svm - svmtrain - unable to solve the optimization problem -

i using svmtrain discriminate between several pairs of data. although svmtrain works desired in 1 case (outputting classifier object ~70 % accuracy verified svmclassify), other cases seem fail. feature vectors 134 dimensions , using between 300 , 800 data points each class. (each class not have same number of data points). have tried using default kernel svmtrain using method

svm = svmtrain(double(train{k}), group_train{k},'showplot',true); 

in case error:

unable solve optimization problem: maximum number of iterations exceeded; increase options.maxiter. continue solving problem current solution starting point, set x0 = x before calling quadprog.

i have tried extending number of iterations , specifying kernel using call:

options = optimset('maxiter',1000,'largescale','on'); svm = svmtrain(double(train{k}),group_train{k},'kernel_function','mlp','method','qp',...    'quadprog_opts',options); 

in case, error:

unable solve optimization problem: exiting: solution unbounded , @ infinity; constraints not restrictive enough.

in case did work, have 338 data points first class , 476 data points second class. examples, in 3 of cases don't work, have 828, 573, , 333 data points in second class, while first class remains same , has 338 data points. neither method call seems work.

could please me? have been trying solve problem week , have had no luck. using matlab 7.9.0 r2009b on virtual machine windows xp 1 ghz processor , 2 gb ram.

thank much! -vivek

make :

options = optimset('maxiter',1000); svmtrain(totalresult,yresultstotal,'kernel_function','mlp','method','qp',...             'quadprog_opts',options);