Changeset 18

Timestamp:

06/15/11 15:05:46 (14 years ago)

Author:

bduin

Message:

clevald, parzenddc, parzend_map and kem added

Location:

distools

Files:

• Contents.m (modified) (1 diff)
• clevald.m (added)
• disnorm.m (modified) (3 diffs)
• distools_long.m (added)
• kem.m (added)
• parzend_map.m (added)
• parzenddc.m (added)

distools/Contents.m

@@ -53 +53 @@
 %  Routines supporting in learning from dissimilarity matrices
 %  -----------------------------------------------------------------------
-%  CROSSVALD    Cross-validation error for dissimilarity representations
+%  CROSSVALD    Cross-validation error for dissimilarity data
+%  CLEVALD      Classifier evaluation (learning curve) for dissimilarity data
 %  DISSPACES    Compute various spaces out of a dissimilarity matrix
 %  GENDDAT      Generate random training and test sets for dissimilarity data

distools/disnorm.m

@@ -1 +1 @@
 %DISNORM Normalization of a dissimilarity matrix
 %
-%       V = disnorm(D,OPT)
+%       V = DISNORM(D,OPT)
 %   F = E*V
 %
 % INPUT
 %   D    NxN dissimilarity matrix or dataset, which sets the norm
-%   E    Matrix to be normalized
+%   E    Matrix to be normalized, e.g. D itself
 %       OPT  'max' : maximum dissimilarity is set to 1 by global rescaling
 %                  'mean': average dissimilarity is set to 1 by global rescaling (default)
@@ -17 +17 @@
 %   OPT = 'mean'
 %
+% DESCRIPTION
+% Operation on dissimilarity matrices, like the computation of classifiers
+% in dissimilarity space, may depend on the scaling of the dissimilarities
+% (a single scalar for the entire matrix). This routine computes a scaling
+% for a giving matrix, e.g. a training set and applies it to other
+% matrices, e.g. the same training set or based on a test set.
 
 % Copyright: Elzbieta Pekalska, ela.pekalska@googlemail.com
@@ -41 +47 @@
 %DEFINE mapping
 if isstr(opt)
-        discheck(D);
+%       discheck(D);
         opt = lower(opt);
         if strcmp(opt,'mean')