Design and Evaluation of GAN based Regression Model

Generative Adversarial Networks (GANs) are capable of generating realistic photos of objects, scenes and people that do not exist in real life. This is made possible due to the successful ability of GANs in modeling high dimensional data, handling missing data, providing multi-modal outputs and multi plausible answers. These positive features and capabilities of GANs have spearheaded research in the area of visual modeling using GAN. In this paper, an attempt is made to design a GAN model for solving regression problems. In order to assess the performance evaluation of proposed GAN model for regression problem, four basic functions and seven datasets from standard repositories are employed. It is observed that the proposed GAN model gave satisfactory results and can be employed for various other regression problems too.