Please enable JavaScript.
Coggle requires JavaScript to display documents.
[213] A comprehensive survey on optimizing deep learning models by…
[213]
A comprehensive survey on optimizing deep learning models by metaheuristics
optimization problems
1 architecture optimization,
2 hyper-parameter optimization,
3 training and feature representation level optimization
DNN Architectures
CNN
Unsupervised pretained networks
auto-encoders
Restricted Boltzmann Machine & Deep belief networks
Generative adversarial neural networks
RNN(Recurrent neural networks)
Recursive nerual networks
Metaheuristic algorithms
Paper
Problems of Optimization
HyperParameter Optimizatoin
Training DNNs
Architecture Optimization(Architecture Search)
Optimization at feature representation level
尋找paper 的方法, 條件
使用metaheuristics
Hyper-Parameter optimization
training deep neural networks
architecture optimization
feature representation level
只統計了使用metaheuristics的papers
defined main optimization problems in DL field and presented representation schemes
to encode a DNN structure for metaheuristics
