Abstract
This paper presents a novel dimensionality reduction technique named Tensor Distance based Multilinear Multidimensional Scaling (TD-MMDS). First, we propose a new distance metric called Tensor Distance (TD) to build a relationship graph of data points with high-order. Then we employ an iterative strategy to sequentially learn the transformation matrices that can best keep pair-wise TDs of the high-order data in the low-dimensional embedded space. By integrating both tensor distance and tensor embedding, TD-MMDS provides a uniform framework of tensor based dimensionality reduction, which preserves the intrinsic structure of high-order data through the whole learning procedure. Experiments on standard image and video datasets validate the effectiveness of the proposed TD-MMDS.
Original language | English |
---|---|
Title of host publication | MM'09 - Proceedings of the 2009 ACM Multimedia Conference, with Co-located Workshops and Symposiums |
Pages | 577-580 |
Number of pages | 4 |
DOIs | |
Publication status | Published - 28 Dec 2009 |
Event | 17th ACM International Conference on Multimedia, MM'09, with Co-located Workshops and Symposiums - Beijing, China Duration: 19 Oct 2009 → 24 Oct 2009 |
Conference
Conference | 17th ACM International Conference on Multimedia, MM'09, with Co-located Workshops and Symposiums |
---|---|
Country/Territory | China |
City | Beijing |
Period | 19/10/09 → 24/10/09 |
Keywords
- Dimensionality reduction
- Image and video analysis
- Tensor distance
- Tensor distance based multilinear multidimensional scaling
ASJC Scopus subject areas
- Computational Theory and Mathematics
- Computer Science Applications
- Computer Vision and Pattern Recognition
- Software