Abstract—In this correspondence, we proposed an intelligent faces classification method for the robust face recognition. All the face Recognition algorithms introduced till yet are computationally very expensive and situation become more critical when search space is large (Large Population). Proposed technique reduced the search space instead of feature space in the off line mode by classifying the data set using self organizing maps unsupervised neural networks and then face recognition algorithm is applied on reduced search space. More than 6X104 Histogram equalized (HE) front face cropped images are used for experiments. HE enhances the facial features and reduces the illumination effects. As for recognition any robust algorithm can be used on the reduced search space so recognition rate for that used algorithm will be considered. For proving the Histogram Equalized Intelligent Classification using Self Organized Maps (HEIC SOM) robust in throughput traditionally used PCA based face recognition algorithm is used and about 98% accuracy in terms of classification is achieved with improved efficiency of about 24 times in execution timings (for recognition ). As much as the data set would be large this efficiency in time will be increased. Error in terms of false classification is discussed in results.