Two-stage and gradually decreasing quantization, stochastic precision sampling, and joint teacher-student distillation each improve low-bit CNN accuracy on ImageNet and CIFAR-100, with the largest gains when combined.
Pruning filters for efficient convnets,
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
-
Effective Training of Convolutional Neural Networks with Low-bitwidth Weights and Activations
Two-stage and gradually decreasing quantization, stochastic precision sampling, and joint teacher-student distillation each improve low-bit CNN accuracy on ImageNet and CIFAR-100, with the largest gains when combined.