文献类型：期刊文献

英文题名：Sparse coding with earth mover's distance for multi-instance histogram representation

作者：Zhang, Mohua[1,2];Peng, Jianhua[1];Liu, Xuejie[3]

第一作者：Zhang, Mohua;张墨华

通讯作者：Zhang, MH[1]

机构：[1]Natl Digital Switching Syst Engn & Technol Res Ct, Zhengzhou 450002, Henan, Peoples R China;[2]Henan Univ Econ & Law, Coll Comp & Informat Engn, Zhengzhou 450002, Henan, Peoples R China;[3]SUNY Buffalo, Buffalo, NY 14260 USA

第一机构：Natl Digital Switching Syst Engn & Technol Res Ct, Zhengzhou 450002, Henan, Peoples R China

通讯机构：[1]corresponding author), Natl Digital Switching Syst Engn & Technol Res Ct, Zhengzhou 450002, Henan, Peoples R China.

年份：2017

卷号：28

期号：12

起止页码：3697-3708

外文期刊名：NEURAL COMPUTING & APPLICATIONS

收录：;EI(收录号：20161202141857);Scopus(收录号：2-s2.0-84961204758);WOS:【SCI-EXPANDED(收录号:WOS:000412842200004)】；

基金：The work was support partly by the foundation for innovative research groups of the national natural science foundation of China (Grant No. 61521003), partly by science and technique foundation of HeNan province, China (Grant No. 152102210087), partly by foundation of educational committee of HeNan province, China (Grant No. 14A520040).

语种：英文

外文关键词：Multi-instance learning; Histogram representation; Sparse coding; Earth mover's distance

摘要：Sparse coding (SC) has been studied very well as a powerful data representation method. It attempts to represent the feature vector of a data sample by reconstructing it as the sparse linear combination of some basic elements, and a norm distance function is usually used as the loss function for the reconstruction error. In this paper, we investigate using SC as the representation method within multi-instance learning framework, where a sample is given as a bag of instances and further represented as a histogram of the quantized instances. We argue that for the data type of histogram, using norm distance is not suitable, and propose to use the earth mover's distance (EMD) instead of norm distance as a measure of the reconstruction error. By minimizing the EMD between the histogram of a sample and the its reconstruction from some basic histograms, a novel sparse coding method is developed, which is refereed as SC-EMD. We evaluate its performances as a histogram representation method in tow multi-instance learning problems-abnormal image detection in wireless capsule endoscopy videos, and protein binding site retrieval. The encouraging results demonstrate the advantages of the new method over the traditional method using norm distance.