id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_y6p3hegtendnvedua5ze75kcie	John W. Chinneck	Integrated classifier hyperplane placement and feature selection	2012	25	.pdf	application/pdf	10003	1165	70	data points is equivalent to the following problem: given an infeasible set of linear inequalities, inequalities for the data points, the complete set of constraints integrates both hyperplane Algorithm 2: Adding features before hyperplane placement (AddB). Algorithm 4: Selecting features by adding them after a separation is available (AddA). algorithm that includes the feature constraints directly in Alg. 1 along with the data point Algorithm 5: Selecting features by removing them after a linear separation is available (DelA). data sets have more than 10 features: musk1, sonar, vehicle, wdbc, wine, and wpbc. Three new algorithms select features using the SINF-based filter prior to placing the separating and number of features shown relative to the 10-fold test set results for the Orig algorithm. The Int algorithm integrates hyperplane placement and feature selection, as shown in Alg. 6.  New methods for selecting features after a separating hyperplane has already been 	./cache/work_y6p3hegtendnvedua5ze75kcie.pdf	./txt/work_y6p3hegtendnvedua5ze75kcie.txt