Changeset 66
- Timestamp:
- 07/08/13 18:17:33 (11 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
distools/protselfd.m
r30 r66 15 15 % 16 16 % OUTPUT 17 % W Selection mapping ('feature selection') 17 % W Selection mapping ('feature selection') or prototype indices. 18 18 % E Error stimate as a function of number of selected prototypes 19 19 % (for supervised selection only reliable for prototype sizes >= class size) … … 46 46 % In case of unsupervised selection the maximum or the mean distances to 47 47 % the nearest prototype are minimized. These criteria are the same as used 48 % in the KCENTRE and KMEDIOD cluster procedures. 48 % in the KCENTRE and KMEDIOD cluster procedures. What is returned now in W 49 % is the (ordered) list of prototype indices and not a mapping. 49 50 % 50 51 % REFERENCE … … 60 61 % P.O. Box 5031, 2600 GA Delft, The Netherlands 61 62 62 %63 64 63 function [R,e,D] = protselfd(D,ksel,type) 65 64 … … 77 76 [R,e,D,J,nlab,clab] = protselfd(D,ksel,type); 78 77 case {'maxdist','meandist'} 79 R= protselfd_unsuper(D,ksel,type);78 [R,e] = protselfd_unsuper(D,ksel,type); 80 79 otherwise 81 80 error('Unknown selection type') … … 182 181 % D Square dissimilarity matrix, zeros on diagonal 183 182 % P Number of prototypes to be selected 184 % CRIT ' dist' or 'centre'183 % CRIT 'maxdist' or 'meandist' 185 184 % 186 185 % OUTPUT … … 197 196 % 198 197 % SEE ALSO 199 % KCENTRES 198 % KCENTRES, KMEDIODS 200 199 201 200 % Copyright: R.P.W. Duin, r.p.w.duin@prtools.org … … 203 202 % P.O. Box 5031, 2600 GA Delft, The Netherlands 204 203 205 function N= protselfd_unsuper(d,p,crit)204 function [N,e] = protselfd_unsuper(d,p,crit) 206 205 207 206 d = +d;
Note: See TracChangeset
for help on using the changeset viewer.