Research On Dimensionality Reduction Methods Based On Random Projection

Dimensionality reduction techniques play important roles in the analysis of big data.Traditional dimensionality reduction approaches,such as Principle Component Analysis(PCA)and Linear Discriminant Analysis(LDA),have been studied extensively in the past few decades.However,as the dimension of huge data increases,the computational cost of traditional dimensionality reduction approaches grows dramatically and becomes prohibitive.It has also triggered the development of Random Projection(RP)technique which maps high-dimensional data onto low-dimensional subspace within short time.However,RP generates transformation matrix without considering intrinsic structure of original data and usually leads to relatively high distortion.Therefore,in the past few years,some approaches based on RP have been proposed to address this problem.We summarized these approaches in different applications to help practitioners to employ proper approaches in their specific applications.Also,we enumerated their benefits and limitations to provide further references for researchers to develop novel RP-based approaches.Experimental results have proved that feature extraction methods including LDA,Bag of Words and other application-specified methods can significantly improve the performance of RP.Fast increase of big genomics data leads to burdensome computation and cannot meet the need of real-time analysis for massive data.We proposed several methods,where traditional dimension reduction methods including PCA,LDA,and feature selection(FS)were incorporated into RP to improve the performance of RP.We compared classification accuracy and running time of the proposed methods on three gene expression profiles.With the surge of gene expression profiles,physicians can make more accurate diagnosis.Past studies have revealed that similar treatment can be applied to patients whose expression patterns are similar.Here,we developed a light-weighted and cross-platform web application so as to find similar patients easily.In this web application,several machine learning algorithms are adopted.In addition,dimensionality reduction methods based on RP are applied to gene expression profiles to reduce the computational complexity.