def read_landsat():
array = []
f = open("database/landsat.txt", "r")
for line in f.read().splitlines():
lista = line.split(" ")
classe = lista.pop(len(lista) - 1)
parameters = [float(i) for i in lista]
array.append(Sample(parameters, classe))
bd = Dataset("landsat", array)
return bd
def read_database(name, index):
array = []
f = open("database/" + name + ".txt", "r")
for line in f.read().splitlines():
lista = line.split(",")
classe = lista.pop(index)
parameters = [float(i) for i in lista]
array.append(Sample(parameters, classe))
bd = Dataset(name, array)
return bd
def read_yeast():
array = []
f = open("database/yeast.data", "r")
for line in f.read().splitlines():
lista = line.split()
classe = lista.pop(len(lista) - 1)
lista.pop(0)
parameters = [float(i) for i in lista]
array.append(Sample(parameters, classe))
bd = Dataset("yeast", array)
return bd
def read_xls_database(name):
df = pd.read_excel("database/" + name + ".xls")
array = []
nparray = df[1:].as_matrix()
for k in nparray:
infos = k
classe = infos[-1]
parameters = [float(i) for i in infos[:-1]]
array.append(Sample(parameters, classe))
data_base = Dataset(name, array)
return data_base
def cluster_centers_pcm(self, numero_de_centros, vizinhos):
parameters = np.transpose(
np.array([k.parameters for k in self.initial_dataset.samples]))
cntr, u, u0, d, jm, p, fpc = fuzzy.cluster.cmeans(
parameters, numero_de_centros, 2.0, 0.005, 2000)
new_samples = []
local_dataset = self.dataset
for k in cntr:
amostras = self.get_k1_closest(k, vizinhos)
for sample_pega in amostras:
new_samples.append(sample_pega)
self.clustered_dataset = Dataset("Clustered", new_samples)
# Add the cost and accuracy to summary
tf.summary.scalar('loss', cost)
tf.summary.scalar('accuracy', accuracy)
# Merge all summaries together
merged_summary = tf.summary.merge_all()
path = "../dqn" #The path to save our model to.
num_epochs = 100
batch_size=32
saver = tf.train.Saver()
with tf.Session() as sess:
dset=Dataset()
image_count=dset.getNumExamples()
# Initialize all variables
sess.run(tf.global_variables_initializer())
print('Loading Model...')
ckpt = tf.train.get_checkpoint_state(path)
saver.restore(sess,ckpt.model_checkpoint_path)
# Loop over number of epochs
for epoch in range(num_epochs):
start_time = time.time()
train_accuracy = 0
for batch in range(0, int(image_count/batch_size)):
# Run the optimizer using this batch of training data.
dataBatch=dset.next_batch(batch_size)
def split_train_test(dataset, i):
test = dataset[i]
train_dataset = Dataset(
"Treino", array_of_samples_to_db([dataset[:i] + dataset[i + 1:]]))
test_dataset = Dataset("Teste", test)
return test_dataset, train_dataset
else:
print("Loading configuration from file %s" % inputfile)
print()
inputParser = InputParser(inputfile)
dimensions = inputParser.get_dimensions()
clusters = inputParser.get_clusters()
export_name = inputParser.get_export_name()
print("Dimensions = %s" % dimensions)
print("Cluster count = %s" % len(clusters))
for cluster in clusters:
print(" - %s" % cluster)
print()
dataset = Dataset(dimensions, clusters, export_name)
print("Generating random values for clusters...")
dataset.generate_values()
print("Balancing clusters... (this may take a while)")
dataset.balance_clusters()
print()
filewriter = FileWriter(dataset)
print("Writing SOMToolbox files...")
filewriter.export_for_somtoolbox()
if 1 <= dimensions <= 2 and not quiet_mode:
print()
print("Input has <= 2 dimensions - showing plot")
plotter = Plotter(dataset)
plotter.plot()
# using teacher forcing
dec_input = tf.expand_dims(targ[:, t], 1)
batch_loss = (loss / int(targ.shape[1]))
variables = encoder.trainable_variables + decoder.trainable_variables
gradients = tape.gradient(loss, variables)
optimizer.apply_gradients(zip(gradients, variables))
return batch_loss
EPOCHS = 5
steps_per_epoch = 5
dataset = Dataset()
for epoch in range(EPOCHS):
start = time.time()
enc_hidden = encoder.initialize_hidden_state()
enc_cell = encoder.initialize_cell_state()
total_loss = 0
for (batch, (inp, targ)) in enumerate(dataset(steps_per_epoch)):
batch_loss = train_step(inp, targ, enc_hidden, enc_cell)
total_loss += batch_loss
if batch % 1 == 0:
print('Epoch {} Batch {} Loss {:.4f}'.format(
epoch + 1, batch, batch_loss.numpy()))