Post training 4-bit quantization of convolutional networks for rapid-deployment.
Ron BannerYury NahshanDaniel SoudryPublished in: NeurIPS (2019)
Keyphrases
- uniform quantization
- recurrent networks
- training process
- network design
- successive approximation
- training set
- adaptive quantization
- echo state networks
- quantization error
- data sets
- training phase
- heterogeneous networks
- network architecture
- network analysis
- sparse coding
- community structure
- motion estimation
- active learning
- social networks