Redatuming physical systems using symmetric autoencoders
This addresses the challenge of handling unmodeled nuisances in physical system observations, which is incremental as it builds on autoencoder methods with symmetry constraints.
The paper tackles the problem of disentangling coherent information from nuisance parameters in physical systems using a symmetry-informed autoencoder called SymAE, enabling redatuming to uniformize nuisances across measurements.
This paper considers physical systems described by hidden states and indirectly observed through repeated measurements corrupted by unmodeled nuisance parameters. A network-based representation learns to disentangle the coherent information (relative to the state) from the incoherent nuisance information (relative to the sensing). Instead of physical models, the representation uses symmetry and stochastic regularization to inform an autoencoder architecture called SymAE. It enables redatuming, i.e., creating virtual data instances where the nuisances are uniformized across measurements.