An Efficient Method for Model Pruning Using Knowledge Distillation with Few Samples.
ZhaoJing ZhouYun ZhouZhuqing JiangAidong MenHaiying WangPublished in: ICASSP (2022)
Keyphrases
- prior knowledge
- probabilistic model
- mathematical model
- objective function
- statistical model
- cost function
- network model
- sensitivity analysis
- high order
- hybrid model
- theoretical analysis
- input data
- statistical information
- prediction model
- modeling method
- em algorithm
- computationally efficient
- classification method
- similarity measure
- closed form
- classification algorithm
- test data
- conceptual model
- linear model
- study proposes
- pairwise
- evaluation method
- recognition algorithm
- bp neural network
- evaluation model
- knowledge base
- monte carlo simulation
- reconstruction method
- pruning algorithm
- data sets
- pruning method
- gaussian distribution
- parameter space
- prior information
- tree structure
- energy function
- detection method
- parameter estimation
- training samples
- knowledge representation
- domain knowledge
- expert systems
- bayesian networks
- learning algorithm