基于克隆选择的模糊分类规则提取算法

分类是许多研究领域的关键问题，模糊规则的提取质量对分类器的性能又有着极大影响．所提取的规则不仅在分类能力上要达到最优，同时在规则数量上也不能太多，否则会影响规则搜索和匹配的速度．结合人工免疫的克隆选择原理，采用克隆选择算法，提取通过多精度模糊分割产生的大量模糊if—then规则中的少数精华规则，从而建立了模糊分类所需要的有效规则集合，同时还对优化目标函数进行了改进．经仿真实验证明，该方法所提取的模糊规则具有分类准确率高，规则数目较少等特点。

Extracting fuzzy classification rules using clonal selection algorithm

Classification is crucial for many research domains, but the quality of extracted fuzzy rules has a great influence on the performance of classifiers. It is not only necessary that extracted rules have optimal performance in classification, but also the number of rules must be as small as possible, otherwise, rule searching and matching becomes slow. In this paper, using the clone selection algorithm, the best rules were extracted from massive sets of fuzzy if-then rules generated from multiple precision fuzzy partitions. Thus a set of effective rules for fuzzy classification were developed. Also the optimal objective function was improved. Test results prove that the proposed method uses fewer rules and has high classification preci sion.