard with simple nearest neighbor and achieve a better experimental result than traditional dynamic time warping （DTW）.

关 键 词：时间序列匹配 仿生模式识别 动态时间弯折 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]