Analyzing the discrepancy principle for kernelized spectral filter learning algorithms

Open in new window