Locality preserving robust regression for jointly sparse subspace learning

摘要: As the extended version of conventional Ridge Regression, 2,1 L -norm based ridge regression learning methods have been widely used in subspace since they are more robust than Frobenius norm and meanwhile guarantee joint sparsity. However, encounter small-class problem ignore local geometric structures, which degrade their performances. To address these problems, we propose a novel method called Locality Preserving Robust Regression (LPRR). In addition to using for jointly sparse regression, also utilize capped 2L loss function further enhance robustness proposed algorithm. Moreover, make use structure information, integrate property locality preservation into our model it is great importance dimensionality reduction. The convergence analysis computational complexity iterative algorithm presented. Experimental results on four datasets indicate that LPRR performs better some famous classification tasks.