文章摘要
基于双流形正则化的迁移稀疏算法研究
The Approach of Dual-manifold Regularized Transfer Sparse Algorithm
投稿时间:2020-03-03  修订日期:2020-03-03
DOI:
中文关键词: k均值  迁移稀疏编码  流形正则化  拉普拉斯图
英文关键词: K-means  Transfer sparse coding  Manifold regularization  Laplacian graph
基金项目:安徽省教育厅2019自然科学基金重点项目(KJ2019A0603);医学物理与技术安徽省重点实验室开放基金资助项目(LMPT201706 )
作者单位E-mail
孟欠欠 淮北师范大学 mq_ahu@126.com 
摘要点击次数: 17
全文下载次数: 
中文摘要:
      针对迁移稀疏编码算法仅考虑样本间的几何流形结构,忽略特征间流形关系,以及随机初始化基向量的问题,提出双流形正则化的迁移稀疏编码算法。算法首先通过K均值聚类均衡选择基向量;其次将特征间局部流形结构信息构建拉普拉斯图,并将该拉普拉斯图作为正则化项引入到迁移稀疏编码算法的目标函数中,同时考虑了跨域图像的几何流形结构信息和分布差异信息,保证编码的稳定性和鲁棒性。为了证实算法的有效性和可行性,在通用跨域图像数据集进行实验,实验表明本文方法可有效提高图像的分类准确率。
英文摘要:
      To solve the problem that transfer sparse coding only rely too much on the geometric manifold structure of samples, ignore the manifold information of sample features and adopts initialize dictionary randomly, a dual-manifold regularized transfer sparse concept coding is proposed. Firstly, K-means clustering method is used to balance the initial dictionary; Secondly, the local geometrical information of features is used to construct Laplacian graph,and the Laplacian graph terms is introduced into the objective function of transfer sparse coding , What’s more, the information of geometric features and the distribution differences are considered, which can learn a more stable and robust image representation. To verify the effectiveness and feasibility of the algorithm,we construct the experimental on universal cross domain image datasets,the experimental results show the proposed algorithm obtain the significant improvement on the classification accuracy.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭

手机扫一扫看 分享按钮