Indexing Algorithm Based on Improved Sparse Local Sensitive Hashing

Journal of Multimedia

关键词: PCA;    Sparse;    Local Sensitive Hashing;    Indexing;   

Others  :  1017287
DOI  :  10.4304/jmm.9.1.35-42

【 摘 要 】

In this article, we propose a new semantic hashing algorithm to address the new-merging problems such as the difficulty in similarity measurement brought by high-dimensional data. Based on local sensitive hashing and spectral hashing, we introduce sparse principal component analysis (SPCA) to reduce the dimension of the data set which exclude the redundancy in the parameter list, and thus make high dimensional indexing and retrieval faster and more efficient. In the meanwhile, we employ Boosting algorithm in machine learning to determine the threshold of hashing, so as to improve its adaptive ability to real data and extend its range of application. According to experiments, this method not only has satisfying performance on multimedia data sets such as images and texts, but also performs better than the common indexing methods. 

【 授权许可】

   
@ 2006-2014 by ACADEMY PUBLISHER – All rights reserved.