一种基于规则的模式分类器设计方法

A Designing Method of Rule-Based Classification

针对复杂场景下的模式分类问题,该文提出了一种基于规则的模式分类器设计方法。其基本思想是:首先运用非参数统计方法建立描述样本特征分布的规则集;然后构造一种链式结构将规则集中的元素组织起来,形成模式分类器;最后在以训练样本识别结果为指导的前提下,优化规则集的制定方法和分类器结构。该设计方法的有效性在某对海监视雷达目标识别实验中得到了验证。

This paper proposes a rule-based classification which is deduced fr om nonparametric statistics.To build such a classification,firstly every patte rn's distribution in the feature space is translated into several independent r ules in the way of nonparametric statistics to form a rule set.And then a decis ion chain is built based on the rule set as a classification.Finally this class ification can be easily optimized under the conduct of the training result.The validity of this method is proved by the experimental data which is from a stati oned radar target recognition.