Fig. 3: Data processing pipeline and inference architecture for super-resolution Thomson Scattering.
From: Multimodal super-resolution: discovering hidden physics and its application to fusion plasmas

a During the training, the inputs are aligned and resampled with respect to TS timing, to have a ground truth for training and validation of the model. b During the inference, the inputs are resampled to 1âMHz, which will be the resolution of synthetic SRTS. Since there is no ground truth for data-driven validation, we validate the SRTS by studying its behavior during fast physics phenomena, which are challenging to analyze with measured TS due to its low temporal resolution.