Neural Network Compression: Difference between revisions
No edit summary |
|||
| Line 7: | Line 7: | ||
* Mozer and Smolensky (1988)<ref name="mozer1988skeletonization"></ref> use a gate for each neuron. Then the sensitivity and be estimated with the derivative w.r.t the gate. | * Mozer and Smolensky (1988)<ref name="mozer1988skeletonization"></ref> use a gate for each neuron. Then the sensitivity and be estimated with the derivative w.r.t the gate. | ||
* Karnin estimates the sensitivity by monitoring the change in weight during training. | |||
==Factorization== | ==Factorization== | ||
Revision as of 20:42, 2 February 2021
Brief survey on neural network compression techniques.
Pruning
Sensitivity Methods
The idea here is to measure how sensitive each neuron is.
I.e., if you remove the neuron, how will it change the output?
- Mozer and Smolensky (1988)[1] use a gate for each neuron. Then the sensitivity and be estimated with the derivative w.r.t the gate.
- Karnin estimates the sensitivity by monitoring the change in weight during training.
Factorization
Resources
Surveys
- Pruning algorithms a survey (1993) by Russel Reed
- A Survey of Model Compression and Acceleration for Deep Neural Networks (2017) by Cheng et al.