Transcription of L2,1-Norm Regularized Discriminative Feature Selection …
{{id}} {{{paragraph}}}
Example: barber
L2,1-Norm Regularized Discriminative Feature Selection …
2,1- norm Regularized Discriminative FeatureSelection for Unsupervised LearningYi Yang1, Heng Tao Shen1, Zhigang Ma2, Zi Huang1, Xiaofang Zhou11 School of Information Technology & ElectricalEngineering, The University of of Information Engineering & Computer Science, University of with supervised learning forfeature Selection , it is much more difficultto select the Discriminative features in un-supervised learning due to the lack oflabel information. Traditional unsuper-vised Feature Selection algorithms usuallyselect the features which best preservethe data distribution, , manifold struc-ture, of the whole Feature set. Under theassumption that the class label of inputdata can be predicted by a linear classi-fier, we incorporate Discriminative anal-ysis and 2,1- norm minimization into ajoint framework for unsupervised featureselection. Different from existing unsu-pervised Feature Selection algorithms, ouralgorithm selects the most discriminativefeature subset from the whole Feature setin batch mode.
rithms, e.g., Fisher score [Duda et al., 2001] , robust regres-sion [Nie et al., 2010], sparse multi-output regression [Zhao et al., 2010] and trace ratio [Nie et al., 2008], usually select featuresaccordingto labels of the training data. Because dis-criminative informationis enclosed in labels, supervised fea-
Tags:
Information
Domain:
Source:
Link to this page:
Please notify us if you found a problem with this document: