117 - Learning Fast Converging, Effective Conditional Generative Adversarial Networks with a Mirror