山西大学学报（自然科学版）2017，Vol.40Issue(2)：234-243,10.DOI:10.13451/j.cnki.shanxi.univ(nat.sci.).2017.02.008
面向等维独立多流形的增量学习算法IMM-ISOMAP
Incremental Learning Algorithm IMM-ISOMAP for Well-separated Multi-manifolds with Same Intrinsic Dimension
摘要
Abstract
Manifold learning is an important research issne in the fields of machine learning and data mining.The classical algorithms always assume that the high dimensional batched data exist in a single manifold,and can not effectively deal with the high dimensional multi-manifold data.We propose an incremental learning algorithm IMM-ISOMAP for equal dimension independent multi-manifolds.The algorithm computes the dynamic neighborhood of the new sample,and then divides the new samples into sub-manifolds by extending the tangent space.Finally it realizes the classification of the new samples and calculates the final low dimensional embedding.The experimental results show that the proposed algorithm can be effectively applied to artificial data and real image data.关键词
流形学习/增量学习/等维独立多流形/动态邻域/切空间/IMM-ISOMAPKey words
manifold learning/incremental learning/equal dimension independent multi-manifolds/dynamic neighborhood/tangent space/IMM-ISOMAP分类
信息技术与安全科学引用本文复制引用
高小方,刘杰飞..面向等维独立多流形的增量学习算法IMM-ISOMAP[J].山西大学学报（自然科学版）,2017,40(2):234-243,10.基金项目
国家自然科学基金(61303091 （）
61201453) （）
高等学校博士学科点专向科研基金(20131401120004) （20131401120004）
山西省回国留学人员科研资助项目(2016-002) （2016-002）
山西省自然科学基金(2015021091) （2015021091）
山西省基础研究计划项目(2014021022-2) （2014021022-2）
山西省高校科技创新项目(2015108 （）
2015109) （）
