Improvements in the Large p, Small n Classification Issue

Abstract

Classifying gene expression data is known to contain keys for solving the fundamental problems in cancer studies. However, this issue is a complex task because of the large p, small n issue on gene expression data analysis. In this paper, we propose the improvements in the large p, small n classification issue for the study of human cancer. First, a new enhancing sample size method with generative adversarial network is proposed to improve classification algorithms. Second, we suggest a classification approach with over-sampling technique using features extracted by deep convolutional neural network. Numerical test results on fifty very-high-dimensional and low-sample-size gene expression data datasets from the Kent Ridge Biomedical and Array Expression repositories illustrate that the proposed models are more accurate than state-of-the-art classifying models. In addition, we also have explored the performance of support vector machines, k nearest neighbors and random forests, which have improved when apply our approaches.

Novel feature selection method via kernel tensor decomposition for improved multi-omics data analysis

“…Among the numerous proposed methods adapted to multi-omics data analysis [ 1 , 2 ], only few are capable of performing feature selection. Most of these methods fail to implement feature selection because multi-omics data analysis has a strong tendency to involve a small number ( ) of samples with a large number ( ) of features, commonly referred to as the large p small n problem [ 3 ], posing difficulty for accurate feature selection. Features should have a sufficiently small P value to be selected under the null hypothesis.…”

A Novel Algorithm for Feature Selection Using Penalized Regression with Applications to Single-Cell RNA Sequencing Data

“…When , we call the data having the “large-p-small-n” problem. Statistical models could result in poor prediction performance due to over-fitting when training data contain fewer samples compared to the number of features [ 6 ]. There are several methods to deal with the “large-p-small-n” problem in Machine Learning (ML), of which feature selection is the most useful.…”

Novel feature selection method via kernel tensor decomposition for improved multi-omics data analysis

“…Feature selection with multi-omics datasets has always been difficult; numerous proposed methods adapted to multi-omics data analysis (Reel et al ., 2021; Subramanian et al ., 2020) include a limited number of methods that perform feature selection. Most methods fail to implement feature selection because multi-omics data analysis has a strong tendency to have a large p small n problem (Huynh et al ., 2020) to which feature selection is difficult. Large p small n indicates the situation where there is only a small number (= n ) of samples with a large number (= p ) of features.…”

“…Among the numerous proposed methods adapted to multi-omics data analysis [1,2], only few are capable of performing feature selection. Most of these methods fail to implement feature selection because multi-omics data analysis has a strong tendency to involve a small number (= n) of samples with a large number (= p) of features, commonly referred to as the large p small n problem [3], posing difficulty for accurate feature selection. Features should have a sufficiently small P -value to be selected under the null hypothesis.…”