The cosmic microwave background holds multiple clues to the development of the universe. Because of its cosmic nature, it is almost impossible for researchers to access this information by conventional means. However, with the help of a generative adversarial network (GAN) augmented with a deep learning approach, they could finally change this. A generative adversarial network is a rule-based, machine-learning technique that uses opportunities in data and fantasy worlds. The author of this research implements GAN for the cosmic microwave background (CMB) separation problem. They report the results by leveraging machine learning premised on GAN. To validate the robustness of our network, they perform several tests against different foreground models by increasing the amplitude of each component.