Reshaping mapper that flattens multidimensional arrays into 1D vectors.
This mapper performs relatively cheap reshaping of arrays from ND into 1D and back upon reverse-mapping. The mapper has to be trained with a data array or dataset that has the first axis as the samples-separating dimension. Mapper training will set the particular multidimensional shape the mapper is transforming into 1D vector samples. The setting remains in place until the mapper is retrained.
At present this mapper is only designed (and tested) to work with C-ordered arrays.
Available conditional attributes:
(Conditional attributes enabled by default suffixed with +)
shape : tuple
maxdims : int or None
enable_ca : None or list of str
disable_ca : None or list of str
auto_train : bool
force_train : bool
space: str, optional :
postproc : Node instance, optional
descr : str