RENT—Repeated Elastic Net Technique for Feature Selection

Jenul, Anna; Schrunner, Stefan; Liland, Kristian Hovde; Indahl, Ulf Geir; Futsæther, Cecilia Marie; Tomić, Oliver

doi:10.1109/access.2021.3126429

Cited by 17 publications

(22 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The original RENT [ 17 ] exploited three statistical characteristics of the covariate coefficients fitted by elastic net for feature selection: (i) the empirical probability of the weights being non-zero; (ii) the rate of the weights having stable signs (positive/negative) across resampled runs; and (iii) the statistical significance obtained using the Student’s t -test to reject the null hypothesis of weight = 0. Motivated by the well-established agreement in the literature that ensemble models generally have better representation power than singular models [ 27 ], we propose the bagged-boosted RENT (BB-RENT) model, which further improves the feature selection stability of RENT based on the combination of the bagging and boosting ensemble techniques, as summarized below.…”

Section: Methodsmentioning

confidence: 99%

“…In this study, we boosted a maximum of N elastic nets with boosting coefficients fitted by AdaBoost [ 28 ]. Radiomic features with elastic net coefficients weighted by boosting coefficients that matched the abovementioned three criteria [ 17 ] after fitting the K boosted elastic nets were included.…”

Section: Methodsmentioning

confidence: 99%

“…In light of this, methods to improve feature selection stability for medical imaging have rapidly gained attention. Several recent publications have highlighted the use of ensemble methods to improve the feature selection stability [ 16 , 17 , 18 , 19 ], mainly investigating three ensemble techniques, namely resampling, bagging, and boosting; however, their combined use has not been studied in radiomics. It is possible that using these techniques in combination could further improve the stability.…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, the purpose of this study was to investigate the role of radiomics analysis in discrimination between early-stage T1 NPC and BH on MRI. We built a radiomics model for discrimination using features selected using the technique proposed in this study, which combines bagging, boosting, and the repeated elastic net technique (RENT) [ 17 ], which we call bagged-boosted RENT (BB-RENT). Using radiomic features extracted from the non-contrast-enhanced, T2-weighted, fat-suppressed (T2w-fs) MRI, the model was trained using a patient cohort scanned with a 3 T MRI scanner (3 T MRI training cohort) with fivefold cross validation and tested on a patient cohort scanned with 1.5 T MRI scanners (1.5 T MRI testing cohort).…”

Section: Introductionmentioning

confidence: 99%

“…Figure 3. Illustration of the original repeated elastic net technique (RENT), a technique previously reported in[17], and the proposed bagged-boosted RENT (BB-RENT) technique based on it. K, N, Q, and η are hyperparameters of the algorithm, where K is the number of times the elastic net or the boosted elastic net is repeated, N decides the maximum number of weak learners (depth) used when boosting the elastic nets, Q decides the number of times the boosted RENT is bootstrapped and then aggregated, and η is the percentage threshold for including a feature in the final selected subset.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Radiomics for Discrimination between Early-Stage Nasopharyngeal Carcinoma and Benign Hyperplasia with Stable Feature Selection on MRI

Wong

Zhang

et al. 2022

Cancers

View full text Add to dashboard Cite

Discriminating early-stage nasopharyngeal carcinoma (NPC) from benign hyperplasia (BH) on MRI is a challenging but important task for the early detection of NPC in screening programs. Radiomics models have the potential to meet this challenge, but instability in the feature selection step may reduce their reliability. Therefore, in this study, we aim to discriminate between early-stage T1 NPC and BH on MRI using radiomics and propose a method to improve the stability of the feature selection step in the radiomics pipeline. A radiomics model was trained using data from 442 patients (221 early-stage T1 NPC and 221 with BH) scanned at 3T and tested on 213 patients (99 early-stage T1 NPC and 114 BH) scanned at 1.5T. To verify the improvement in feature selection stability, we compared our proposed ensemble technique, which uses a combination of bagging and boosting (BB-RENT), with the well-established elastic net. The proposed radiomics model achieved an area under the curve of 0.85 (95% confidence interval (CI): 0.82–0.89) and 0.80 (95% CI: 0.74–0.86) in discriminating NPC and BH in the 3T training and 1.5T testing cohort, respectively, using 17 features selected from a pool of 422 features by the proposed feature selection technique. BB-RENT showed a better feature selection stability compared to the elastic net (Jaccard index = 0.39 ± 0.14 and 0.24 ± 0.06, respectively; p < 0.001).

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Radiomics for Discrimination between Early-Stage Nasopharyngeal Carcinoma and Benign Hyperplasia with Stable Feature Selection on MRI

Wong

Zhang

et al. 2022

Cancers

View full text Add to dashboard Cite

show abstract

Iterative re‐weighted covariates selection for robust feature selection modelling in the presence of outliers (irCovSel)

Mishra

Liland

2022

Journal of Chemometrics

Self Cite

View full text Add to dashboard Cite

A new method for feature selective modelling in the presence of outliers is presented. The method is a combination of iterative re-weighted partial least squares and the covariates selection approach. The method relies on iterative down-weighting of the outlying samples prior to estimating the squared covariance for covariates selection. In this way, the outlying samples carrying low weights have minimal influence on the squared covariance estimation, while the inliers have the maximum influence on the squared covariance estimation.The method allows selecting robust features, and models based on such features in general perform better in terms of prediction accuracy (lower error) than selecting features using equal sample weights for all samples. The algorithm description and tests of the method in single and multiple response scenarios are presented. Method performance is also demonstrated on a real spectral data set.

show abstract