New paper: SPATE-GAN: Improved Generative Modeling of Dynamic Spatio-Temporal Patterns with an Autoregressive Embedding Loss

[JackKelly]

This new paper might be relevant:

SPATE-GAN: Improved Generative Modeling of Dynamic Spatio-Temporal Patterns with an Autoregressive Embedding Loss by Konstantin Klemmer, Tianlin Xu, Beatrice Acciaio, Daniel B. Neill. 30 Sept 2021.

The abstract says they:

test this new objective on a diverse set of complex spatio-temporal patterns: turbulent flows, log-Gaussian Cox processes and global weather data. We show that our novel embedding loss improves performance without any changes to the architecture of the COT-GAN backbone, highlighting our model's increased capacity for capturing autoregressive structures. We also contextualize our work with respect to recent advances in physics-informed deep learning and interdisciplinary work connecting neural networks with geographic and geophysical sciences.

(But maybe not a priority!)

[jacobbieker]

Awesome! The official PyTorch implementation is here: https://github.com/konstantinklemmer/spate-gan so we should be able to get it up and running fairly quickly. But yeah, probably a bit lower priority until we have the data pipeline more setup again