def start(self): # testing() dataset = self.dataset print (dataset) image_x, image_y = get_image_size(dataset) amount_classes = get_amount_of_classes(dataset) # Pooling size poolsize_x = self.pooling_x poolsize_y = self.pooling_y # Learning rate # Epochs to be trained and batch size user_learning_rate = self.learn_r user_nepochs = self.epochs user_batch = self.batch # Size of the convolution filter windows user_filter_x = self.filter_x user_filter_y = self.filter_y # Treshhold for model training user_treshhold = 0.995 #amount of layers given by user n=self.layers print ('Please, wait until all iterations are completed.') evaluate_lenet5(learning_rate=self.learn_r, n_epochs=self.epochs, dataset=self.dataset, nkerns=[20, 50], batch_size=self.batch)
def test_convolutional_mlp(): convolutional_mlp.evaluate_lenet5(n_epochs=1, nkerns=[5, 5])
dataset[n, N] = files_data[n][1] return dataset dataset = createDataSet() # outputFile = "c:\\temp\\foo.csv" # np.savetxt(outputFile, np.asarray(dataset), delimiter=",", fmt='%d') def getFormattedDataSet(dataset): train_set = (dataset[0:10, :-1], dataset[0:10, -1]) valid_set = (dataset[10:14, :-1], dataset[10:14, -1]) test_set = (dataset[14:19, :-1], dataset[14:19, -1]) train_set_x, train_set_y = shared_dataset(train_set) valid_set_x, valid_set_y = shared_dataset(valid_set) test_set_x, test_set_y = shared_dataset(test_set) rval = [(train_set_x, train_set_y), (valid_set_x, valid_set_y), (test_set_x, test_set_y)] return rval evaluate_lenet5(getFormattedDataSet(dataset), learning_rate=0.01) # plt.imshow(bla2, cmap=plt.cm.gray) # l = misc.lena() # plt.show()
def test_convolutional_mlp(): t0=time.time() convolutional_mlp.evaluate_lenet5(n_epochs=5,nkerns=[5,5]) print >> sys.stderr, "test_convolutional_mlp took %.3fs expected 168s in our buildbot"%(time.time()-t0)
logistic_cg.cg_optimization_mnist(mnist_pkl_gz=c10) sys.stdout = open('results/cifar-10_results/lsgd.out', 'w') logistic_sgd.sgd_optimization_mnist(dataset=c10) sys.stdout = open('results/cifar-10_results/lsgd_gau.out', 'w') logistic_sgd_gaussian.sgd_optimization_mnist(dataset=c10) sys.stdout = open('results/cifar-10_results/lsgd_bin.out', 'w') logistic_sgd_binomial.sgd_optimization_mnist(dataset=c10) sys.stdout = open('results/cifar-10_results/mlp.out', 'w') mlp.test_mlp(dataset=c10) sys.stdout = open('results/cifar-10_results/mlpO.out', 'w') # mlp_dropOut.test_mlp(p=0.8, n_hidden = 100) mlp_dropOut.test_mlp(dataset=c10) sys.stdout = open('results/cifar-10_results/mlpC.out', 'w') mlp_dropConnect.test_mlp(dataset=c10) sys.stdout = open('results/cifar-10_results/convo.out', 'w') convolutional_mlp.evaluate_lenet5(dataset=c10) sys.stdout = open('results/cifar-10_results/convoC.out', 'w') # con_mlp_dropConnect.evaluate_lenet5(p=0.8) con_mlp_dropConnect.evaluate_lenet5(dataset=c10) sys.stdout = open('results/cifar-10_results/convoO.out', 'w') con_mlp_dropOut.evaluate_lenet5(dataset=c10)
def conv_mlp_train_and_predict(train_set_x,train_set_y,valid_set_x,valid_set_y,test_set_x,test_set_y,rng,isrbg=False): PRED = evaluate_lenet5(np.copy(train_set_x),np.copy(train_set_y),np.copy(valid_set_x),np.copy(valid_set_y),np.copy(test_set_x),np.copy(test_set_y),rng,isrbg=isrbg) return PRED