Facial retouching detection using Subclass Restricted Boltzmann Machines

Our project is to build and explore subclass Restricted Boltzmann Machines (subclass RBMs), a combination of Subclass Discriminant Analysis (SDA) and RBMs - extending the approach in [1]. To demonstrate the subclass RBM, we plan to use it for facial retouching detection problem, where we classify a given image of a face as unaltered or retouched. This is an important problem in face recognition domain since retouched photos can affect face recognition and thus, may pose a threat to security.