NormMaxMin Class Reference
[Processing]
Normalize by mapping min/max range to user specified range.
More...
#include <NormMaxMin.h>
Inherits Marsyas::MarSystem.
Detailed Description
Normalize by mapping min/max range to user specified range.
You can normalize over the entire slice, within an entire observation channel, or over a sample by using the "domain" control.
You can do the normalization all at once by setting the "mode" control to twopass. Alternatively you can train the normalizer with several sets of data by setting the mode control to "train", and then switch to "predict" to normalize a slice to the maximums and minimums found in the training slices.
Controls:
- mrs_real/lower [] : The lower bound you want to normalize the data to (default 0.0)
- mrs_real/upper [] : The upper bound you want to normalize the data to (default 1.0)
- mrs_realvec/maximums [] : The current maximum values for each observation or sample
- mrs_realvec/minimums [] : The current minimum values for each observation or sample
- mrs_string/mode [] : Are we training, predicting or in twopass mode? (train, predict, twopass) (default "train")
- mrs_string/domain [] : Are we normalizing over observations, samples or entire slices? (observations, samples, slices) (default "observations")
- mrs_bool/ignoreLast [] : Should we ignore the last observation or sample?
- mrs_bool/init [] :
Definition at line 56 of file NormMaxMin.h.
The documentation for this class was generated from the following files:
