MAPSD

Train mapping between two representations

     [W,V] = MAPSD(S,D,P,Q)

Input
S	Dataset, source representation, typically high-dimensional
D	Dataset, destination representation, typically low-dimensional
P	Non-linearity parameter for W between 0 and 1, default 0.01
Q	Non-linearity parameter for V between 0 and 1, default 0.01

Output
W	Mapping, such that D = S*W
V	Mapping, such that S = D*V

Description

This mapping is useful to generalise mappings like MDS and TSNEM such  that they can be applied to new datasets. Once by such routines a proper  low-dimensional representation D is found for the original dataset S then  W can be applied to find an approximate representation D2 of a similar to  S source representation S2 by D2 = S2*W.

See also

datasets, mappings, mds, tsnem,