Pre-trained ImageNet classifiers encode age more strongly than race or gender in their activations, but the evidence is limited because linear probes overfit and do not generalize to unseen faces.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Assessing Intersectional Bias in Representations of Pre-Trained Image Recognition Models
Pre-trained ImageNet classifiers encode age more strongly than race or gender in their activations, but the evidence is limited because linear probes overfit and do not generalize to unseen faces.