A pattern recognition method based on linguistic ordered weighted distance measure

作者：Cai Mei^*; Gong Zaiwu; Wu DaQin; Wu Minjie

来源：Journal of Intelligent and Fuzzy Systems, 2014, 27(4): 1897-1903.

DOI：10.3233/IFS-141155

• Pattern recognition (PR)
• ordered weighted distance measure
• maximizing deviation method
• linguistic symbolic computational model

Pattern recognition (PR) can roughly be defined as an art of assigning to individual objects proper names of patterns consisting of similar to them in an assumed sense objects. Distance measure is a key concept for all attempts to deal with pattern recognition. This paper addresses the issue of linguistic ordered weighted distance measure for PR. The method enables the handling of imprecise information given by linguistic variables. Firstly we construct a non-linear programming model to identify weights of attributes based on the principle that the relative weights of positions should maximize deviations of unknown objects, in order to distinguish the objects as far as possible. Then we utilize the LOWD operator to aggregate the global evaluation of an unknown object and classify it to the pattern with the shortest distance. Finally, a numerical example is used to show the process and effects of our new proposed method.

• 出版日期2014
• 单位南京审计大学; 南京信息工程大学