Fig. 3From: Stochastic convex sparse principal component analysisRegularization path for Cvx-SPCA. We checked whether the known principal component can be recovered through the path to be able confirm that this is a valid regularization path. When regularization term was around −0.11 (dashed line) in logarithmic scale, we could exactly recover the non-zero loading values of the known principal component which was used to generate the dataBack to article page