首页|Cross-Media Correlation Analysis with Semi-supervised Graph Regularization
Cross-Media Correlation Analysis with Semi-supervised Graph Regularization
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
Springer Nature
With the rapid development of multimedia data such as text， image， cross-media retrieval has become increasingly important， because users can retrieve the results with various types of media by submitting a query of any media type。 The measure of relevance among different media is a basic problem。 Existing methods usually only consider the original media instances (such as images， texts) but ignore their patches。 In fact， cross-media patches can emphasize the important parts and improve the precision of cross-media correlation。 What's more， existing cross-media retrieval methods often focus on modeling the pairwise correlation with the similarity matrix is a constant matrix， while the similarity matrix which is not a constant matrix can improve the accuracy。 In this paper， we propose a novel algorithm for cross-media data， called cross-media correlation analysis with semi-supervised graph regularization (CMCA)， which can not only take full advantage of both the media instances and their patches in one graph， but also explore the similarity matrix which can improve the correlation between data。 CMCA explores the sparse and semi-supervised regularization for different media types， and integrates them into a unified optimization matter， which increases the performance of the algorithm。 Comparing with the current state-of-the-art methods on two datasets (i。e。， Wikipedia， XMedia)， the comprehensive experimental results demonstrate the effectiveness of our proposed approach。
Cross-mediaThe measure of relevanceSimilarity matrix Cross-media patches
Hong Zhang、Tingting Qi
展开 >
College of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan 430081, China,Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System, Wuhan, China
College of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan 430081, China
NETL
Springer Nature
