This transformation can provide convergence and robustness improvements for the optimisation and sampling processes when hyper-parameters are not calibrated.
Possible implementation
Add an instance of the ScaledTransformation that scales the parameter space to an amplitude of 1 and range of [0,1].
Feature description
Normalising the parameter space to a common amplitude and scale can be viewed as a special-case for the
ScaledTransformationinstance. For more information, see https://en.wikipedia.org/wiki/Normalization_(statistics)Motivation
This transformation can provide convergence and robustness improvements for the optimisation and sampling processes when hyper-parameters are not calibrated.
Possible implementation
Add an instance of the
ScaledTransformationthat scales the parameter space to an amplitude of 1 and range of [0,1].Additional context
No response