BONNs add a Gaussian-mixture kernel loss and a center-loss-style feature loss to 1-bit CNNs, reaching 59.3% ImageNet top-1 accuracy with a binary ResNet-18.
Title resolution pending
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
-
Bayesian Optimized 1-Bit CNNs
BONNs add a Gaussian-mixture kernel loss and a center-loss-style feature loss to 1-bit CNNs, reaching 59.3% ImageNet top-1 accuracy with a binary ResNet-18.