摘要
针对SVDD处理大数据样本时存在时间复杂度较大的问题,提出一种随机蚕食快速增量式支持向量域数据描述(RGInc-SVDD)算法.RGInc-SVDD首先利用随机抽样定理将样本训练集分割为多个子集,然后将其中一子集用于建模Inc-SVDDi分类器,最后利用迭代蚕食算法合并增长Inc-SVDDi分类器,以生成整个训练集的SVDD分类器.RGInc-SVDD算法使得SVDD的时间复杂度从O(N3)降到O(N2r/Gn2).实验结果验证了RGInc-SVDD算法的正确性和有效性.
To address the problem of SVDD（support vector data description） processing larget samples dataset with huge time complexity,a novel RGInc-SVDD（random greed incremental SVDD） algorithm was proposed.Firstly,using the SL（sampling lemma） to divide the training samples dataset into several small samples subsets;secondly,create an Inc-SVDDi model with one of samples subsets.Then,the rule of interactive random greed was applied to grow the Inc-SVDDi until the SVDD being created with whole training samples information.The RGInc-SVDD algorithm makes the time complexity significantly decrease from O（N3） to O（N2r/Gn2/k）,where respectively denote the number of training samples and the number of random greed in each interactive step.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第12期94-97,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家重点基础研究发展计划资助项目(2007CB307102)
国家自然科学基金资助项目(60975052
F030508)
关键词
支持向量
抽样
随机
迭代蚕食
增量式
support vector sampling random interactive random greed incremental
