Face recognition is the very significant field in pattern recognition area. It has multiple applications in military and finance, to name a few. In this paper, the combination of the sparse PCA with the nearest-neighbor method (and with the kernel ridge regression method) will be proposed and will be applied to solve the face recognition problem. Experimental results illustrate that the accuracy of the combination of the sparse PCA method (using the proximal gradient method and the FISTA method) and one specific classification system may be lower than the accuracy of the combination of the PCA method and one specific classification system but sometimes the combination of the sparse PCA method (using the proximal gradient method or the FISTA method) and one specific classification system leads to better accuracy. Moreover, we recognize that the process computing the sparse PCA algorithm using the FISTA method is always faster than the process computing the sparse PCA algorithm using the proximal gradient method.