Adversarial autoencoders and adversarial LSTM for improved forecasts of urban air pollution simulations
This work addresses forecasting accuracy for urban air pollution simulations, which is an incremental improvement in a domain-specific application.
This paper tackled the problem of improving forecasts for computational fluid dynamics simulations of urban air pollution by using adversarial training with a PCA-based adversarial autoencoder and adversarial LSTM networks, resulting in the adversarially trained LSTM outperforming a classically trained LSTM.
This paper presents an approach to improve the forecast of computational fluid dynamics (CFD) simulations of urban air pollution using deep learning, and most specifically adversarial training. This adversarial approach aims to reduce the divergence of the forecasts from the underlying physical model. Our two-step method integrates a Principal Components Analysis (PCA) based adversarial autoencoder (PC-AAE) with adversarial Long short-term memory (LSTM) networks. Once the reduced-order model (ROM) of the CFD solution is obtained via PCA, an adversarial autoencoder is used on the principal components time series. Subsequentially, a Long Short-Term Memory network (LSTM) is adversarially trained on the latent space produced by the PC-AAE to make forecasts. Once trained, the adversarially trained LSTM outperforms a LSTM trained in a classical way. The study area is in South London, including three-dimensional velocity vectors in a busy traffic junction.