A low-cost prediction strategy using curve fitting and SVMs on initial-final accuracy pairs from multiple trainings enables efficient hyper-parameter optimization for CNNs on MNIST and CIFAR-10.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Mise en abyme with artificial intelligence: how to predict the accuracy of NN, applied to hyper-parameter tuning
A low-cost prediction strategy using curve fitting and SVMs on initial-final accuracy pairs from multiple trainings enables efficient hyper-parameter optimization for CNNs on MNIST and CIFAR-10.