Theoretical Insights into the Use of Structural Similarity Index In Generative Models and Inferential Autoencoders
This work addresses image quality improvement for generative modeling applications, but it appears incremental as it focuses on theoretical insights and adaptations of existing methods.
The paper tackles the problem of generating perceptually better images in generative models and inferential autoencoders by theoretically discussing the use of Structural Similarity Index (SSIM) instead of the ℓ₂ norm, showing that the SSIM kernel is universal and proposing its application in various models like GANs and autoencoders.
Generative models and inferential autoencoders mostly make use of $\ell_2$ norm in their optimization objectives. In order to generate perceptually better images, this short paper theoretically discusses how to use Structural Similarity Index (SSIM) in generative models and inferential autoencoders. We first review SSIM, SSIM distance metrics, and SSIM kernel. We show that the SSIM kernel is a universal kernel and thus can be used in unconditional and conditional generated moment matching networks. Then, we explain how to use SSIM distance in variational and adversarial autoencoders and unconditional and conditional Generative Adversarial Networks (GANs). Finally, we propose to use SSIM distance rather than $\ell_2$ norm in least squares GAN.