Regressing CNN image features to random pseudo-targets during fine-tuning acts as an effective regularizer, improving transfer learning accuracy on par with concrete auxiliary-task methods.
Factors of transferability for a generic convnet representation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Regularizing CNN Transfer Learning with Randomised Regression
Regressing CNN image features to random pseudo-targets during fine-tuning acts as an effective regularizer, improving transfer learning accuracy on par with concrete auxiliary-task methods.