Abstract
Graph-based semi-supervised classification heavily depends on a well-structured graph. In this paper, we investigate a mixture graph and propose a method called semi-supervised classification based on mixture graph (SSCMG). SSCMG first constructs multiple k nearest neighborhood (kNN) graphs in different random subspaces of the samples. Then, it combines these graphs into a mixture graph and incorporates this graph into a graph-based semi-supervised classifier. SSCMG can preserve the local structure of samples in subspaces and is less affected by noisy and redundant features. Empirical study on facial images classification shows that SSCMG not only has better recognition performance, but also is more robust to input parameters than other related methods.
Highlights
Many real-world pattern classification and data mining applications are often confronted with a typical problem that is lacking of sufficient labeled data, since labeling samples usually requires domain-specific experts [1,2]
In this paper, motivated by these observations, we investigate a mixture graph combined by multiple k nearest neighborhood (kNN) graphs constructed in the subspaces, and study a semi-supervised classification method on this mixture graph ((semi-supervised classification based on mixture graph SSCMG))
The main difference between SSCMG and aforementioned Gaussian fields and harmonic functions (GFHF), local and global consistence (LGC), linear neighborhood propagation (LNP), robust multi-class graph transductive (RMGT) and compact graph based semi-supervised learning (CGSSL) is that SSCMG uses a mixture graph combined by multiple kNN graphs constructed in the subspaces, whereas the other methods just utilize a single graph alone
Summary
Many real-world pattern classification and data mining applications are often confronted with a typical problem that is lacking of sufficient labeled data, since labeling samples usually requires domain-specific experts [1,2]. The main difference between SSCMG and aforementioned GFHF, LGC, LNP, RMGT and CGSSL is that SSCMG uses a mixture graph combined by multiple kNN graphs constructed in the subspaces, whereas the other methods just utilize (or optimize) a single graph alone. The main contributions of this paper are summarized as follows: (1) A semi-supervised classification based on mixture graph (SSCMG) method is proposed for high-dimensional data classification; (2) Extensive experimental study and analysis demonstrate that SSCMG achieves higher accuracy and it can be more robust to input parameters than other related methods; (3) Mixture graph works better than the graph optimized by a single kNN graph alone, and it can be used as a good alternative graph for GSSC methods on high dimensional data.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call