The paper introduces a maxitive Donsker-Varadhan formulation for possibilistic variational inference, deriving learning rules and CBOpt optimizers that achieve competitive performance on image classification.
Olivier Catoni
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Maxitive Donsker-Varadhan Formulation for Possibilistic Variational Inference
The paper introduces a maxitive Donsker-Varadhan formulation for possibilistic variational inference, deriving learning rules and CBOpt optimizers that achieve competitive performance on image classification.