A Fast Iterative Algorithm for Improved Unsupervised Feature Selection

Ordozgoiti, Bruno; Canaval, Sandra Gómez; Mozó, Alberto

doi:10.1109/icdm.2016.0050

Cited by 12 publications

(12 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, its performance was quite consistent over different disease types in achieving a high ROC-AUC score with a low FN rate, as shown in Figure 2 . We also benchmarked four other methods twoPhase [ 3 ]; iterFS [ 4 ]; Barabasi [ 9 ]; and LIMMA . We found that the performance of these algorithms was inconsistent, and that the performance degraded—especially when a small number of features were selected ( Figure 2 ).…”

Section: Resultsmentioning

confidence: 99%

“…We also benchmarked our results with two other feature selection methods, twoPhase [ 3 ], iterFS [ 4 ]; one geneset selection method Barabasi [ 9 ]; and one DE analysis method, LIMMA . For twoPhase, iterFS, and Barabasi we follow the same preprocessing as our method to select the subset of features.…”

Section: Methodsmentioning

confidence: 99%

“…Several of these feature selection techniques have been used in the past for different purposes. For example, twoPhase [ 3 ]; iterFS [ 4 ]; WeiBi [ 5 ] are based on information gain, LASSO, or Fisher score. Feature selection algorithms in the biological domain were briefly discussed in [ 2 , 6 ].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Relevant and Non-Redundant Feature Selection for Cancer Classification and Subtype Detection

Rana

Thai

Dinh

et al. 2021

Cancers

View full text Add to dashboard Cite

Biologists seek to identify a small number of significant features that are important, non-redundant, and relevant from diverse omics data. For example, statistical methods such as LIMMA and DEseq distinguish differentially expressed genes between a case and control group from the transcript profile. Researchers also apply various column subset selection algorithms on genomics datasets for a similar purpose. Unfortunately, genes selected by such statistical or machine learning methods are often highly co-regulated, making their performance inconsistent. Here, we introduce a novel feature selection algorithm that selects highly disease-related and non-redundant features from a diverse set of omics datasets. We successfully applied this algorithm to three different biological problems: (a) disease-to-normal sample classification; (b) multiclass classification of different disease samples; and (c) disease subtypes detection. Considering the classification of ROC-AUC, false-positive, and false-negative rates, our algorithm outperformed other gene selection and differential expression (DE) methods for all six types of cancer datasets from TCGA considered here for binary and multiclass classification problems. Moreover, genes picked by our algorithm improved the disease subtyping accuracy for four different cancer types over state-of-the-art methods. Hence, we posit that our proposed feature reduction method can support the community to solve various problems, including the selection of disease-specific biomarkers, precision medicine design, and disease sub-type detection.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Relevant and Non-Redundant Feature Selection for Cancer Classification and Subtype Detection

Rana

Thai

Dinh

et al. 2021

Cancers

View full text Add to dashboard Cite

show abstract

“…To investigate its theoretical performance, Bhaskara et al (2016) studied the equivalent problem that maximizes SS + A 2 F , and proved an approximation ratio of (1 − ) w.r.t. the optimal function value Ordozgoiti et al (2016) proposed a simple local search algorithm, which starts from k randomly selected columns, and iteratively replaces one selected column with the best among n − k unselected columns, until no improvement can be yielded. This algorithm has been shown empirically to outperform other state-of-the-art methods, and an approximation bound w.r.t.…”

Section: Related Workmentioning

confidence: 99%

Unsupervised Feature Selection by Pareto Optimization

Feng

Qian

Tang

2019

AAAI

View full text Add to dashboard Cite

Dimensionality reduction is often employed to deal with the data with a huge number of features, which can be generally divided into two categories: feature transformation and feature selection. Due to the interpretability, the efficiency during inference and the abundance of unlabeled data, unsupervised feature selection has attracted much attention. In this paper, we consider its natural formulation, column subset selection (CSS), which is to minimize the reconstruction error of a data matrix by selecting a subset of features. We propose an anytime randomized iterative approach POCSS, which minimizes the reconstruction error and the number of selected features simultaneously. Its approximation guarantee is well bounded. Empirical results exhibit the superior performance of POCSS over the state-of-the-art algorithms.

show abstract

“…On one hand, not much is known about the hardness of approximation of column subset selection problems in general, as tight examples exist only for best rank-k subspaces. On the other, traditional combinatorial algorithms are generally much more convenient for practical implementation and exhibit good empirical performance [28,27,62]. Furthermore, recent works provide approximation guarantees for methods in this category [1,16,63].…”

Section: Heuristicsmentioning

confidence: 99%

Column subset selection in practice : efficient heuristics and regularization

Rubio¹

View full text Add to dashboard Cite

A Fast Iterative Algorithm for Improved Unsupervised Feature Selection

Cited by 12 publications

References 29 publications

Relevant and Non-Redundant Feature Selection for Cancer Classification and Subtype Detection

Relevant and Non-Redundant Feature Selection for Cancer Classification and Subtype Detection

Unsupervised Feature Selection by Pareto Optimization

Column subset selection in practice : efficient heuristics and regularization

Contact Info

Product

Resources

About