Hi,
I have a few problems with svms.
This works fine:
load(’fruit’)
>> out = sdsvc(a);
one-against-all: [’apple’ ....................sigma=0.96624 C=0.336 err=0.063 SVs=68] [’banana’ ....................sigma=0.96137 C=2.98 err=0.113 SVs=56] [’stone’ ....................sigma=3.22551 C=12.7 err=0.010 SVs=30]
However:
>> out = a*sdsvc;
Warning: In the directory “C:\Documents and
Settings\MATLAB\prsd_toolbox”, svmtrain.mexw32 now
shadows svmtrain.dll.
Please see the MATLAB 7.1 Release Notes.
PRSD Studio Commercial 2.2.2 (22-Jun-2010), Copyright (C) 2007-2010, PR Sys Design, All rights reserved
Error in ==> C:\Documents and Settings\MATLAB\prsd_toolbox\sdsvc.p>sdsvc at 30
??? Output argument “p” (and maybe others) not
assigned during call to “C:\Documents and
Settings\MATLAB\prsd_toolbox\sdsvc.p>sdsvc”.
Is this due to svmtrain.mexw32 shadowing svmtrain.dll?
And when I try it with my own dataset I also get an error (only with sdsvc and not with other sdclassifiers). Did I forget to set a certain property for sdsvc?
>> ds
998 by 52 sddata, 2 classes: ‘b’(937) ‘n’(61)
>> struct(ds)
ans =
format: ‘sddata1’
data: [1x1 struct]
>> sdsvc(ds)
Warning: In the directory “C:\Documents and
Settings\MATLAB\prsd_toolbox”, svmtrain.mexw32 now
shadows svmtrain.dll.
Please see the MATLAB 7.1 Release Notes.
??? Error using ==> vertcat at 67
Uknown property format
Error in ==> C:\Documents and
Settings\MATLAB\prsd_toolbox\sdsvc.p>sdsvc at 109
p = sdsvc;
Error in ==> C:\Documents and Settings\MATLAB\prsd_toolbox\sdsvc.p>sdsvc at 30
??? Output argument “p” (and maybe others) not
assigned during call to “C:\Documents and
Settings\MATLAB\prsd_toolbox\sdsvc.p>sdsvc”.

