A filter-smoothness regularizer makes CNN latent codes from image frames behave like samples of a continuous trajectory, so neural ODEs fit them and predict future frames better.
Rezende, Andrew Jaegle, S ´ebastien Racani `ere, Aleksandar Botev, and Irina Higgins
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Continuity-Preserving Convolutional Autoencoders for Learning Continuous Latent Dynamical Models from Images
A filter-smoothness regularizer makes CNN latent codes from image frames behave like samples of a continuous trajectory, so neural ODEs fit them and predict future frames better.