论文部分内容阅读
泛系模拟是用以表征两个系统间结构或功能的某种相似性的一个概念。在模式识别中,我们经常需要用一个向量空间或者用一组句子集来表示模式集。泛系模拟为这种用一个系统朱模拟另一个系统的探讨提供了一个统一的描述框架。本文在文献[1,2,3,4]的基础上,进一步探讨了如何通过泛系模拟关系由一个集合的分类导出另一个集合分类的问题。具体地探讨了泛系会诊的形式,推广了泛系树搜索原理。前者从若干个侧面对模式进行分析,然后加以综合,确定识别结果。后者用于处理模式集很大的情况,首先用较简单的方法对模式集进行部分识别,然后再对剩下的难以识别的模式用复杂一些的方法进行识别,这样既使计算量不致于过大,又保证识别的相对准确性。另外,本文讨论了泛系半等价分类系统,把上述的讨论结果推广到半等价分类的情形。
Pansystems simulation is a concept used to characterize some similarity in the structure or function between two systems. In pattern recognition, we often need to use a vector space or a set of sentence sets to represent the pattern set. Pansystems simulation provides a unified framework for this discussion of simulating another system with one system. Based on the literature [1, 2, 3, 4], this paper further explores how to derive another set classification from one set of classification through pantomimetic simulation. This paper discusses the form of pan-phy consultation and generalizes the principle of pan-tree search. The former from several aspects of the pattern analysis, and then integrated to determine the recognition results. The latter is used to deal with a large set of patterns, first of all, a relatively simple method to identify part of the pattern set, and then identify the remaining difficult to identify patterns with more complex methods, so that even if the amount of computation does not result in Too large, but also to ensure the relative accuracy of recognition. In addition, we discuss the pheromone semi-equivalence classification system and generalize the above discussion to semi-equivalence classification.