摘要
提出了一种可视化 IGA模型 ,结合 GA在 n维空间上快速搜索的优点和人类在 2维空间上把握数据整体分布的能力 ,采用可视化的方法使用户主动参与搜索过程以加快遗传算法的收敛速度 ,从而减轻用户疲劳 .采用主元分析的方法将 n维空间中的个体向量映射到 2维空间 ,并显示出来 ,用户可以在这个 2维空间中选择一个好的个体加入遗传过程 ,以此来加速算法的收敛 .实验证明可视化 IGA较一般的 IGA有更快的收敛速度 ,对减轻用户疲劳有很好的作用 .该模型用于图像的感性检索 。
A visualized IGA model is proposed in this paper, which combines the ability of GA to search in n-D space fast and that of human to grasp the distribution of data in 2-D space. By a visualized method, user can take part in searching process initiatively to quicken the convergence of GA and alleviate user fatigue. Adopting the principle component analysis to map individuals of n-D space to 2-D Space and visualizing these individuals, user can select a good individual in 2-D mapping space. Thus quicken the convergence. Experiments have proved that visualized IGA has a faster convergence velocity and is very helpful to ease user fatigue. This model is applied into kansei-based image retrieval and has a good performance.
出处
《小型微型计算机系统》
CSCD
北大核心
2004年第3期399-403,共5页
Journal of Chinese Computer Systems
基金
国家"973"资助课题 ( G19980 3 0 5 0 )资助