摘要
自模糊c回归模型(FCRM)聚类算法提出以来,其在收敛速度和鲁棒性等方面的改进一直是研究的热点。为此,M.S.Yang等提出模糊c回归模型α(FCRMα)算法,该算法引入参数α,对FCRM算法进行了快速迭代,提高了算法的鲁棒性。然而该算法存在参数α选值的问题。针对这种情况,基于相似关系理论提出一种自适应的α参数取值方法,得到了自适应迭代过程的SAFCRM算法。多个实验表明,相对于FCRMα算法,SAFCRM算法具有更强的鲁棒性,收敛速度更快,得到的回归效果也更好。
Since the presentation of fuzzy c-regression models (FCRM) clustering algorithm, the improvement on its convergence speed and robustness has been the focus of research. For this reason, M.S. Yang proposed FCRMα algorithm. In this algorithm, parameter α is introduced to expedite the iterative operation of FCRM algorithm and improves the robustness of it. However, the algorithm has the problem of parameter α selection. To solve the problem, we present an adaptive parameter α value assignation method based on similarity relation theory, and derive the SAFCRM algorithm for adaptive iteration process. Several experiments show that the SAFCRM algorithm has stronger robustness, faster convergence speed and better regression results than FCRMα algorithm.
出处
《计算机应用与软件》
CSCD
2016年第1期330-333,共4页
Computer Applications and Software
基金
浙江省温州市科技计划项目(G20130031)
浙江省高职高专院校专业领军项目(lj2013146)
温州市公益性科技计划项目(G20140049)
关键词
切换回归
模糊聚类
参数优化
自适应
Switching regression Fuzzy clustering Parameter optimisation Adaptive