Margin based feature selection - theory and algorithms

By Ran Gilad-Bachrach, Amir Navot, Naftali Tishby. International Conference on Machine Learning, 2004

453 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/894967541ba49c416e81b426770d612cb1d4d135

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Novel Speech Signal Processing Algorithms for High-Accuracy Classification of Parkinson's Disease (2012, methodology): “well, in practice, in this kind of application because, internally, it incorporates a (nonlinear, nearest-neighbor) classifier [30].”
Searching for Interacting Features (2007, background): “Many features can be removed without performance deterioration [Gilad-Bachrach et al., 2004].”
Automated taxonomic classification of phytoplankton sampled with imaging‐in‐flow cytometry (2007, methodology): “We have used the Greedy Feature Flip Algorithm (G-flip) as described by Gilad-Bachrach et al. (2004b) and available in a MATLAB implementation (Gilad-Bachrach et al. 2004a).”
Machine Learning in Major Depression: from Classification to Treatment Outcome Prediction. (2018, methodology): “Besides, some proposed approaches 24,25 provide an intermediate solution by adopting geometric distance in feature space between different groups in the training data.”
Large-margin nearest neighbor classifiers via sample weight learning (2011, background): “In addition, Roberto et al. developed a prototype weighting algorithm derived by approximately minimizing the leaving-one-out classification error of the given training set [20].”
Nightmare at test time: robust learning by feature deletion (2006, methodology): “The problem may be approximated via difierent relaxations as in ( Gilad-Bachrach et al., 2004; Weston et al., 2000).”
Discriminative prototype selection methods for graph embedding (2013, unknown): “[38,39]).”
Sensor Positioning for Activity Recognition Using Wearable Accelerometers (2011, methodology): “2) Simba Feature Selection: The iterative search marginbased algorithm (Simba) for feature selection [37] is similar to RELIEF in terms of updating feature weights to provide maximum margins.”
A Survey of Text Classification Algorithms (2012, methodology): “Therefore, a number of techniques were proposed in [134] in order to learn the associations between the words and the categories.”
Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications (2007, background): “E-mail: sun@dsp.ufl.edu .”
A Hybrid Intelligent System Framework for the Prediction of Heart Disease Using Machine Learning Algorithms (2018, methodology): “Relief uses the same techniques as in K-NN that determines the weights of features (see Algorithm 1) [22].”
Class Conditional Nearest Neighbor and Large Margin Instance Selection (2008, methodology): “Margins play an important role in machine learning research, as a tool for performing theoretic analysis [5], [34], for developing new machine learning algorithms [10], and for improving the accuracy performance of nearest neighbor based classiﬁers [12], [20].”
Neighborhood Component Feature Selection for High-Dimensional Data (2012, methodology): “Recently, several nearest neighbor-based feature weighting methods, including RELIEF [7], Simba [8], RGS [9], IRELIEF [10], LMFW [11], Lmba [12] and FSSun [13], have been successfully developed and shown the better performance on high-dimensional data analysis.”
Searching for interacting features in subset selection (2009, unknown): “…features using normalized mutual information) 4 [21], CFS (selecting features using correlation based measure and Best-First search [20,32]), ReliefF (selecting features by hypothesis maximization [14,33]), and FOCUS (selecting features using consistency based measure and exhaustive search) [18].”