Robust One-Class Kernel Spectral Regression

The kernel null-space technique is known to be an effective one-class classification (OCC) technique. Nevertheless, the applicability of this method limited due its susceptibility possible training data corruption and inability rank observations according their conformity with model. This article addresses these shortcomings by regularizing solution Fisher methodology in context regression-based formulation. In respect, first, effect Tikhonov regularization Hilbert space analyzed, where learning problem presence contamination set posed as a sensitivity analysis problem. Next, sparsity studied. For both alternative schemes, iterative algorithms are proposed which recursively update label confidences. Through extensive experiments, found enhance robustness against compared baseline method, well other existing approaches OCC paradigm, while providing functionality samples effectively.