tune

The tune program is used for hyper-parameter optimization of the Support Vector Machine parameters C and gamma, as well as other predictor hyper-parameters. The standard options used in CPSign are normally good when using the signatures descriptors in SVM problems, but here you can optionally run tuning of these.

Usage manual

The full usage manual can be retrieved by running command:

> ./cpsign-[version]-fatjar.jar tune

Picking parameter search space

Parameter search space is fully configurable either by supplying specific values for each parameter, or by giving a range specification. See further information at Specifying lists of numbers as parameter. The grid of parameters are set using the -g, --grid CLI parameter using one of the following syntaxes;

--grid C # Specify to include SVM Cost param, using the default set of parameter values to try
--grid C=10:100:10 # List syntax, enumerates the list: 10,20,..,100
--grid C=1,10,50,100 # Explicit list of numbers 1, 10, 50, 100
--grid C=b2:1:10:2 # List syntax, with different "base": 2^1, 2^3, 2^5, 2^7, 2^9 (note 10 is excluded)

Example: tuning β

The smoothing factor, β, of the logarithmically normalized nonconformity measure introduced in Nonconformity measures can be optimized with the tune program. This is done slightly different than with the C and gamma values, here you can simply add a list of β values that you wish to test (given that you have set the logarithmically normalized nonconformity measure in a regression case):

> java -jar cpsign-[version].jar tune \
  --grid beta=0.0,0.1,0.2,0.5 \
  ..