In unconditional image generators, measured attribute bias shifts are small and are strongly influenced by whether the attribute classifier's decision boundary falls in a dense or sparse region of the attribute's distribution.
Gritsenko, William Chan, Mohammad Norouzi, and David J
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Bias Analysis in Unconditional Image Generative Models
In unconditional image generators, measured attribute bias shifts are small and are strongly influenced by whether the attribute classifier's decision boundary falls in a dense or sparse region of the attribute's distribution.