DISSIMT

Fixed DISsimilarity-SIMilarity transformations

     T = P*DISSIMT(TYPE,PAR)

Input
P	NxN or NxM proximity matrix or dataset
TYPE	Type of the transformation that acts on proximity data  (optional; default: 'POW'):
'I',	'identity': P
'SCALE',	'scale': P ./ PAR
'NEG',	'negative': -P
'NEG2',	'negative2': -P.^2
'REV',	'reverse': 1./(PAR + P)
'REV2',	'reverse': 1./(PAR^2 + P.^2)
'D2S',	'S2D': 1-P
'D2SSQ','S2DSQ':	sqrt(1-P)
'POW',	'power': P.^PAR
'EXP',	'exponent': exp(-P./PAR)
'EXP2',	'exponent2': exp((-P.^2)./(PAR^2))
'GAUSS':	exp((-P.^2)./(2*PAR^2))
'LOG',	'logarithm': log(P/PAR)
'NLOG',	'neglog': -log((P+1)./PAR) 'NLOG2','neglog2': -log((P.^2+1)./PAR^2)
'SIGM',	'sigmoid': 2./(1+exp(-P./PAR))-1  'SIGM2','sigmoid2': 2./(1+exp(-(P.^2)./PAR^2))-1
'SIM2DIS':	inner products to Euclidean distances
'DIS2SIM':	Euclidean distances to inner products
PAR	Parameter, PAR > 0 (optional; default: 1)

Output
T	NxN or NxM proximity matrix or dataset

Description

A fixed mapping that transforms proximity data. Note that in some cases  similarity is transformed into dissimilarity, or vice versa.