Face is one of the most widely used biometric in security systems. Despite its wide usage, face recognition is not a fully solved problem due to the challenges associated with varying illumination conditions and pose. In this paper, we address the problem of face recognition under non-uniform illumination using deep convolutional neural networks (CNN). The ability of a CNN to learn local patterns from data is used for facial recognition. The symmetry of facial information is exploited to improve the performance of the system by considering the horizontal reflections of the facial images. Experiments conducted on Yale facial image dataset demonstrate the efficacy of the proposed approach. © 2015 IEEE.