摘要
通过对已标示和未标示数据的学习和分类,提出一种改进微分进化算法的半监督模糊聚类。先从大量的数据中选取一小部分进行标记,然后利用标记数据来指导进化过程,实现对未标记数据的分类。通过参考粒子群算法惯性权重思想,引入惯性加权系数,在计算初期能够维持个体的多样性,后期能够加快算法的收敛速度,有效提高了算法的性能。遥感图像数据实验结果显示该方法可以提高分类精度。
Through studying and classifying labeled and unlabeled data, this paper proposed a modified differential evolution algorithm for semi-supervised fuzzy clustering. Firstly, a small part of data was labeled from the whole dataset, and then these labeled data were used to guide the evolution process to partition unlabeled data. The modified algorithm introduces inertia-weighted coefficient by considering inertia-weighted idea of particle swarm algorithm, which keeps diversity of individual at early stages and quickens convergent speed at later stages, and at the same time improves the performance of the algorithm. The experimental results for remote sensing data indicate that the proposed approach can improve classification accuracy.
出处
《计算机应用》
CSCD
北大核心
2009年第4期1046-1047,1051,共3页
journal of Computer Applications
基金
国家自然科学基金资助项目(6047206060572034)
2006年教育部新世纪优秀人才计划项目(NCET-06-0487)
关键词
模糊聚类
标示数据
未标示数据
微分进化算法
半监督学习
fuzzy cluster
labeled data
unlabeled data
differential evolution algorithm
semi-supervised learning