
Decoder decomposition is a post-processing method that can aid the interpretability of nonlinear autoencoder models by connecting the latent variables to the coherent structures of the flow.


coremsg

interpreting-the-latent-space-of-nonlinear-autoencoders-using-decoder-decomposition-and-wind-tunnel-experimental-data


Interpreting the Latent Space of Nonlinear Autoencoders Using Decoder Decomposition and Wind-Tunnel Experimental Data
