Improving Reconstruction Fidelity and Robustness in Physics-Integrated Generative Modeling using Attentive Planar Normalizing Flow based Variational Autoencoder
The core message of this work is to improve the fidelity of reconstruction and robustness to noise in physics-integrated generative modeling by using a variational autoencoder with planar normalizing flow based latent posterior distribution and an attention-based encoder architecture.