Sparse Dict Learning loss

• $$L(X) = n^{-1}\Sigma_i^n ||x_i - Dr_i ||^2 + \lambda \Sigma_i |r_i|$
• $\lambda {\Sigma }_i|r_i|$ is Lasso/L1 Lp Regularization
• Predictions : $r=argmi{n}_r||x-D{r}_i|{|}^2+\lambda {\Sigma }_i|r_i|$