SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization.
Juyong KimYookoon ParkGunhee KimSung Ju HwangPublished in: ICML (2017)
Keyphrases
- statistical model
- learning scheme
- learning models
- prior knowledge
- computational model
- network structure
- mathematical model
- machine learning
- linear model
- experimental data
- neural network
- active learning
- learning algorithm
- parallel processing
- connectionist networks
- fully connected
- learning phase
- hidden variables
- high level
- parameter values
- parameter estimation
- learning problems
- graphical models
- online learning
- knowledge acquisition
- generative model
- learning systems