CircleGAN: Generative Adversarial Learning across Spherical Circles
This work addresses the problem of improving sample realness and diversity in GANs, which is a common challenge for researchers and practitioners working with generative models.
This paper introduces a new GAN discriminator that uses spherical circles to embed samples, improving both the realness and diversity of generated outputs. By populating realistic samples around a great circle and pushing unrealistic ones towards the poles, the method achieves state-of-the-art results on standard unconditional and conditional generation benchmarks.
We present a novel discriminator for GANs that improves realness and diversity of generated samples by learning a structured hypersphere embedding space using spherical circles. The proposed discriminator learns to populate realistic samples around the longest spherical circle, i.e., a great circle, while pushing unrealistic samples toward the poles perpendicular to the great circle. Since longer circles occupy larger area on the hypersphere, they encourage more diversity in representation learning, and vice versa. Discriminating samples based on their corresponding spherical circles can thus naturally induce diversity to generated samples. We also extend the proposed method for conditional settings with class labels by creating a hypersphere for each category and performing class-wise discrimination and update. In experiments, we validate the effectiveness for both unconditional and conditional generation on standard benchmarks, achieving the state of the art.