/Hyperparameter-robust-simulation-optimization

Hyperparameter tuning using a robust simulation optimization framework

Primary LanguagePython

Day and Night

See Readme From https://github.com/cameronfabbri/Day-Night-Classification

MNIST

Set mnist_on on in RSO.py

  1. Run mylhs.py to get the LHS design points for a given range of Hypes
  2. Run RSO.py with RSO_use flag on to get RSO results into res/mnist_rso.csv
  3. Run RSO.py with hyperopt_use flag on to get HP result
  4. Run RSO.py with hype_given flag on to get final results of RSO vs HP.

We don't need to make data folder etc, directly run RSO.py with desired flags, so set

  1. hyperopt_use=1 to use HyperOpt library.
  2. RSO_use=1 to get the experimental points results for RSO.
  3. hype_given=1 to get the test accuracies for a given Hyperparameter (like output of RSO) on different noise levels.