Coupled generative adversarial stacked Auto-encoder: CoGASA

Coupled Generative Adversarial Network (CoGAN) was recently introduced in order to model a joint distribution of a multi modal dataset. The CoGAN model lacks the capability to handle noisy data as well as it is computationally expensive and inefficient for practical applications such as cross-domain image transformation. In this paper, we propose a new method, named the Coupled Generative Adversarial Stacked Auto-encoder (CoGASA), to directly transfer data from one domain to another domain with robustness to noise in the input data as well to as reduce the computation time. We evaluate the proposed model using MNIST and the Large-scale CelebFaces Attributes (CelebA) datasets, and the results demonstrate a highly competitive performance. Our proposed models can easily transfer images into the target domain with minimal effort.