基于双链量子遗传优化的分类规则挖掘算法

针对采用传统智能优化算法挖掘分类规则时易出现分类精度不理想、噪声容忍度差等情况,提出一种基于双链量子遗传优化分类规则挖掘算法.采用双链量子位对分类规则进行实数编码,通过解空间变换将量子位概率幅映射到相应实数集,根据目标函数梯度变化确定量子旋转门转角,并利用量子非门进行个体变异.选取UCI数据库中9组分类数据集对所提出算法分类性能进行测试,结果表明,所提出算法具有较好的分类精度和噪声容忍度.

Mining algorithm for classification rule based on double chain quantum genetic optimization

In order to solve the problems of unsatisfactory classification accuracy, poor noise tolerance and other shortcomings when using traditional intelligent optimization algorithm for classification rules mining, a mining algorithm for classification rule based on double chain quantum genetic optimization was proposed. Double chain quantum bit was adopted for the real number coding of classification rules, and solution space transformation was employed for the mapping of quantum bits probability amplitude to the corresponding real number set. In addition, the rotation angle of quantum rotation gate was determined according to the gradient change of objective function, and the quantum non-gate was utilized for individual mutation. Nine data sets in UCI database were selected to test the classification performance of as-proposed algorithm. Results indicate that the as-proposed algorithm has higher classification accuracy and better noise tolerance.