Variable Selection Using SVM-based Criteria

By Alain Rakotomamonjy. 2003

747 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Benchmark and Survey of Automated Machine Learning Frameworks (2019, methodology): “Many ML models provide some sort of feature ranking that can be utilized, e.g., SVMs (Guyon et al., 2002; Rakotomamonjy, 2003), perceptrons (Mej́ıa-Lavalle et al., 2006) or random forests (Tuv et al., 2009).”
COMPARE: Classification of Morphological Patterns Using Adaptive Regional Elements (2007, methodology): “initially proposed for a cancer classification problem [29], and was later extended by introducing SVM-based leave-one-out error bound criteria in [30].”
Correlation and variable importance in random forests (2013, unknown): “[15], Rakotomamonjy [31], Svetnik et al.”
Feature selection via sensitivity analysis of SVM probabilistic outputs (2008, methodology): “2.2 Past work in SVM feature selection Several feature-selection methods for SVM have been proposed in the literature (Bradley and Mangasarian 1998; Weston et al. 2001; Guyon et al. 2002; Rakotomamonjy 2003; Neumann et al. 2005).”
Broad Learning System: An Effective and Efficient Incremental Learning System Without the Need for Deep Architecture (2018, background): “Common approaches, which have the advantage of easy implementation and outstanding efficiency, for feature selection include variable ranking [28], feature subset selection [29], penalized least squares [30], random feature extractions such as nonadaptive random projections [31] and random forest [3”
MIXED-INTEGER SUPPORT VECTOR MACHINE (2009, background): “…of the i th data point, and y i ∈ {− 1 , +1 } is the corresponding label, we can formulate SVM learning into the following optimization problem: where L { f ( x ) , y } = [1 − yf ( x )] + is the hinge loss, f ( x ) = wx − γ is the decision function, w is the weight vector, γ is the bias term.”
A Survey on Malware Detection Using Data Mining Techniques (2017, methodology): “…some embedded methods perform more efﬁciently than wrapper methods by directly optimizing an objective function, often deﬁned by two or more parameters – one to encourage the goodness-of-ﬁt and the other to penalize for a large number of variables [Rakotomamonjy 2003; Perkins et al. 2003].”
Unsupervised Feature Selection Using Nonnegative Spectral Analysis (2012, background): “In the wrapper approaches (Guyon and Elisseeff 2003; Rakotomamonjy 2003), feature selection is “wrapped” in a learning algorithm and the classiﬁcation performance of features is taken as the evaluation criterion.”
Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images (2015, unknown): “By allowing feature interactions, RFE-SVM (39, 40) explicitly models the correlations between features, thereby leading to robust classifier performance.”
Combining multivariate voxel selection and support vector machines for mapping and classification of fMRI spatial patterns (2008, unknown): “Recursive Feature Elimination (RFE) has been compared to other multivariate feature selection strategies (Rakotomamonjy, 2003) and has been successfully applied to gene selection and sample classification in combination with Support Vector Machine classifiers (Guyon et al.”
PyMVPA: a Python Toolbox for Multivariate Pattern Analysis of fMRI Data (2009, background): “Feature sensitivity is then expressed in terms of the difference between computed mea - sures with and without noise added to a feature ( see Rakotomamonjy , 2003 ; Hanson et al .”
Feature selection for high-dimensional class-imbalanced data sets using Support Vector Machines (2014, methodology): “For nonlinear SVM classifiers with Gaussian kernel we used C = 1000 and σ = 3, as suggested in Rakotomamonjy [31].”