HIGH DIMENSIONALITY REDUCTION ON GRAPHICAL DATA

Smita J.Khelukar, Manisha.M.Naoghare

Abstract: In spite of the fact that graph embedding has been an intense instrument for displaying data natural structures, just utilizing all elements for data structures revelation may bring about noise amplification. This is especially serious for high dimensional data with little examples. To meet this test, a novel effective structure to perform highlight determination for graph embedding, in which a classification of graph implanting routines is given a role as a slightest squares relapse issue. In this structure, a twofold component selector is acquainted with normally handle the component cardinality at all squares detailing. The proposed strategy is quick and memory proficient. The proposed system is connected to a few graph embedding learning issues, counting administered, unsupervised and semi supervised graph embedding.

Keywords: Efficient feature selection, High dimensional data, Sparse graph embedding, Sparse principal component analysis, Subproblem Optimization.

DOI: https://doi.org/10.15623/ijret.2015.0411029