Conditional entropy between label sequences predicts task transferability and hardness without training any models, with strong correlations on CelebA, AwA2, and CUB.
Learning bounds for domain adaptation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Transferability and Hardness of Supervised Classification Tasks
Conditional entropy between label sequences predicts task transferability and hardness without training any models, with strong correlations on CelebA, AwA2, and CUB.