def main(): kernelFile = '/afs/cs.stanford.edu/u/rwitten/projects/multi_kernel_spl/data/allkernels_info.txt' trainFile = './train.newsmall_1_reducedy.txt' #TODO: Rafi, move this to your data directory so it doesn't clutter things up params = Params() spl_params = Params() spl_params.spl_mode = 0 params.max_outer_iter = 1337 #TODO: make this user input loadKernelFile(kernelFile, params) loadTrainFile(trainFile, params) w = ImagePsi.PsiObject(params) LSSVM.optimize(w, params, spl_params) return params
def train_and_predict_task(self, t, train_X, train_y, eval_X, param_dict):
kernel_func = self.get_kernel_func(param_dict['kernel'],
param_dict['beta'])
self.models[t] = lssvm.LSSVM(C=param_dict['C'],
kernel_func=kernel_func)
converged = self.models[t].fit(train_X, train_y)
if converged:
preds = self.models[t].predict(eval_X)
else:
# predict majority class
preds = np.sign(np.mean(train_y)) * np.ones(len(eval_X))
return preds
def main():
try:
params = UserInput.getUserInput('train')
ExampleLoader.loadExamples(params)
CommonApp.setExampleCosts(params)
w = None
if params.initialModelFile:
w = CacheObj.loadObject(params.initialModelFile)
else:
w = CommonApp.PsiObject(params,False)
globalSPLVars = SPLSelector.SPLVar()
if params.splParams.splMode != 'CCCP':
SPLSelector.setupSPL(params)
w = LSSVM.optimize(w, globalSPLVars, params)
CacheObj.cacheObject(params.modelFile,w)
Performance.printStrongAndWeakTrainError(params, w)
except Exception, e :
import traceback
traceback.print_exc(file=sys.stdout)
Test = numpy.loadtxt('dataset///monks_2_test.txt')
X_t = Test[:,1:-1]
Y_t = Test[:,0]
X_N_t = NorX.fT(X_t)
Y_N_t = NorY.fT(Y_t)
Y_N_t = Y_N_t * 2 -1
(alpha,b,K) = LSSVM.LSSVM_CV(X_N,Y_N,'RBF',[0.01,0.1,0.5,1,2,10,25,50,100],[0.001,0.01,0.1,0.2,0.5,1,5,10,40,100],arg2 = None)
#(alpha,b,K) = LSSVM.LSSVM_CV(X_N,Y_N,'LINEAR',[0.001,0.005,0.01,0.05,0.1,0.5,1,4,10,25,100])
#(alpha,b,K) = LSSVM.LSSVM_CV(X_N,Y_N,'POLY',[0.01,0.1,0.5,1,2,10,25,50,100],[0.001,0.01,0.1,0.2,0.5,1,5,10,40,100],[1,2,3,4,5])
#(alpha,b,K) = LSSVM.LSSVM_CV(X_N,Y_N,'TANH',[0.01,0.1,0.5,1,2,10,25,50,100],[-10,-5,-3,-2,-1,-0.5,-0.1,-0.05,-0.01,0.01,0.05,0.1,0.5,1,3,5,10],[0.1,1,2,3,10])
#(alpha,b,K) = LSSVM.LSSVM_CV(X_N,Y_N,'TL1',[0.001,0.005,0.01,0.03,0.1,0.5,1,2,10,25,50,100],[0.001,0.005,0.01,0.05,0.1,0.2,0.5,1,2,5,10,20])
Y_predict = LSSVM._LSSVMpredict(X_N_t,K,alpha,b,Y_N)
acc = LSSVM._compare(Y_N_t,Y_predict)
print(acc)
