S3RP: Self-Supervised Super-Resolution and Prediction for Advection-Diffusion Process
This addresses a domain-specific problem in computational physics or engineering where high-resolution data is scarce, offering a novel approach but likely incremental in the broader ML context.
The paper tackles super-resolution for advection-diffusion processes without high-resolution ground-truth data by using a Recurrent Convolutional Network with physics-based regularizations, and models uncertainty with a Recurrent Wasserstein Autoencoder.
We present a super-resolution model for an advection-diffusion process with limited information. While most of the super-resolution models assume high-resolution (HR) ground-truth data in the training, in many cases such HR dataset is not readily accessible. Here, we show that a Recurrent Convolutional Network trained with physics-based regularizations is able to reconstruct the HR information without having the HR ground-truth data. Moreover, considering the ill-posed nature of a super-resolution problem, we employ the Recurrent Wasserstein Autoencoder to model the uncertainty.